Engine start injection quantity control method

Abstract

The present application relates to the technical field of automobile control, and particularly relates to an engine starting fuel injection amount control method. M related parameters of engine operation fuel injection amount are used to define an M-dimensional space, and K self-learning intervals of neurons are established based on the M-dimensional space; the probability Z NORM [k] of engine operation in each neuron interval is calculated; the increment value Z out of the basic fuel injection amount is calculated and output according to a pre-constructed neural network model; wherein, the synaptic parameter weight factor W k (m) is iteratively operated, and after each iteration round, the new W k (m) is re-input into the neural network model, and Z out is updated; when the engine starts, Z out is used to correct the basic fuel injection amount, and the corrected value is used as the engine starting fuel injection amount. The LAMBDA correction coefficient can be continuously calculated and updated in the whole vehicle life cycle, the control based on the correction coefficient achieves the optimal fuel injection amount at the start, effectively improves the starting combustion condition, and can successfully start and reduce the starting fuel consumption.

Description

Technical Field

[0001] This invention relates to the field of automotive control technology, and more specifically to a method for controlling the amount of fuel injected during engine startup. Background Technology

[0002] The LAMBDA (Air-Fuel Ratio-to-Delta) control of the engine control system is crucial, significantly impacting emissions control, vehicle power, and fuel economy. Traditional control strategies primarily utilize two-dimensional adjustments based on engine speed and load, or employ self-learning to divide the system into several two-dimensional intervals based on operating conditions. Within each interval, appropriate calibration is applied, followed by proportional or differential corrections based on varying external environments. This correction method is limited to interpolation, resulting in a complex control strategy and a massive calibration workload. For fuel quality variations, it relies solely on closed-loop fuel injection differential (PDI) control, offering no protection during cold starts, warm-up, or open-loop control. Furthermore, significant differences in injector quality, engine component aging, or fuel quality can lead to the closed-loop correction coefficient exceeding its maximum or minimum value. This results in slow system adjustment, an inability to quickly correct the LAMBDA value, and ultimately, excessively rich or lean air-fuel ratios, causing misfires and deteriorating emissions. Summary of the Invention

[0003] The purpose of this invention is to address the shortcomings of existing technologies by providing an engine start-up fuel injection quantity control method. Based on a neural network self-learning strategy, the system can continuously calculate and learn the LAMBDA correction coefficient throughout the vehicle's life cycle and continuously update it. During engine start-up, the LAMBDA correction coefficient enriches or thins the base fuel injection quantity to achieve the optimal injection quantity during start-up, effectively improving the start-up combustion condition. This ensures successful start-up while reducing start-up fuel consumption and effectively reducing initial emissions.

[0004] This invention provides a method for controlling the amount of fuel injected during engine startup, comprising:

[0005] An M-dimensional space is defined by M relevant parameters of the engine's fuel injection quantity, and K self-learning intervals of neurons are established based on the M-dimensional space, with each self-learning interval corresponding to a neuron.

[0006] Calculate the probability Z of the engine operating within each neuron interval. NORM [k], where k is any natural number from 1 to K;

[0007] The incremental value Z of the base injection quantity is calculated and output based on a pre-built neural network model. out ,in, In the summation calculation, the value of m takes values ​​sequentially from the interval of the first relevant parameter to the interval of the Mth relevant parameter, W k[1] is the weighting factor of the synaptic parameter corresponding to the first relevant parameter, W norm (m) is a standardized constant value, and x(m) is the input value of m under different values;

[0008] Weighting factor W for synaptic parameters k (m) performs iterative calculations, and after each iteration, the new W is generated. k (m) Re-input the neural network model to Z out Update;

[0009] When starting the engine, use Z out The base fuel injection quantity is corrected, and the corrected value is used as the fuel injection quantity for engine startup.

[0010] Preferably, the relevant parameters include engine coolant temperature T, engine speed N, and engine load R, and the defined M-dimensional space is a three-dimensional space.

