Urea injection control method for tail gas aftertreatment of extended-range lean-burn engine

Abstract

The invention provides a urea injection control method for tail gas after-treatment of an extended-range lean-burn engine, which comprises the following steps: after a tail gas after-treatment system of the extended-range lean-burn engine is powered on and runs, starting a urea system to build pressure, and building the pressure of a urea pump to a first level; working condition parameters of the lean-burn engine are obtained, and an actual nitrogen-oxygen error e is calculated; constructing a feature vector X describing the current working condition of the engine according to the working condition parameters of the lean-burn engine and the actual nitrogen-oxygen error e, and inputting the feature vector X into the three-layer BP back-propagation neural network to obtain dynamically adjusted control parameters; determining a PID basic correction coefficient CPID based on the control parameter and the actual nitrogen-oxygen error e; performing long-term stability compensation on the basic correction coefficient CPID by fusing a temperature-time function to obtain a fusion correction coefficient C; and finally, according to the fusion correction coefficient C and the basic injection amount, the urea injection amount is dynamically adjusted in a zoning mode in combination with nitrogen and oxygen concentration data collected by a nitrogen and oxygen sensor, accurate control is achieved, and the balance between environmental protection standard reaching and fuel economy is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of engine exhaust treatment, and in particular to a urea injection control method for exhaust aftertreatment of an extended-range lean-burn engine. Background Art

[0002] As global energy consumption continues to rise, reserves of traditional fossil fuels like oil are gradually decreasing, and the problem of energy supply shortages is becoming increasingly prominent. Lean-burn technology allows engines to operate with a low gasoline content in the fuel mixture, improving fuel efficiency and reducing energy consumption. This is of great significance in alleviating the energy crisis. The use of lean-burn engines in extended-range vehicles can further improve the overall efficiency of energy utilization.

[0003] For example, in congested urban traffic conditions, extended-range vehicles can rely on pure electric mode to reduce fuel consumption; when traveling long distances or when the battery is low on power, the lean-burn engine starts to generate electricity, which is more energy-efficient and fuel-efficient than traditional fuel engines.

[0004] Lean-burn engines have a higher air-fuel ratio than traditional engines, offering advantages such as high combustion efficiency and excellent economic performance. Under lean-burn conditions, it's more difficult for the mixture to ignite than under the stoichiometric air-fuel ratio, making flash combustion even less likely. Therefore, a higher compression ratio design can be used to improve heat conversion efficiency and fully burn the gasoline in the excess air, extracting every drop of energy from each drop of gasoline. However, in terms of emissions, due to the low combustion temperature of the lean-burn mixture and the high oxygen content in the combustion products, the three-way catalytic converter cannot purify nitrogen oxides from the exhaust gases. This is because the three-way catalytic converter uses hydrocarbons (HC) or carbon monoxide (CO) in the exhaust to reduce nitrogen oxides. However, during lean-burn conditions, a high amount of oxygen remains in the exhaust gases, preventing an ideal nitrogen oxide reduction reaction.

[0005] The formation of nitrogen oxides is closely related to factors such as combustion temperature and oxygen concentration. In lean-burn engines, higher combustion temperatures and excess oxygen easily lead to the formation of large amounts of nitrogen oxides. Furthermore, lean-burn engines operate under complex conditions, with air-fuel ratios varying significantly under different operating conditions, making nitrogen oxide emission control even more difficult.

[0006] In traditional nitrogen oxide emission control methods, adjustments are made based on changes in nitrogen oxide values. For example, the urea injection amount is dynamically adjusted based on PI control strategies, PID control strategies, etc. However, the operating conditions of the lean-burn engine of an extended-range vehicle are complex and transiently changeable. Traditional PID control relies on historical error adjustment and is insufficiently responsive to nitrogen oxide emission fluctuations under transient conditions of the lean-burn engine (such as rapid acceleration and sudden load changes). This makes the traditional intelligent PID control strategy for adjusting the real-time injection amount of the urea injection system insufficiently adaptable to complex operating conditions. Summary of the Invention

[0007] In view of the fact that the existing technology for adjusting the urea injection amount by using PID control strategy for the exhaust aftertreatment of extended-range lean-burn engines has response lag and adaptability problems under complex working conditions, the purpose of the present invention is to provide a urea injection control method for the exhaust aftertreatment of extended-range lean-burn engines. Based on the PID control strategy, the weights of temperature, time, and nitrogen oxide errors in the control coefficient are automatically adjusted according to the real-time working conditions, and the temperature-time function is integrated to compensate the PID basic correction coefficient for long-term stability, thereby improving the dynamic response performance of the model and its adaptability to complex working conditions (transient, steady state, extreme working conditions, etc.), realizing instant feedback control of nitrogen oxide errors, and combining the real-time ambient temperature and system running time to supplement and correct the PID control results, thereby improving long-term stability.

