Method and system for ammonia slip detection and correction of nox measurement values for heavy-duty diesel vehicles

Abstract

The application discloses a heavy diesel vehicle ammonia leakage identification and NOx measurement value correction method and a correction system. First, by constructing and training a long short-term memory network model, the missing urea injection rate is predicted by using engine and exhaust parameters which are highly correlated with urea injection. Second, based on the stoichiometric relationship of urea and NOx reaction, the application innovatively proposes a reducing agent excess characteristic value to quantify the degree of ammonia supply excess. A dynamic sliding window is used for comprehensive judgment to accurately identify ammonia leakage events and to subdivide the diagnosis of non-leakage states. Finally, during the ammonia leakage period, a fixed high-efficiency conversion rate is used to correct the sensor readings. The application does not require additional hardware, and only uses existing remote monitoring data to achieve high-precision ammonia leakage identification and effective correction of contaminated NOx readings, avoiding false over-standard misjudgment of vehicle emission compliance. The application has the advantages of moderate calculation load, strong engineering practicability and easy large-scale promotion.

Description

Technical Field

[0001] This invention relates to the field of diesel vehicle emission detection technology, and in particular to a method and system for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles. Background Technology

[0002] Heavy-duty diesel vehicles are the mainstay of road freight transport, but their nitrogen oxide (NOx) emissions are one of the major air pollutants. To meet increasingly stringent emission regulations (such as GB 17691-2018), the vast majority of heavy-duty diesel vehicles adopt selective catalytic reduction (SCR) aftertreatment systems. This system generates ammonia (NH3) by injecting a urea aqueous solution, which, under the action of a catalyst, reduces NOx into harmless nitrogen and water.

[0003] With regulations increasingly requiring real-world emissions monitoring, emissions management platforms based on in-vehicle remote monitoring have become a crucial tool for overseeing the compliance of vehicles in use. Currently, this platform relies entirely on electrochemical NOx sensors installed downstream of the SCR (Selective Catalytic Reduction) system for NOx emissions monitoring.

[0004] However, this technology has a fatal flaw: the NOx sensor exhibits significant cross-sensitivity to ammonia. Its working principle is that the catalyst on the sensor electrode surface oxidizes ammonia to NO (4NH3 + 5O2 → 4NO + 6H2O), resulting in the output signal being the sum of the NOx and ammonia concentrations. When an ammonia leak occurs in the SCR system, the readings of the downstream NOx sensor will be severely overestimated, causing "false exceedances" and misjudging vehicles as non-compliant, seriously undermining the impartiality of regulation and the credibility of remote monitoring data.

[0005] In existing technologies, the following are the main approaches to solving this problem: 1) Adding an ammonia sensor: This method is limited by the accuracy and cost of the current ammonia sensor itself, making it difficult to install on a large scale in the existing fleet, and it also introduces new measurement uncertainties.

[0006] 2) Observers based on complex models, such as the dynamic cross-sensitivity model coupled with the SCR model, or the use of square root unscented Kalman filtering to estimate ammonia concentration. These methods have complex models, high computational load, require high-frequency data and high-performance processors, and are difficult to integrate into existing remote monitoring terminals for large-scale fleet deployment.

[0007] 3) The excitation perturbation method identifies NOx and ammonia components by perturbing the ammonia-to-nitrogen ratio. This method interferes with normal engine operation and urea injection strategy, and is not suitable for continuous, undisturbed monitoring during actual driving.

[0008] The closest existing technology can be considered a simple threshold-based SCR efficiency monitoring method, which suspects system malfunction or failure when the calculated SCR efficiency is abnormally low. However, this method cannot distinguish whether the efficiency decline is due to ammonia leakage or insufficient NOx conversion capacity, and it completely ignores the fundamental cause of excessive urea supply. More importantly, existing remote monitoring data only includes basic data that is mandated by regulations, while in reality, the key parameter of urea injection rate is usually not a regularly uploaded item, resulting in a lack of direct evidence to determine whether ammonia is excessive in existing methods.

[0009] Therefore, existing technologies generally suffer from problems such as strong hardware dependence, computational complexity making them impractical, or inaccurate identification due to missing key parameters. There is an urgent need for a high-precision, low-computational-complexity ammonia leak identification and correction scheme that can be achieved using only existing remote monitoring data. Summary of the Invention

[0010] Therefore, the purpose of this invention is to provide a method and system for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles, enabling accurate identification of ammonia leak events even when urea injection rate data is missing; and eliminating measurement errors of NOx sensors to avoid false exceedances.