[0011] Preferably, the K self-learning intervals are intervals between various typical operating conditions in three-dimensional space. The typical operating conditions are coordinate points (T, N, R) formed in three-dimensional space with the values ​​of engine coolant temperature T, engine speed N, and engine load R as coordinates. The engine coolant temperature T, engine speed N, and engine load R each include multiple values, wherein the unit of engine coolant temperature T is °C, the unit of engine speed N is rmp, and the unit of engine load R is N·m.

[0012] More preferably, the typical operating points include (90, 1200, 25), (90, 5000, 50), (90, 2500, 50), (90, 5250, 90), (90, 2000, 200), (90, 5500, 200), (50, 1200, 25), (60, 2250, 60), (50, 5000, 30), (60, 750, 100), (60, 5250, 90), (30, 1200, 25), (30, 2250, 60), (30, 5000, 30), (20, 750, 100).

[0013] Preferably, the probability Z of the engine operating within each neuron interval is... NORM [k] is calculated using the following formula:

[0014]

[0015] Where x(m) follows a normal distribution, C k [m] represents the amplitude at the center of the normal distribution, σ k [m] represents the variance of the normal distribution.

[0016] Preferably, the engine operates within the sum of probabilities of each neuron interval.

[0017] Preferably, the synaptic parameter weighting factor W k (m) Iteratively calculate according to the following formula:

[0018]

[0019] Fak is the adjustment factor for a traditional closed-loop PID controller for fuel systems.

[0020] More preferably, the incremental value Z out The value is either positive or negative, and the Z value is used. out Correcting the base fuel injection quantity means adding the positive or negative value to the base fuel injection quantity.

[0021] The beneficial effects of this invention are as follows:

[0022] 1. This method incorporates fuel self-learning, based on a neural network self-learning strategy. Throughout the vehicle's lifecycle, the system continuously calculates and learns the LAMBDA correction coefficient and updates it, achieving better and more precise pre-control. It can successfully start the engine while reducing fuel injection and lowering fuel consumption, offering broader adaptability to different engine types and different stages of engine use. This method constructs an M-dimensional space using M parameters related to the engine's starting fuel injection quantity, establishes K self-learning intervals for neurons based on this M-dimensional space, and establishes the incremental value Z of the base fuel injection quantity. out The neural network model calculates the probability Z of the engine operating within each neuron interval. NORM [k] Calculate Z out The value is calculated using the synaptic parameter weighting factor W. k (m) This value is iteratively updated to achieve fuel self-learning. This method has a unique model structure and inherent nonlinear simulation capability, as well as highly adaptive fault-tolerant characteristics, and can model any nonlinear physical model. This system can take into account multiple input dimensions, making the engine operating conditions more detailed and the control more precise. During engine startup, the LAMBDA correction coefficient enriches or leans the base fuel injection quantity at startup to achieve the optimal injection quantity, effectively improving the startup combustion condition. This ensures successful startup, reduces startup fuel consumption, effectively reduces raw emissions, and has broader adaptability to different individual engines and different stages of engine use.

[0023] 2. A three-dimensional neural network is constructed using engine load R, engine speed N, and engine coolant temperature T, and then self-learned. This approach achieves high accuracy while maintaining high learning efficiency, better reflecting actual vehicle operating conditions. outThe revised fuel injection quantity is better suited to the vehicle's current operating conditions. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the process of the present invention;

[0025] Figure 2 This is a schematic diagram illustrating the calculation of the incremental value of the basic fuel injection quantity in this invention. Detailed Implementation

[0026] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0027] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0028] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0029] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0030] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0031] References to "one embodiment" or "some embodiments" in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized. "A plurality" means "two or more."

[0032] The present invention provides an engine start-up fuel injection quantity control method, which specifically sets a LAMBDA correction coefficient. This coefficient takes into account the influence of five factors on the mixture concentration during engine combustion: intake air quantity error or leakage, fuel density or fuel rail pressure error, tolerance of injectors and other components, pipeline aging and carbon deposits, and the difference between the ethanol content in the fuel and the reference value. The mixture is controlled by enriching or thinning the base fuel injection quantity.