[0008] According to a first aspect of the present invention, a method for controlling urea injection for exhaust aftertreatment of an extended-range lean-burn engine is provided, comprising the following steps:

[0009] After the extended-range lean-burn engine exhaust after-treatment system is powered on, the urea system pressure is started and the urea pump pressure is built up to the first level;

[0010] Obtain the operating parameters of the lean-burn engine, including engine state parameters, exhaust gas parameters, and environmental parameters, and calculate the actual NOx error e based on the front NOx concentration collected in real time by the front NOx sensor and the target NOx concentration of the rear NOx sensor;

[0011] According to the working condition parameters of the lean-burn engine and the actual nitrogen and oxygen error e, a characteristic vector X describing the current working condition of the engine is constructed and input into the three-layer BP back propagation neural network to realize the mapping output of the working condition characteristic vector input to the control parameters of the PID controller, and obtain the dynamically adjusted control parameters [K p , K i , K d ], respectively represent the proportional, integral and differential control coefficients;

[0012] Dynamic control parameters based on PID controller [K p , K i , K d ] and the actual nitrogen and oxygen error e, determine the PID controller basic correction coefficient C PID ;

[0013] Combine the real-time ambient temperature and system operation time, and integrate the temperature-time function to the basic correction coefficient C PID Perform long-term stability compensation to obtain the fusion correction coefficient C; and

[0014] The urea injection amount is dynamically adjusted based on the fusion correction coefficient C and the basic injection amount, combined with the nitrogen oxide concentration data partition collected by the front nitrogen oxide sensor.

[0015] As an optional embodiment, the pre-training process of the three-layer BP back propagation neural network includes bench simulation test, data label calibration and neural network training process, wherein:

[0016] First, typical operating conditions are simulated on an engine test bench. Sensors are used to collect engine state parameters, exhaust gas parameters, and environmental parameters under each operating condition, and the NOx error is determined. The typical operating conditions include:

[0017] Steady-state operating conditions: idling, economical cruising, and high-speed cruising;

[0018] Transient working conditions: rapid acceleration, rapid deceleration, and climbing;

[0019] Extreme working conditions: high altitude, high temperature, and extreme cold;

[0020] Then, for each operating point, the optimal PID parameters are determined through calibration and emission compliance back-calculation;

[0021] Next, a feature vector X describing the current engine operating condition is constructed based on the operating parameters at each operating point and the actual NOx error e. After normalization, this feature vector is fed into a three-layer BP back-propagation neural network for training. During training, an L2 regularization term is added to the hidden layer to suppress parameter overfitting, and model pruning is performed to remove connections in the hidden layer with an absolute weight value less than 0.1.

[0022] When the validation set loss during training does not decrease for 5 consecutive epochs, the training is terminated.

[0023] As an optional embodiment, the dynamic control parameter [K p , K i , K d ] and the actual nitrogen and oxygen error e, determine the PID controller basic correction coefficient C PID ,include:

[0024] C PID =K p e+K i ·∫edt+K d ·dt / de

[0025] Among them, K p , K i , K d Represent the proportional, integral, and differential control coefficients respectively.

[0026] As an optional embodiment, the real-time ambient temperature and system operating time are combined to fuse the temperature-time function to the basic correction coefficient C PID Perform long-term stability compensation to obtain the fusion correction coefficient C, including:

[0027] Combine the real-time ambient temperature and system operating time to determine the temperature-time function f(T amb ,t):

[0028] f(T amb ,t)=α1·T amb +α2·log(t+1)

[0029] Among them, T amb represents the ambient temperature collected in real time, t represents the running time of the system after power-on, α1 and α2 represent the preset temperature and time sensitivity coefficients respectively, which are calibrated through bench tests;

[0030] The basic correction coefficient C is calculated based on the temperature-time function. PID Perform long-term stability compensation to obtain the fusion correction coefficient C, specifically:

[0031] C=w1·C PID +w2·f(T amb ,t)

[0032] Among them, w1 and w2 represent the PID control weight and the temperature-time function weight respectively, and w1+w2=1.

[0033] As an optional embodiment, the PID control weight w1 and the temperature-time function weight w2 are set to be dynamically adjusted based on different working conditions:

[0034] Steady-state condition: w1 is less than or equal to 0.4;

[0035] Transient condition: w1 is greater than or equal to 0.5;

[0036] Extreme working conditions: w1 is greater than or equal to 0.6;

[0037] Moreover, w1+w2=1.

[0038] As an optional embodiment, the dynamically adjusting the urea injection amount according to the fusion correction coefficient C and the basic injection amount in combination with the nitrogen oxide concentration data partition collected by the front nitrogen oxide sensor includes:

[0039] (1) When A1≥β≥A2, that is, in the medium concentration range, the basic injection amount Q base Determine the actual injection quantity Q by combining with the correction coefficient C actual And continuously spray, where Q actual =C·Q base; and limit 150ml / h≥Q actual ≥10ml / h; Q base It is set to be calibrated according to the ideal operating conditions of the engine;

[0040] (2) When β is greater than A1, that is, in the high concentration range, the high-dose injection mode is triggered, with a Q of 1.2 to 1.5 times actual Determine the injection amount and inject continuously;

[0041] (3) When β is less than A2, that is, in the low concentration range, the pulse injection mode is triggered and a fixed injection amount is executed at the set time interval ΔT: [0.1, 0.5]*Q base

[0042] Among them, the aforementioned A1 and A2 respectively represent the upper and lower limit values ​​of the preset margin of nitrogen and oxygen concentration, and A1>λ·A2, λ represents the margin allowable intensity coefficient, λ≥2.