[0011] To achieve the above objectives, this invention proposes a method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles, comprising the following steps: S1. Obtain the raw monitoring data uploaded by the vehicle to the monitoring platform, and preprocess the raw monitoring data; S2. Input the preprocessed original monitoring data into the trained prediction model to predict the missing urea injection rate. S3. Based on the stoichiometric relationship between all urea injection rates and NOx reactions, calculate the excess characteristic value R_NH3 of the reducing agent to quantify the degree of ammonia supply surplus. S4. Using a dynamically sliding observation window, within each observation window, the ammonia leakage period is identified based on the excess reducing agent characteristic value R_NH3 and the SCR efficiency; S5. During the identified ammonia leak period, the NOx sensor readings downstream of the SCR are corrected.

[0012] More preferably, in S1, the raw monitoring data includes at least: vehicle speed, engine speed, net torque, fuel flow rate, intake air flow rate, NOx sensor concentration values ​​upstream and downstream of SCR, SCR inlet and outlet temperatures, engine coolant temperature, and atmospheric pressure.

[0013] More preferably, in S1, the raw monitoring data is preprocessed, including the following steps: The key parameters are calculated from the original monitoring data using the following formula: Excess air coefficient λ: ; SCR airspeed SV: NOx mass flow rate upstream of SCR : in, Intake airflow rate, unit: kg / h; Fuel flow rate, unit: kg / h; The values ​​are: exhaust mass flow rate (kg / h); R is the gas constant; T_SCR is the catalyst temperature (K); P is the exhaust pressure (Pa); and V_SCR is the catalyst volume (m³). 3 M_exh represents the molar mass of the exhaust gas, in g / mol; C_NOx_up represents the NOx concentration value from the upstream NOx sensor of the SCR, in ppm.

[0014] Further preferably, in S2, the preprocessed original monitoring data is input into the trained prediction model to predict the missing urea injection rate, including: 12-15 parameters strongly correlated with urea injection rate were selected from the dataset as input features of the LSTM model. These parameters strongly correlated with urea injection rate include: vehicle speed, net torque, engine speed, fuel flow rate, intake air flow rate, SCR inlet temperature, SCR outlet temperature, coolant temperature, upstream NOx concentration, excess air coefficient, SCR air velocity, and SCR upstream NOx mass flow rate. The collected datasets were filtered, and only data segments with SCR inlet temperatures higher than the urea injection temperature of the vehicle model were retained for model training. Construct an LSTM network containing an input layer, two hidden layers, and an output layer as a prediction model; Collect measured data containing the true value of the actual urea injection rate over a period of time; train and construct a prediction model using the measured data; The trained prediction model is used to predict the urea injection rate from real-time monitoring data that does not contain the true value of the urea injection rate.

[0015] More preferably, in S3: the excess characteristic value R_NH3 of the reducing agent is calculated using the following formula: R_NH3= (m_UWS - m_NOx_up × γ) / (δ × V_SCR) Wherein, m_UWS is the cumulative urea mass within the sliding window, in g; m_NOx_up is the cumulative upstream NOx mass within the window, in g; V_SCR is the SCR catalyst volume, in L; γ is the theoretical urea to NOx mass ratio coefficient, ranging from 1.9 to 2.1, which can be adjusted according to the urea purity; δ is the urea to ammonia conversion coefficient, ranging from 5.3 to 5.6.

[0016] More preferably, in S4, the following steps are included: A dynamically sliding observation window is used to observe the preprocessed raw monitoring data, and the SCR efficiency within the window is calculated. ; Within each sliding window, calculate the average reductant excess characteristic value R_NH3 within the window; Based on the SCR efficiency and average R_NH3 obtained within the window, the following process is used for judgment: When the SCR system is within its effective operating temperature range, if R_NH3>0 and If a preset efficiency threshold is set, the current window is determined to be in an "ammonia leakage" state; otherwise, it is determined to be in a "normal" state.

[0017] Furthermore, when determining whether the current window is in an "ammonia leak" state, based on a continuous confirmation mechanism, an ammonia leak event is finally confirmed to have started only when two or more consecutive sliding windows meet the leak determination conditions.