[0033] The implementation method utilizes the probe variables during engine operation to establish a neural network self-learning model. The model consists of interconnected neurons forming a non-cyclic system—the fuel self-learning model. The outputs of some neurons serve as the inputs of other neurons, forming a layered connection. Each layer contains multiple neurons. Each layer is further divided into an input layer, a hidden layer, and an output layer.

[0034] Example 1

[0035] Figure 1 A preferred embodiment of this application is shown. Figure 1 The diagram shows a schematic flow chart of an engine start-up fuel injection quantity control method according to the first embodiment of this application. For ease of explanation, only the parts related to this embodiment are shown, and the details are as follows:

[0036] When the engine starts, LAMBDA correction is not entered into the closed-loop control. The engine system control input values ​​have multiple dimensions. The establishment of the self-learning model can be divided into the following steps:

[0037] Step 1: Define an M-dimensional space based on M relevant parameters of the engine's fuel injection quantity, and establish K self-learning intervals for neurons based on the M-dimensional space, wherein each self-learning interval corresponds one-to-one with a neuron.

[0038] Step 2: Calculate the probability Z of the engine operating within each neuron interval.NORM [k], where k is any natural number from 1 to K;

[0039] Step 3: Calculate and output the incremental value Z of the base injection quantity based on the pre-built neural network model. out ,in, In the summation calculation, the value of m takes values ​​sequentially from the interval of the first relevant parameter to the interval of the Mth relevant parameter, W k [1] is the weighting factor of the synaptic parameter corresponding to the first relevant parameter, W norm (m) is a standardized constant value, and x(m) is the input value of m under different values;

[0040] Step 4, adjust the weighting factor W of the synaptic parameter. k (m) performs iterative calculations, and after each iteration, the new W is generated. k (m) Re-input the neural network model to Z out Update;

[0041] Step 5, when starting the engine, use Z... out The base fuel injection quantity is corrected, and the corrected value is used as the fuel injection quantity for engine startup.

[0042] Example 2

[0043] This embodiment uses three parameters—engine coolant temperature T, engine speed N, and engine load R—as examples to construct a three-dimensional space to illustrate steps 1 to 3 of this scheme.

[0044] The control system positions any engine operating condition within a three-dimensional space defined by engine coolant temperature T, engine speed N, and engine load R. This three-dimensional space is a three-dimensional map composed of engine coolant temperature T, engine speed N, and engine load R, where m takes values ​​from 1 to 3, corresponding to engine coolant temperature T, engine speed N, and engine load R, respectively.

[0045] The system calculates the engine's actual operating conditions (i.e., the probability of the engine operating within each neuron interval calculated in step 2) in the order of m=1, 2, 3. First, it calculates the probability of the engine coolant temperature interval (m=1), then the probability of the engine speed interval (m=2), and finally the probability of the engine load interval (m=3). The interval division takes into account engine operating conditions such as idling and full load, variable valve lift (VVT), and variable valve timing (VVL); engine speeds such as idling and acceleration; engine coolant temperature rise and adjustment; and the characteristics of the fuel injection system, i.e., the different injection modes for single and multiple injections under different operating conditions. These intervals may overlap. However, based on the three-dimensional space and interval division, the actual operating range of the engine can be defined.

[0046] The K self-learning intervals of the neuron are intervals between various typical operating conditions in three-dimensional space. The typical operating conditions are coordinate points (T, N, R) formed in three-dimensional space with the values ​​of engine coolant temperature T, engine speed N, and engine load R as coordinates. The engine coolant temperature T, engine speed N, and engine load R each include multiple values. The unit of engine coolant temperature T is °C, the unit of engine speed N is rmp, and the unit of engine load R is N·m.

[0047] In one embodiment, typical operating points include (90, 1200, 25), (90, 5000, 50), (90, 2500, 50), (90, 5250, 90), (90, 2000, 200), (90, 5500, 200), (50, 1200, 25), (60, 2250, 60), (50, 5000, 30), (60, 750, 100), (60, 5250, 90), (30, 1200, 25), (30, 2250, 60), (30, 5000, 30), (20, 750, 100), as detailed in Table 1.