[0043] According to a second aspect of the present invention, a computer system is provided, comprising:

[0044] one or more processors; and

[0045] Memory, which stores instructions that can be operated;

[0046] When the instruction is executed by one or more processors, the one or more processors are caused to perform operations, including executing the process of the aforementioned method for controlling urea injection in exhaust aftertreatment of an extended-range lean-burn engine.

[0047] According to a third aspect of the present invention, a computer-readable storage medium is provided for storing one or more programs, wherein the one or more programs include instructions or instruction sets that can be executed by one or more processors, and when the instructions or instruction sets are executed by one or more processors, they perform the process of the aforementioned method for controlling urea injection in exhaust aftertreatment of an extended-range lean-burn engine.

[0048] Combining the above aspects, the urea injection control method for exhaust aftertreatment of extended-range lean-burn engines adopts the PID algorithm as the basic correction mechanism, and utilizes its proportional, integral, and differential adjustment characteristics of the error to ensure immediate feedback control of nitrogen oxide emission errors. At the same time, a three-layer BP neural network is introduced to dynamically optimize the PID parameters and combine dynamic weights to solve the defects and response lag of traditional PID parameters. The nonlinear mapping of the neural network is used to process the complex coupling relationship between engine operating conditions (speed, torque, temperature, etc.) and optimal control parameters, thereby improving the system's adaptability to transient, steady-state, and extreme operating conditions. Compared with existing technologies, its significant advantages are:

[0049] (1) By constructing multi-dimensional features (engine status, exhaust composition, environmental parameters, nitrogen and oxygen errors) to describe the real-time working conditions, real-time working condition mapping is achieved, and the response lag and working condition adaptability problems caused by the fixed parameters of traditional PID are solved, achieving accurate fitting of the working conditions and dynamic and accurate PID control system output;

[0050] (2) By integrating the immediate error response of the PID (processing the current emission deviation) with the long-term stability compensation of the temperature-time function (such as gradually increasing the injection amount during the warm-up process), the fluctuation amplitude of the nitrogen oxide concentration is reduced (compared with the fixed parameter PID), balancing short-term error control and long-term operating condition adaptation, and improving the compliance rate of the rear nitrogen oxide sensor concentration;

[0051] (3) At the same time, the interval injection control strategy is used, combined with the dynamic coefficient of fusion supplementation, to avoid excessive injection in the high concentration range (for example, the actual injection volume does not exceed 150ml / h at high load), and pulse injection is used in the low concentration range (executed at intervals of 0.1-0.5 times the basic volume) to adapt to the different operating conditions of the extended-range vehicle's lean-burn engine and the changes in nitrogen oxide emissions under different operating conditions, to achieve precise control, and significantly reduce urea consumption compared to traditional solutions, reduce urea pump power consumption, and improve engine fuel economy.

[0052] It should be understood that all combinations of the foregoing concepts and the additional concepts described in more detail below, as long as such concepts are not mutually inconsistent, can be considered part of the inventive subject matter of this disclosure. In addition, all combinations of the claimed subject matter are considered part of the inventive subject matter of this disclosure.

[0053] The foregoing and other aspects, embodiments, and features of the present invention will be more fully understood from the following description in conjunction with the accompanying drawings. Other additional aspects of the present invention, such as features and / or beneficial effects of the exemplary embodiments, will become apparent from the following description or through practice of specific embodiments according to the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In the accompanying drawings, each identical or nearly identical component shown in various figures may be represented by the same reference numeral. For clarity, not every component may be labeled in every figure.

[0055] Figure 1 The figure is a flow chart of a method for controlling urea injection in exhaust aftertreatment of a range-extended lean-burn engine according to an embodiment of the present invention. Figure 2 Schematic diagram of a lean-burn engine exhaust after-treatment urea injection control method according to a specific example of an embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to better understand the technical content of the present invention, specific embodiments are given and described below with reference to the accompanying drawings.

[0057] Various aspects of the present invention are described in this disclosure with reference to the accompanying drawings, in which a number of illustrative embodiments are shown. The embodiments of the present disclosure are not necessarily intended to include all aspects of the present invention. It should be understood that the various concepts and embodiments introduced above, as well as those described in more detail below, can be implemented in any of many ways, because the concepts and embodiments disclosed herein are not limited to any embodiment. In addition, some aspects of the present disclosure may be used alone or in any appropriate combination with other aspects disclosed herein.