[0018] Further preferably, the SCR efficiency within the calculation window The following formula is used for calculation: ; Wherein, C_NOx_down is the NOx concentration value of the downstream SCR sensor; C_NOx_up is the NOx concentration value of the upstream SCR sensor.

[0019] Further preferably, in S5, during the identified ammonia leak period, the NOx sensor reading downstream of the SCR is corrected, including: using a preset fixed high-efficiency SCR conversion rate η_fixed, the actual downstream NOx concentration is calculated based on the upstream NOx concentration, and the correction formula is: ; Where C_NOx_up is the NOx concentration value of the upstream NOx sensor of the SCR, and C_NOx_down_cor is the corrected NOx concentration value of the downstream SCR.

[0020] Furthermore, in S4, a comprehensive diagnosis of non-ammonia leakage conditions is also included: When the SCR system is within its effective operating temperature range like If η_SCR < preset efficiency threshold, it is determined to be a "non-ammonia fault state", triggering a check of the urea supply system or catalyst activity. like If η_SCR ≥ the preset efficiency threshold, it is determined to be in "ammonia adsorption storage state" and is used as an early warning state for ammonia leakage for dynamic tracking.

[0021] This invention also provides a heavy-duty diesel vehicle ammonia leak detection and NOx measurement correction system, wherein the steps of the above-mentioned heavy-duty diesel vehicle ammonia leak detection and NOx measurement correction method include: The data acquisition and preprocessing module acquires the raw monitoring data uploaded by the vehicle to the monitoring platform and preprocesses the raw monitoring data. The urea injection rate prediction module is used to receive the preprocessed raw monitoring data and input it into the trained prediction model to predict the missing urea injection rate. The reducing agent excess characteristic value calculation module is used to define and calculate the reducing agent excess characteristic value R_NH3 based on the stoichiometric relationship between urea injection and NOx reaction, so as to quantify the degree of ammonia supply excess. The ammonia leak detection module uses a dynamically sliding observation window to make a comprehensive judgment based on the excess reducing agent characteristic value R_NH3 and SCR efficiency within each window to identify ammonia leak events. The NOx measurement correction module is connected to the ammonia leak identification module and is used to correct the NOx sensor readings downstream of the SCR during the identified ammonia leak period.

[0022] The method and system for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles disclosed in this application have at least the following advantages compared to the prior art:

[0023] 1) High-precision ammonia leak identification was achieved even with missing key parameters. The LSTM model successfully predicted the urea injection rate (not directly uploaded), with a cumulative prediction error within ±10%, providing reliable input for subsequent identification. This was further enhanced by combining an innovative excess reducing agent feature value R_NH3 with a dual-condition criterion (R_NH3>0 and...). A sliding window criterion with a preset efficiency threshold can accurately distinguish between ammonia leaks and actual NOx conversion failures. Experimental verification shows that in multiple real-world road emission test cycles, the recall rate for ammonia leak events reaches 77%~97%, and the overall classification accuracy exceeds 93%.

[0024] 2) Significantly eliminated measurement errors from the NOx sensor, avoiding false exceedances. After calibration, the NOx sensor readings closely matched the PEMS benchmark measurements, eliminating the positive bias caused by ammonia cross-sensitivity. Experimental data showed that in the assessment of the 90th percentile of NOx emissions, a key indicator of vehicle emissions compliance, the corrected absolute average error was reduced by 94% (from 0.50 g / kWh to 0.03 g / kWh), and the width of the error distribution (interquartile range) was reduced by 78% (from 1.00 g / kWh to 0.22 g / kWh). In multi-vehicle testing, five false exceedances caused by ammonia interference were successfully avoided.

[0025] 3) It possesses high engineering practicality and scalability. First, the method is entirely based on existing remote monitoring data streams, requiring no additional sensors, making it particularly suitable for low-cost and rapid deployment in existing fleets. Second, compared to complex algorithms such as EKF and SRUKF, LSTM and sliding window algorithms have moderate computational loads and can run stably on existing remote monitoring terminals or cloud platforms, meeting the processing needs of large-scale fleets. Furthermore, the entire identification and correction process does not affect the vehicle's ECU control strategy or normal urea injection, making it suitable for real-world, continuous driving monitoring. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating the method for identifying ammonia leaks in heavy-duty diesel vehicles and correcting NOx measurements according to the present invention. Figure 2 This is a schematic diagram of the heavy-duty diesel vehicle ammonia leak identification and NOx measurement correction system of the present invention. Detailed Implementation

[0027] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0028] like Figure 1 As shown, one embodiment of the present invention provides a method for identifying ammonia leaks in heavy-duty diesel vehicles and correcting NOx measurements, which includes the following steps: S1. Obtain the raw monitoring data uploaded by the vehicle to the monitoring platform and preprocess the raw monitoring data; in S1, the raw monitoring data includes at least: vehicle speed, engine speed, net torque, fuel flow, intake air flow, NOx sensor concentration values ​​upstream and downstream of SCR, SCR inlet and outlet temperatures, engine coolant temperature, and atmospheric pressure. Furthermore, the raw monitoring data is preprocessed, including the following steps: The key parameters are calculated from the original monitoring data using the following formula: Excess air coefficient λ: ; SCR airspeed SV: NOx mass flow rate upstream of SCR : in, Intake airflow rate, unit: kg / h Fuel flow rate, unit: kg / h The values ​​are: exhaust mass flow rate (kg / h), R (gas constant), T_SCR (catalyst temperature, K), P (exhaust pressure, Pa), and V_SCR (catalyst volume, m³). 3 M_exh represents the exhaust molar mass in g / mol, and C_NOx_up represents the NOx concentration value from the upstream NOx sensor of the SCR in ppm.

[0029] S2. Input the preprocessed raw monitoring data into the trained prediction model to predict the missing urea injection rate; before using the prediction model for prediction, the following steps are also included: Data preparation: Through on-vehicle testing, a short data segment containing the true value of the actual urea injection rate was collected. This collected data is used for subsequent model training to determine the relationship between the urea injection rate and remote monitoring data. This relationship primarily depends on the manufacturer's calibration; for the same vehicle model and engine type, this relationship is generally consistent and unlikely to change significantly in the short term. To improve efficiency, only data segments with SCR inlet temperatures higher than the urea injection start-up temperature of this vehicle model were used.

[0030] 12-15 parameters strongly correlated with urea injection rate were selected from the dataset as input features of the LSTM model. These parameters strongly correlated with urea injection rate include: vehicle speed, net torque, engine speed, fuel flow rate, intake air flow rate, SCR inlet temperature, SCR outlet temperature, coolant temperature, upstream NOx concentration, excess air coefficient, SCR air velocity, and SCR upstream NOx mass flow rate. The collected datasets were filtered, and only data segments with SCR inlet temperatures higher than the urea injection temperature of the vehicle model were retained for model training. Construct an LSTM network containing an input layer, two hidden layers, and an output layer as a prediction model; Measured data containing the true values ​​of actual urea injection rates were collected over a period of time. A prediction model was trained using this measured data. The dataset containing the true values ​​was divided into a training set (first 80%) and a test set (last 20%) in chronological order. The training set was further divided into a training subset and a validation set. The memory window length and the number of hidden layer nodes were optimized by minimizing the mean squared error (MSE) on the validation set. An optimized example is a 60-second memory window with 40 and 60 hidden layer nodes.

[0031] The trained prediction model is used to predict the urea injection rate from real-time monitoring data that does not contain the true value of the urea injection rate.

[0032] S3. Based on the stoichiometric relationship between all urea injection rates and NOx reactions, calculate the excess characteristic value R_NH3 of the reducing agent to quantify the degree of ammonia supply surplus. The excess characteristic value of the reducing agent R_NH3 is calculated using the following formula: R_NH3= (m_UWS - m_NOx_up × γ) / (δ × V_SCR) Wherein, m_UWS is the cumulative urea mass within the sliding window, in g; m_NOx_up is the cumulative upstream NOx mass within the window, in g; V_SCR is the SCR catalyst volume, in L; γ is the theoretical urea to NOx mass ratio coefficient, which is taken as γ=2.01 in this embodiment; δ is the urea to ammonia conversion coefficient, which is taken as δ=5.43 in this embodiment.

[0033] This invention innovatively proposes a characteristic value R_NH3 for excess reducing agent to quantify the degree of ammonia supply surplus.