[0058] {Example 1}

[0059] Combined with attachment Figure 1 As shown, according to an embodiment of the present invention, a method for controlling urea injection for exhaust aftertreatment of a range-extended lean-burn engine includes the following steps:

[0060] S1. After the extended-range lean-burn engine exhaust after-treatment system is powered on, the urea system is started to build pressure, and the urea pump pressure is built up to the first level;

[0061] S2. Obtain operating parameters of the lean-burn engine, including engine state parameters, exhaust gas parameters, and environmental parameters, and calculate an actual nitrogen oxide error e based on the front nitrogen oxide concentration collected in real time by the front nitrogen oxide sensor and the target nitrogen oxide concentration of the rear nitrogen oxide sensor;

[0062] S3. Construct a feature vector X describing the current working condition of the engine based on the working condition parameters of the lean-burn engine and the actual nitrogen and oxygen error e, input it into the three-layer BP back propagation neural network, realize the mapping output of the working condition feature vector input to the control parameters of the PID controller, and obtain the dynamically adjusted control parameters [K p , K i , K d ], respectively represent the proportional, integral and differential control coefficients;

[0063] S4. Dynamic control parameters based on PID controller [K p , K i , K d ] and the actual nitrogen and oxygen error e, determine the PID controller basic correction coefficient C PID ;

[0064] S5. Combine the real-time ambient temperature and system operation time to integrate the temperature-time function to the basic correction coefficient C PID Perform long-term stability compensation to obtain the fusion correction coefficient C; and

[0065] S6. Dynamically adjust the urea injection amount according to the fusion correction coefficient C and the basic injection amount in combination with the nitrogen oxide concentration data collected by the front nitrogen oxide sensor.

[0066] Therefore, by constructing real-time multi-dimensional features (engine status, exhaust composition, environmental parameters, and NOx errors) to describe real-time operating conditions, real-time operating condition mapping is achieved, enabling precise fitting of operating conditions and dynamic, precise PID control system output. This addresses the response lag and operating condition adaptability issues caused by traditional fixed PID parameters. Furthermore, by combining temperature-time functions to provide long-term stability compensation for basic PID parameters, the fluctuation range of NOx concentration is reduced.

[0067] In an embodiment of the present invention, a segmented injection control strategy is used, combined with a fusion-supplemented dynamic coefficient, to avoid excessive injection in high-concentration intervals, and pulse injection with a fixed injection volume is used in low-concentration intervals to adapt to different operating conditions of the lean-burn engine of an extended-range vehicle and changes in nitrogen oxide emissions under different operating conditions, thereby achieving precise control. Compared with traditional solutions, this strategy significantly reduces urea consumption, reduces urea pump power consumption, and improves engine fuel economy.

[0068] As an optional embodiment, in the aforementioned method S1, when the extended-range lean-burn engine exhaust aftertreatment system is in operation, the urea system is activated upon power-up, and the urea pump builds pressure to 6 bar and waits. When the engine speed, current torque, cooling water temperature, and intake air volume meet the set conditions, the urea injection system begins operation.

[0069] As an optional embodiment, in the aforementioned method S2, the engine state parameters in the operating parameters of the lean-burn engine include: speed, torque, cooling water temperature, intake air volume, throttle opening, and cylinder temperature;

[0070] The aforementioned exhaust gas parameters include the front NOx concentration collected by the front NOx sensor, the rear NOx concentration collected by the rear NOx sensor, the HC concentration, and the CO concentration;

[0071] The aforementioned environmental parameters include atmospheric pressure and ambient temperature.

[0072] In an embodiment of the present invention, the operating parameters of the aforementioned lean-burn engine are configured to be acquired through a vehicle ECU and / or a vehicle-mounted sensor bus network.

[0073] The collected input parameters (original signals) of the operating parameters of the lean-burn engine are as follows:

[0074] Engine status: speed n (rpm), torque T q (N·m), cooling water temperature T cool (℃), intake air volume m a (g / s), throttle opening θ (°), cylinder temperature T cyl(K);

[0075] Exhaust parameters: front NOx concentration β (ppm) collected by the front NOx sensor, rear NOx concentration β target (target value, ppm), HC concentration c HC (ppm), CO concentration c CO (%);

[0076] Environmental parameters: atmospheric pressure P atm (kPa), ambient temperature T amb (℃).

[0077] By collecting real-time working conditions, it comprehensively reflects the engine combustion status, exhaust gas composition and environmental impact, and is the basic input for neural network and PID control.

[0078] Furthermore, for signals with good volatility (such as intake volume and oxygen concentration), Kalman filtering can be used to remove high-frequency noise and perform signal preprocessing.

[0079] Then, the parameters of different dimensions are normalized and uniformly mapped to the interval [0, 1].

[0080] Furthermore, the actual nitrogen oxide error e is calculated based on the front nitrogen oxide concentration collected in real time by the front nitrogen oxide sensor and the target nitrogen oxide concentration of the rear nitrogen oxide sensor, including:

[0081] Actual NOx error e=β target -β

[0082] Among them, β represents the front nitrogen oxide concentration collected in real time by the front nitrogen oxide sensor, reflecting the current emission level; β targe Indicates the target nitrogen oxide concentration of the rear nitrogen oxide sensor, which is preset according to environmental regulations and engine operating conditions. For example, the National VI RDE standard requires NO x ≤35ppm.