[0034] The theoretical basis is that, during complete conversion, the mass ratio of urea to ammonia is approximately 5.43, and the stoichiometric mass ratio of ammonia to NOx is approximately 0.37. Therefore, theoretically, the mass of urea required to treat one unit mass of NOx is approximately 5.43 × 0.37 = 2.01. It can be seen that when the mass of urea (m_UWS) is greater than 2.01 times the mass of NOx (m_NOx_up), urea is excess, and R_NH3 > 0; conversely, when the mass of urea (m_UWS) is less than 2.01 times the mass of NOx (m_NOx_up), urea is insufficient, and R_NH3 < 0. That is, R_NH3 characterizes the degree of ammonia excess relative to the current NOx level; the larger the R_NH3, the higher the degree of ammonia excess. The denominator is used for normalization, making R_NH3 a dimensionless number, which is convenient for application to different vehicle models.

[0035] S4. A dynamically sliding observation window is used. Within each observation window, the ammonia leakage period is identified based on the excess reducing agent characteristic value R_NH3 and the SCR efficiency. The window length is preferably 5 to 20 seconds; in this embodiment, 10 seconds is used, with a sliding step of 1 second. The process includes the following steps: A dynamically sliding observation window is used to observe the preprocessed raw monitoring data, and the SCR efficiency within the window is calculated. ; SCR efficiency within the calculation window The following formula is used for calculation: ; Wherein, C_NOx_down is the NOx concentration value of the downstream SCR sensor; C_NOx_up is the NOx concentration value of the upstream SCR sensor.

[0036] Within each sliding window, calculate the average reductant excess characteristic value R_NH3 within the window; Based on the SCR efficiency and average R_NH3 obtained within the window, the following process is used for judgment: When the SCR system is within its effective operating temperature range (recommended 200°C~550°C, preferably 220°C~500°C), the following judgment is made: like If η_SCR < preset efficiency threshold (e.g., 95% to 99.5%, preferably 98% in this embodiment), then the current window is determined to be in the "ammonia leakage" state; otherwise, it is determined to be in the "normal" state.

[0037] To avoid false alarms caused by instantaneous data fluctuations, this embodiment employs a continuous confirmation mechanism: an ammonia leak event is only confirmed to have started when two (or more) consecutive sliding windows meet the aforementioned leak determination conditions. The ammonia leak reset condition is: when When η_SCR > preset efficiency threshold, the leakage event is considered to have ended, and the system is reset.

[0038] S5. During the identified ammonia leak period, the NOx sensor readings downstream of the SCR are corrected, including: using a preset fixed high-efficiency SCR conversion rate η_fixed (e.g., 98%–99.5%, preferably 99% in this embodiment), the actual downstream NOx concentration is calculated based on the upstream NOx concentration, and the correction formula is: ; Wherein, C_NOx_up is the NOx concentration value of the upstream NOx sensor of the SCR, and C_NOx_down_cor is the corrected NOx concentration value of the downstream SCR.

[0039] Example 1

[0040] In this embodiment, based on the SCR efficiency and average R_NH3 within the window calculated above, the ammonia leakage period is determined according to the following process: When the SCR system is within its effective operating temperature range (recommended 220°C - 500°C), the following criteria should be applied: State A: Normal State: If If η_SCR > preset efficiency threshold, then it is determined to be in a "normal" state; no correction is required at this time.

[0041] Status B: "Ammonia Leakage" Status: When the SCR system is within its effective operating temperature range, if... If η_SCR < preset efficiency threshold, then the current window is determined to be in the "ammonia leakage" state.

[0042] State C: Non-ammonia fault state: If If η_SCR < preset efficiency threshold, it is determined that the SCR system has a conversion efficiency decrease due to reasons other than ammonia leakage, triggering urea supply system fault diagnosis or catalyst activity check. The downstream NOx sensor reading is true and valid, and can also be directly used for emission assessment.

[0043] State D: Ammonia adsorption storage state: If If η_SCR ≥ the preset efficiency threshold, then the SCR system is determined to be in the ammonia adsorption and storage stage, and dynamic tracking is performed as an early warning state for ammonia leakage.

[0044] Example 2

[0045] This embodiment provides a detailed explanation and differentiated treatment of the comprehensive judgment of non-ammonia leakage status based on the above embodiment 1.

[0046] For example, during a real-world road test, the monitoring system calculated the following data within a series of sliding windows: Scenario 1: Urea depletion condition When the vehicle reached the highway section, the SCR inlet temperature was 280℃, within the effective operating window. Calculation results from multiple consecutive windows showed that R_NH3 remained negative, and η_SCR gradually decreased from 98% to 85%. Based on the judgment logic, the system determined it to be a "non-ammonia fault state." The system triggered a urea supply system check alarm and simultaneously marked the downstream NOx sensor reading as valid, directly using it for emission compliance calculations. Inspection confirmed that the injection interruption was caused by an excessively low urea tank level.

[0047] Scenario 2: Catalyst Ammonia Storage Condition After the vehicle slowly entered the highway from the city, the SCR inlet temperature rose from 180℃ to 260℃. In the initial stage after the temperature entered the effective window, R_NH3 was +0.003, and η_SCR remained at 99%. The system determined it to be in "ammonia adsorption storage state," and no leak alarm was triggered, but dynamic monitoring of the R_NH3 trend began. As driving continued, R_NH3 gradually decreased to near 0, and η_SCR remained above 98%, with the system smoothly transitioning to "normal state."

[0048] like Figure 2 As shown, the present invention also provides a heavy-duty diesel vehicle ammonia leak identification and NOx measurement correction system, the steps of which implement the above-mentioned heavy-duty diesel vehicle ammonia leak identification and NOx measurement correction method include: The data acquisition and preprocessing module acquires the raw monitoring data uploaded by the vehicle to the monitoring platform and preprocesses the raw monitoring data. The raw monitoring data includes at least: vehicle speed, engine speed, net torque, fuel flow rate, intake air flow rate, NOx sensor concentration values ​​upstream and downstream of SCR, SCR inlet and outlet temperatures, engine coolant temperature, and atmospheric pressure.

[0049] The preprocessing of the raw monitoring data includes the following steps: The key parameters are calculated from the original monitoring data using the following formula: Excess air coefficient λ: ; SCR airspeed SV: NOx mass flow rate upstream of SCR : in, Intake airflow rate, unit: kg / h Fuel flow rate, unit: kg / h The values ​​are: exhaust mass flow rate (kg / h), R (gas constant), T_SCR (catalyst temperature, K), P (exhaust pressure, Pa), and V_SCR (catalyst volume, m³). 3 M_exh represents the exhaust molar mass in g / mol, and C_NOx_up represents the NOx concentration value from the upstream NOx sensor of the SCR in ppm.

[0050] A urea injection rate prediction module is used to receive preprocessed raw monitoring data and input it into a trained prediction model to predict the missing urea injection rate; the urea injection rate prediction module includes: The model training unit is used to collect a dataset containing the true value of the actual urea injection rate through vehicle-following tests, and only uses the data segment with the SCR inlet temperature higher than the urea injection temperature of the vehicle model for model training. 12-15 parameters strongly correlated with urea injection rate were selected from the dataset as input features of the LSTM model. These parameters strongly correlated with urea injection rate include: vehicle speed, net torque, engine speed, fuel flow rate, intake air flow rate, SCR inlet temperature, SCR outlet temperature, coolant temperature, upstream NOx concentration, excess air coefficient, SCR air velocity, and SCR upstream NOx mass flow rate. A model storage unit, connected to the model training unit, is used to store the trained LSTM model parameters; The real-time prediction unit is connected to the data acquisition and preprocessing module and the model storage unit, respectively, and is used to load the trained LSTM model to predict the urea injection rate in real time from new remote monitoring data without the true value of the urea rate.

[0051] The reducing agent excess characteristic value calculation module is used to define and calculate the reducing agent excess characteristic value R_NH3 based on the stoichiometric relationship between urea injection and NOx reaction, so as to quantify the degree of ammonia supply excess. The characteristic value of excess reducing agent R_NH3 is calculated using the following formula: R_NH3= (m_UWS - m_NOx_up × γ) / (δ × V_SCR) Where m_UWS is the cumulative urea mass within the sliding window (in g), m_NOx_up is the cumulative upstream NOx mass within the window (in g), V_SCR is the SCR catalyst volume (in L), γ is the theoretical urea to NOx mass ratio coefficient, and δ is the urea to ammonia conversion coefficient.