[0083] The actual nitrogen oxide error e obtained in this way serves as the core input of PID control. A positive value indicates that the emission exceeds the standard and urea injection needs to be increased; a negative value indicates that the emission is insufficient and injection can be reduced.

[0084] Therefore, PID control based on NOx error provides basic injection quantity correction to ensure a quick response to real-time errors.

[0085] In the aforementioned method S3, a feature vector X describing the current working condition of the engine is constructed based on the working condition parameters of the lean-burn engine and the actual nitrogen and oxygen error e. The engine state, environment, exhaust gas composition and error are integrated to fully describe the current working condition. The three-layer BP back propagation neural network is input to realize the mapping output of the working condition feature vector input to the control parameters of the PID controller, and the dynamically adjusted control parameters [K p, K i , K d ], representing the proportional, integral and differential control coefficients respectively.

[0086] Through the real-time output of the neural network, the PID parameters are adapted to different working conditions such as idling, acceleration, and high load. For example, when accelerating sharply, the K p Improve response speed and increase K during steady-state cruising i Eliminate long-term errors.

[0087] It should be understood that in PID control, the proportional control coefficient K p The purpose is to amplify the impact of the current error, quickly adjust the injection amount, and avoid response lag. Integral control coefficient K i The purpose is to accumulate historical errors and eliminate steady-state emission deviations (such as slight excesses after long periods of idling). The differential control coefficient K d The aim is to suppress the error change rate and prevent excessive injection volume fluctuations during rapid acceleration. Adaptive control is achieved through the coordinated control of the three.

[0088] As an embodiment of the present invention, the network structure of the three-layer BP back propagation neural network includes an input layer, a hidden layer and an output layer, wherein:

[0089] The input layer is 12-dimensional and receives the pre-processed working condition feature vector;

[0090] The hidden layer is 24-dimensional and is used to extract the mapping relationship between the operating condition characteristics and the optimal PID parameters through nonlinear transformation. The number of neurons is determined by cross-validation, and the ReLU activation function is used to solve the gradient disappearance problem.

[0091] The output layer is 3-dimensional and is used to output dynamic PID parameters [K p , K i , K d ], using linear function (directly output continuous values).

[0092] It should be understood that after the input feature vector X passes through the input layer, the following calculation is performed in the hidden layer:

[0093] h=ReLU(W1X+b1)

[0094] Among them, W1 is the weight matrix (12*24) from the input layer to the hidden layer, and b1 is the hidden layer bias (24 dimensions).

[0095] The hidden layer output h is passed to the output layer:

[0096]

[0097] Among them, W2 is the weight matrix (12*24) from the hidden layer to the output layer, and b2 is the output layer bias (3D).

[0098] The pre-training process of the three-layer BP back-propagation neural network includes bench simulation test, data label calibration and neural network training process:

[0099] First, typical operating conditions are simulated on an engine test bench. Sensors are used to collect engine state parameters, exhaust gas parameters, and environmental parameters under each operating condition, and the NOx error is determined. The aforementioned typical operating conditions include:

[0100] Steady-state operating conditions: idling, economic cruising (60-90km / h), high-speed cruising (100-120km / h);

[0101] Transient operating conditions: rapid acceleration (0-100km / h), rapid deceleration, and climbing (slope ≥ 15%);

[0102] Extreme working conditions: high-cold simulation (ambient temperature below -20°C), plateau simulation (altitude ≥ 2000m), high-temperature simulation (ambient temperature ≥ 38°C);

[0103] Then, for each operating point, the optimal PID parameters are determined by calibration and emission standard back-calculation as label data. For example, first determine a set of K p , K i , K d , adjust the injection amount until the concentration of the rear nitrogen oxide sensor reaches the standard (β targe t ≥β); then record the parameters at this time as labels

[0104] Next, a feature vector X describing the current engine operating condition is constructed based on the operating parameters at each operating point and the actual NOx error e. After normalization, this feature vector is fed into a three-layer BP back-propagation neural network for training. During training, an L2 regularization term is added to the hidden layer to suppress parameter overfitting, and model pruning is performed to remove connections in the hidden layer with an absolute weight value less than 0.1.

[0105] When the validation set loss during training does not decrease for 5 consecutive epochs, the training is terminated (to avoid overfitting).

[0106] As an optional implementation, in step S4 of the above method, the dynamic control parameter [K p , K i , K d ] and the above-mentioned actual nitrogen and oxygen error e, determine the basic correction coefficient C of the PID controller PID ,include:

[0107] C PID =K p e+K i·∫edt+K d ·dt / de

[0108] Among them, K p , K i , K d Represent the proportional, integral, and differential control coefficients respectively.