[0052] The ammonia leak detection module uses a dynamically sliding observation window to make a comprehensive judgment based on the excess reducing agent characteristic value R_NH3 and SCR efficiency within each window to identify ammonia leak events. A continuous confirmation unit, connected to the leakage determination unit, is used to introduce a continuous window confirmation mechanism, which requires that the leakage determination conditions be met for multiple consecutive windows before the start of the ammonia leakage event is finally confirmed. A reset determination unit, connected to the leakage determination unit, is used to set leakage reset conditions when... When η_SCR > preset efficiency threshold, the leakage event is considered to have ended.

[0053] For determining the non-ammonia leakage state, refer to Examples 1 and 2 above.

[0054] The NOx measurement correction module is connected to the ammonia leak identification module and is used to correct the NOx sensor readings downstream of the SCR during the identified ammonia leak period.

[0055] SCR efficiency within the calculation window The following formula is used for calculation: ; Wherein, C_NOx_down is the NOx concentration value of the downstream SCR sensor; C_NOx_up is the NOx concentration value of the upstream SCR sensor.

[0056] Using a preset fixed high-efficiency SCR conversion rate η_fixed, the actual downstream NOx concentration is calculated based on the upstream NOx concentration. The correction formula is as follows: ; Wherein, C_NOx_up is the NOx concentration value of the upstream NOx sensor of the SCR, and C_NOx_down_cor is the corrected NOx concentration value of the downstream SCR.

[0057] For details on parameter selection and calculation, please refer to the embodiments of the above methods, which will not be repeated here.

[0058] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles, characterized in that, Includes the following steps: S1. Obtain the raw monitoring data uploaded by the vehicle to the monitoring platform, and preprocess the raw monitoring data; S2. Input the preprocessed original monitoring data into the trained prediction model to predict the missing urea injection rate. S3. Based on the obtained stoichiometric relationship between all urea injection rates and NOx reactions, calculate the excess characteristic value R_NH3 of the reducing agent to quantify the degree of ammonia supply surplus. S4. Using a dynamically sliding observation window, within each observation window, the ammonia leakage period is identified based on the excess reducing agent characteristic value R_NH3 and the SCR efficiency; S5. During the identified ammonia leak period, the NOx sensor readings downstream of the SCR are corrected.

2. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 1, characterized in that, In S1, the raw monitoring data includes at least: vehicle speed, engine speed, net torque, fuel flow rate, intake air flow rate, NOx sensor concentration values ​​upstream and downstream of SCR, SCR inlet and outlet temperatures, engine coolant temperature, and atmospheric pressure.

3. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 1, characterized in that, In S1, the raw monitoring data is preprocessed, including the following steps: The key parameters are calculated from the original monitoring data using the following formula: Excess air coefficient λ: ; SCR airspeed SV: NOx mass flow rate upstream of SCR : in, Intake airflow rate, unit: kg / h; Fuel flow rate, unit: kg / h; The values ​​are: exhaust mass flow rate (kg / h); R is the gas constant; T_SCR is the catalyst temperature (K); P is the exhaust pressure (Pa); and V_SCR is the catalyst volume (m³). 3 M_exh represents the molar mass of the exhaust gas, in g / mol; C_NOx_up represents the NOx concentration value from the upstream NOx sensor of the SCR, in ppm.

4. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 3, characterized in that, In S2, the preprocessed raw monitoring data is input into the trained prediction model to predict the missing urea injection rate, including: A dataset containing true values ​​of actual urea injection rates was collected through vehicle-following tests. 12-15 parameters strongly correlated with urea injection rate were selected from the dataset as input features of the LSTM model. These parameters strongly correlated with urea injection rate include: vehicle speed, net torque, engine speed, fuel flow rate, intake air flow rate, SCR inlet temperature, SCR outlet temperature, coolant temperature, upstream NOx concentration, excess air coefficient, SCR air velocity, and SCR upstream NOx mass flow rate. The collected datasets were filtered, and only data segments with SCR inlet temperatures higher than the urea injection temperature of the vehicle model were retained for model training. Construct an LSTM network containing an input layer, two hidden layers, and an output layer as a prediction model; Collect measured data containing the true value of the actual urea injection rate over a period of time; train and construct a prediction model using the measured data; The trained prediction model is used to predict the urea injection rate from real-time monitoring data that does not contain the true value of the urea injection rate.

5. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 1, characterized in that, In S3: the excess characteristic value R_NH3 of the reducing agent is calculated using the following formula: R_NH3 = (m_UWS - m_NOx_up × γ) / (δ × V_SCR) Wherein, m_UWS is the cumulative urea mass within the sliding window, in g; m_NOx_up is the cumulative upstream NOx mass within the window, in g; V_SCR is the SCR catalyst volume, in L; γ is the theoretical urea to NOx mass ratio coefficient, ranging from 1.9 to 2.1; and δ is the urea to ammonia conversion coefficient, ranging from 5.3 to 5.

6.

6. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 1, characterized in that, S4 includes the following steps: A dynamically sliding observation window is used to observe the preprocessed raw monitoring data, and the SCR efficiency within the window is calculated. ; Within each sliding window, calculate the average reductant excess characteristic value R_NH3 within the window; When the SCR system is within its effective operating temperature range, the following process is used to determine its performance based on the SCR efficiency and excess reductant characteristic value within the obtained window: If R_NH3 > 0 and If a preset efficiency threshold is set, the current window is determined to be in an "ammonia leakage" state. Otherwise, it is judged as "normal".

7. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 6, characterized in that, When determining whether the current window is in an "ammonia leak" state, based on a continuous confirmation mechanism, an ammonia leak event is only confirmed to have started when two or more consecutive sliding windows meet the leak determination criteria; when... When η_SCR > preset efficiency threshold, the ammonia leak event is considered to have ended.

8. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 6, characterized in that... SCR efficiency within the calculation window The following formula is used for calculation: Wherein, C_NOx_down is the NOx concentration value of the downstream SCR sensor; C_NOx_up is the NOx concentration value of the upstream SCR sensor.

9. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 1, characterized in that, In S5, during the identified ammonia leak period, the NOx sensor readings downstream of the SCR are corrected, including: using a preset fixed high-efficiency SCR conversion rate η_fixed, the actual downstream NOx concentration is calculated based on the upstream NOx concentration, and the correction formula is: Where C_NOx_up is the NOx concentration value of the upstream NOx sensor of the SCR, and C_NOx_down_cor is the corrected NOx concentration value of the downstream SCR.

10. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 6, characterized in that, S4 also includes a comprehensive diagnosis of non-ammonia leak conditions: When the SCR system is within its effective operating temperature range like If η_SCR < preset efficiency threshold, it is determined to be a "non-ammonia fault state", triggering a check of the urea supply system or catalyst activity. like If η_SCR ≥ the preset efficiency threshold, it is determined to be in "ammonia adsorption storage state" and is used as an early warning state for ammonia leakage for dynamic tracking.

11. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 6, characterized in that, The length of the dynamically sliding observation window is 5 to 20 seconds, and the sliding step is 1 second; the effective operating temperature range is 200°C to 550°C.

12. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to claim 6, characterized in that, The preset efficiency threshold is 95% to 99.5%; the fixed high-efficiency SCR conversion rate η_fixed is 98% to 99.5%.

13. The method for identifying ammonia leaks and correcting NOx measurements in heavy-duty diesel vehicles according to any one of claims 1 to 12, characterized in that, The prediction model is a long short-term memory network model with a memory window of 60 seconds and 40-60 hidden layer nodes.

14. A heavy-duty diesel vehicle ammonia leak identification and NOx measurement correction system, characterized in that, The steps for implementing the heavy-duty diesel vehicle ammonia leak identification and NOx measurement correction method according to any one of claims 1-13 include: The data acquisition and preprocessing module acquires the raw monitoring data uploaded by the vehicle to the monitoring platform and preprocesses the raw monitoring data. The urea injection rate prediction module is used to receive preprocessed raw monitoring data and input it into the trained prediction model to predict the missing urea injection rate. The reducing agent excess characteristic value calculation module is used to define and calculate the reducing agent excess characteristic value R_NH3 based on the stoichiometric relationship between urea injection and NOx reaction, so as to quantify the degree of ammonia supply excess. The ammonia leak detection module uses a dynamically sliding observation window to make a comprehensive judgment based on the excess reducing agent characteristic value and SCR efficiency within each window to identify ammonia leak events. The NOx measurement correction module is connected to the ammonia leak identification module and is used to correct the NOx sensor readings downstream of the SCR during the identified ammonia leak period.

Images