[0109] Furthermore, in step S5, the temperature-time function is integrated with the real-time ambient temperature and the system operation time to calculate the basic correction coefficient C PID Perform long-term stability compensation to obtain the fusion correction coefficient C, including:

[0110] Combine the real-time ambient temperature and system operating time to determine the temperature-time function f(T amb ,t):

[0111] f(T amb ,t)=α1·T amb +α2·log(t+1)

[0112] Among them, T amb represents the ambient temperature collected in real time, t represents the running time of the system after power-on, α1 and α2 represent the preset temperature and time sensitivity coefficients respectively, which are calibrated through bench tests;

[0113] The basic correction coefficient C is calculated based on the temperature-time function. PID Perform long-term stability compensation to obtain the fusion correction coefficient C, specifically:

[0114] C=w1·C PID +w2·f(T amb ,t)

[0115] Among them, w1 and w2 represent the PID control weight and the temperature-time function weight respectively, and w1+w2=1.

[0116] Therefore, the PID control results are supplemented and corrected based on the real-time ambient temperature and system operating time. On the one hand, the urea hydrolysis efficiency is low at low temperatures, and the injection amount needs to be appropriately increased. On the other hand, the system is in the warm-up stage (t<100s) within a certain period of time after power-on, and the injection amount needs to be gradually increased.

[0117] For example, when w1 is much larger than w2 (e.g. transient conditions): PID control is used as the main method to quickly respond to error changes;

[0118] When w2 is much larger than w1 (such as steady-state conditions): slowly adjust the temperature and time to avoid over-injection.

[0119] Therefore, the short-term error control and long-term working condition adaptation are balanced. For example, in a low temperature environment, even if the error is zero, it will be controlled by f(Tamb ,t) Fine-tune the injection amount to ensure sufficient urea hydrolysis.

[0120] As an optional implementation, the aforementioned PID control weight w1 and the temperature-time function weight w2 are set to be dynamically adjusted based on different working conditions:

[0121] Steady-state condition: w1 is less than or equal to 0.4, the temperature-time function dominates, and injection fluctuations caused by high-frequency adjustments are avoided;

[0122] Transient operating conditions: w1 is greater than or equal to 0.5, PID control is dominant, and it responds quickly to sudden increases in nitrogen and oxygen concentrations;

[0123] Extreme working conditions: w1 is greater than or equal to 0.6, ensuring that PID control is dominant, quickly responding to sudden increases in nitrogen and oxygen concentrations, and ensuring the injection amount in the high injection range;

[0124] Moreover, w1+w2=1.

[0125] Furthermore, in step S6, the urea injection amount is adjusted based on the coefficient C, and the correction coefficient is converted into the actual injection amount. According to the previous nitrogen oxide concentration segmentation logic, combined with the dynamic correction of the coefficient C, the urea injection amount is dynamically controlled and adjusted under different operating conditions of the lean-burn engine to achieve a balance between environmental protection standards and fuel economy.

[0126] In an embodiment of the present invention, the urea injection amount is dynamically adjusted based on the aforementioned fusion correction coefficient C and the basic injection amount, combined with the nitrogen oxide concentration data partition collected by the front nitrogen oxide sensor, including:

[0127] (1) When A1≥β≥A2, that is, in the medium concentration range, the basic injection amount Q base Determine the actual injection quantity Q by combining with the correction coefficient C actual And continuously spray, where Q actual =C·Q base ; and limit 150ml / h≥Q actual ≥10ml / h to avoid excessive injection leading to catalyst crystallization or nozzle clogging; the aforementioned Q base It is set to be calibrated according to the ideal operating conditions of the engine;

[0128] (2) When β is greater than A1, that is, in the high concentration range, the high-dose injection mode is triggered, with a Q of 1.2 to 1.5 times actual Determine the injection amount and inject continuously;

[0129] (3) When β is less than A2, that is, in the low concentration range, the pulse injection mode is triggered and a fixed injection amount is executed at the set time interval ΔT;

[0130] Among them, the aforementioned A1 and A2 respectively represent the upper and lower limit values of the preset margin of nitrogen and oxygen concentration, and A1 > λ·A2, where λ represents the margin allowable strength coefficient and λ≥2.

[0131] Among them, the aforementioned fixed injection amount is set to: [0.1, 0.5]*Q base .

[0132] In this embodiment, A1 is set to 100 ppm and A2 is set to 45 ppm.

[0133] Thus, on the basis of achieving precise PID control adjustment, by using the nitrogen and oxygen concentration segmented control logic, a hierarchical control of global strategy + local correction is achieved, and precise control of urea injection for a lean-burn engine under different working conditions is realized.

[0134] {Example 2}

[0135] Combined with the attached Figure 2 As shown, for the urea injection control method for the exhaust gas aftertreatment of a lean-burn engine according to a specific example of the present invention, when the exhaust gas aftertreatment system of the range-extended lean-burn engine is working, after power-on, the urea system is started, and the urea pump builds pressure to 6 bar and waits.

[0136] When the engine speed, current torque, coolant water temperature, and intake air volume meet the set conditions (at this time ξ = 1), the urea injection system starts to work.

[0137] Adjust the injection amount according to the size of the front nitrogen and oxygen sensor data β in different zones:

[0138] When A1≥β≥A2 ppm, inject B2 ml / h of urea and correct it in real time according to the adaptive adjustment coefficient C (determined by dynamic adjustment according to the real-time working condition);

[0139] If β < A2 ppm, turn on the fixed injection amount injection mode, and perform pulsed (intermittent) quantitative B3 injection at intervals;

[0140] When β > A1 ppm, perform continuous injection of B1 ml / h of urea.

[0141] After the engine stops, urea injection stops. After the vehicle is powered off in T15, the urea pump relieves pressure, and the system is powered off after 10 seconds.

[0142] In this example, B1 is usually 1.2 to 1.५ times of B2 (as the basic amount) to achieve a linear response of nitrogen and oxygen concentration - injection amount.

[0143] In particular, when it is at low temperature, C > 1 to compensate for the decrease in hydrolysis efficiency; when it is at high temperature, C≈1 to maintain the efficiency stability.

[0144] {Example 3}

[0145] In conjunction with the implementation of the above embodiment of the urea injection control method for exhaust aftertreatment of an extended-range lean-burn engine, the present invention further proposes a computer system comprising:

[0146] one or more processors; and

[0147] Memory stores instructions that can be operated.

[0148] When the instruction is executed by one or more processors, the one or more processors are caused to perform operations, including the process of executing the aforementioned embodiment of the method for controlling urea injection in exhaust aftertreatment of an extended-range lean-burn engine.

[0149] {Example 4}

[0150] In conjunction with the implementation of the urea injection control method for exhaust aftertreatment of a range-extended lean-burn engine according to the above embodiment, the present invention further provides a computer-readable storage medium for storing one or more programs.

[0151] The one or more programs include instructions or instruction sets that can be executed by one or more processors, and when these instructions or instruction sets are executed by one or more processors, they execute the process of the aforementioned embodiment of the method for controlling urea injection for exhaust aftertreatment of an extended-range lean-burn engine.

[0152] While the present invention has been disclosed above with reference to preferred embodiments, this is not intended to limit the present invention. Persons skilled in the art will readily appreciate that various modifications and variations can be made without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

Claims

1. A method for controlling urea injection for exhaust aftertreatment of a range-extended lean-burn engine, characterized in that: The following steps are involved: After the extended-range lean-burn engine exhaust after-treatment system is powered on, the urea system pressure is started and the urea pump pressure is built up to the first level; Obtain the operating parameters of the lean-burn engine, including engine state parameters, exhaust gas parameters, and environmental parameters, and calculate the actual NOx error e based on the front NOx concentration collected in real time by the front NOx sensor and the target NOx concentration of the rear NOx sensor; According to the working condition parameters of the lean-burn engine and the actual nitrogen and oxygen error e, a characteristic vector X describing the current working condition of the engine is constructed and input into the three-layer BP back propagation neural network to realize the mapping output of the working condition characteristic vector input to the control parameters of the PID controller, and obtain the dynamically adjusted control parameters [K p , K i , K d ], respectively represent the proportional, integral and differential control coefficients; Dynamic control parameters based on PID controller [K p , K i , K d ] and the actual nitrogen and oxygen error e, determine the PID controller basic correction coefficient C PID ; Combine the real-time ambient temperature and system operation time, and integrate the temperature-time function to the basic correction coefficient C PID Perform long-term stability compensation to obtain the fusion correction coefficient C; as well as The urea injection amount is dynamically adjusted based on the fusion correction coefficient C and the basic injection amount, combined with the nitrogen oxide concentration data partition collected by the front nitrogen oxide sensor.

2. The urea injection control method for exhaust aftertreatment of a range-extended lean-burn engine according to claim 1, characterized in that: The engine state parameters in the operating parameters of the lean-burn engine include: speed, torque, cooling water temperature, intake air volume, throttle opening and cylinder temperature; The exhaust gas parameters include the front nitrogen oxide concentration collected by the front nitrogen oxide sensor, the rear nitrogen oxide concentration collected by the rear nitrogen oxide sensor, the HC concentration and the CO concentration; The environmental parameters include atmospheric pressure and ambient temperature; The operating parameters of the lean-burn engine are configured to be acquired through a vehicle ECU and / or a vehicle-mounted sensor bus network.

3. The urea injection control method for exhaust aftertreatment of a range-extended lean-burn engine according to claim 1, characterized in that: Calculating the actual nitrogen oxide error e based on the front nitrogen oxide concentration collected in real time by the front nitrogen oxide sensor and the target nitrogen oxide concentration of the rear nitrogen oxide sensor includes: Actual NOx error e=β target -β Among them, β represents the front nitrogen and oxygen concentration collected in real time by the front nitrogen and oxygen sensor, β targe Indicates the target NOx concentration of the rear NOx sensor, which is preset according to environmental regulations and engine operating conditions.

4. The urea injection control method for exhaust aftertreatment of a range-extended lean-burn engine according to claim 1, characterized in that: The network structure of the three-layer BP back propagation neural network includes an input layer, a hidden layer and an output layer, wherein: The input layer is 12-dimensional and receives the pre-processed working condition feature vector; The hidden layer is 24-dimensional and is used to extract the mapping relationship between the operating condition characteristics and the optimal PID parameters through nonlinear transformation. The number of neurons is determined by cross-validation and the ReLU activation function is used. The output layer is 3-dimensional and is used to output dynamic PID parameters [K p , K i , K d ].

5. The urea injection control method for exhaust after-treatment of a range-extended lean-burn engine according to claim 4, characterized in that: The pre-training process of the three-layer BP back propagation neural network includes bench simulation test, data label calibration and neural network training process, wherein: First, typical operating conditions are simulated on an engine test bench. Sensors are used to collect engine state parameters, exhaust gas parameters, and environmental parameters under each operating condition, and the NOx error is determined. The typical operating conditions include: Steady-state operating conditions: idling, economical cruising, and high-speed cruising; Transient working conditions: rapid acceleration, rapid deceleration, and climbing; Extreme working conditions: high altitude, high temperature, and extreme cold; Then, for each operating point, the optimal PID parameters are determined through calibration and emission compliance back-calculation; Next, a feature vector X describing the current engine operating condition is constructed based on the operating parameters at each operating point and the actual NOx error e. After normalization, this feature vector is fed into a three-layer BP back-propagation neural network for training. During training, an L2 regularization term is added to the hidden layer to suppress parameter overfitting, and model pruning is performed to remove connections in the hidden layer with an absolute weight value less than 0.

1. When the validation set loss during training does not decrease for 5 consecutive epochs, the training is terminated.

6. The method for controlling urea injection for exhaust aftertreatment of a range-extended lean-burn engine according to claim 1, characterized in that: The dynamic control parameters based on the PID controller [K p , K i , K d ] and the actual nitrogen and oxygen error e, determine the PID controller basic correction coefficient C PID ,include: C PID =K p ·e+K i ·∫edt+K d ·dt / de Among them, K p , K i , K d Represent the proportional, integral, and differential control coefficients respectively.

7. The method for controlling urea injection for exhaust aftertreatment of a range-extended lean-burn engine according to claim 1, characterized in that: The combination of real-time ambient temperature and system operating time, the fusion temperature-time function is used to correct the basic correction coefficient C PID Perform long-term stability compensation to obtain the fusion correction coefficient C, including: Combine the real-time ambient temperature and system operating time to determine the temperature-time function f(T amb ,t): f(T amb ,t)=α1·T amb +α2·log(t+1) Among them, T amb represents the ambient temperature collected in real time, t represents the running time of the system after power-on, α1 and α2 represent the preset temperature and time sensitivity coefficients respectively, which are calibrated through bench tests; The basic correction coefficient C is calculated based on the temperature-time function. PID Perform long-term stability compensation to obtain the fusion correction coefficient C, specifically: C=w1·C PID +w2·f(T amb ,t) Among them, w1 and w2 represent the PID control weight and the temperature-time function weight respectively, and w1+w2=1.

8. The method for controlling urea injection for exhaust aftertreatment of a range-extended lean-burn engine according to claim 7, characterized in that: The PID control weight w1 and the temperature-time function weight w2 are set to be dynamically adjusted based on different working conditions: Steady-state condition: w1 is less than or equal to 0.4; Transient condition: w1 is greater than or equal to 0.5; Extreme working conditions: w1 is greater than or equal to 0.6; Moreover, w1+w2=1.

9. The method for controlling urea injection for exhaust aftertreatment of a range-extended lean-burn engine according to claim 1, characterized in that: The method of dynamically adjusting the urea injection amount according to the fusion correction coefficient C and the basic injection amount in combination with the nitrogen oxide concentration data partition collected by the front nitrogen oxide sensor includes: (1) When A1≥β≥A2, that is, in the medium concentration range, the basic injection amount Q base Determine the actual injection quantity Q by combining with the correction coefficient C actual And continuously spray, where Q actual =C·Q base ; and limit 150ml / h≥Q actual ≥10ml / h; Q base It is set to be calibrated according to the ideal operating conditions of the engine; (2) When β is greater than A1, that is, in the high concentration range, the high-dose injection mode is triggered, with a Q of 1.2 to 1.5 times actual Determine the injection amount and inject continuously; (3) When β is less than A2, that is, in the low concentration range, the pulse injection mode is triggered and a fixed injection amount is executed at the set time interval ΔT; Among them, the aforementioned A1 and A2 respectively represent the upper and lower limit values ​​of the preset margin of nitrogen and oxygen concentration, and A1>λ·A2, λ represents the margin allowable intensity coefficient, λ≥2.

10. The urea injection control method for exhaust aftertreatment of a range-extended lean-burn engine according to claim 9, characterized in that: The fixed injection amount is set to: [0.1, 0.5]*Q base .

Images