[0048] Table 1

[0049] 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 T 90 90 90 90 90 90 50 60 50 60 60 30 20 30 20 N 1200 5000 2500 5250 2000 5500 1200 2250 5000 750 5250 1200 2250 5000 750 R 25 50 90 90 200 200 25 60 30 100 90 25 60 30 100

[0050] The 15 neurons are divided into different intervals according to the engine's operating conditions. Each neuron detects a different operating condition interval. The first neuron detects operating conditions where the engine water temperature is greater than 90°C, the engine speed is 1200 rpm, and the engine load is greater than 25 N·m.

[0051] The second neuron detects operating conditions where the engine coolant temperature is greater than 90°C, the engine speed is 5000 rpm, and the engine load is greater than 50 N·m.

[0052] The third neuron detects operating conditions where the engine water temperature is greater than 90℃, the engine speed is 2500 rpm, and the engine load is greater than 90 N·m.

[0053] The fourth neuron detects operating conditions where the engine water temperature is greater than 90℃, the engine speed is 5250 rpm, and the engine load is greater than 90 N·m.

[0054] The fifth neuron detects operating conditions where the engine water temperature is greater than 90℃, the engine speed is 2000 rpm, and the engine load is greater than 200 N·m.

[0055] The sixth neuron detects operating conditions where the engine water temperature is greater than 90℃, the engine speed is 5500 rpm, and the engine load is greater than 200 N·m.

[0056] The seventh neuron detects operating conditions where the engine water temperature is greater than 50°C, the engine speed is 1200 rpm, and the engine load is greater than 25 N·m.

[0057] The eighth neuron detects operating conditions where the engine water temperature is greater than 60℃, the engine speed is 2250 rpm, and the engine load is greater than 60 N·m.

[0058] The ninth neuron detects operating conditions where the engine water temperature is greater than 50℃, the engine speed is 5000 rpm, and the engine load is greater than 30 N·m.

[0059] The tenth neuron detects operating conditions where the engine water temperature is greater than 60℃, the engine speed is 750 rpm, and the engine load is greater than 100 N·m.

[0060] The eleventh neuron detects operating conditions where the engine water temperature is greater than 60℃, the engine speed is 5250 rpm, and the engine load is greater than 90 N·m.

[0061] The twelfth neuron detects operating conditions where the engine water temperature is greater than 30°C, the engine speed is 1200 rpm, and the engine load is greater than 25 N·m.

[0062] The thirteenth neuron detects operating conditions where the engine water temperature is greater than 20°C, the engine speed is 2250 rpm, and the engine load is greater than 60 N·m.

[0063] The fourteenth neuron detects operating conditions where the engine water temperature is greater than 30°C, the engine speed is 5000 rpm, and the engine load is greater than 30 N·m.

[0064] The fifteenth neuron detects operating conditions where the engine water temperature is greater than 20°C, the engine speed is 750 rpm, and the engine load is greater than 100 N·m.

[0065] Building the model first requires calculating the probability of the engine operating within each neuron interval, defined as Z. NORM [K]. In modeling, the input value x[m] can be assumed to follow a normal distribution across different dimensions. From the Gaussian function, if the random variable x[m] follows a probability distribution with location parameter u and scale parameter σ, then its probability density function is:

[0066]

[0067]

[0068] Define C k [m] represents the amplitude at the center of the normal distribution, which can be calibrated, σ nk [m] is the variance of the normal distribution, which can also be calibrated, then:

[0069]

[0070] Since the system cannot calculate the exponential function internally, the calculation of the exponential function can be directly converted into a special calibrable two-dimensional table (CUR), resulting in:

[0071]

[0072] Define the interval name as Z NORM [k], k = 1....15;

[0073] Simultaneously, the sum of probabilities of the engine operating within each neuron interval.

[0074] like Figure 2 As shown, the incremental value Z of the base injection quantity is calculated and output based on a pre-built neural network model. out ,in,

[0075] Weighting factor W for synaptic parameters k (m) performs iterative calculations, and after each iteration, the new W is generated. k (m) Re-input the neural network model to Z out Update;

[0076] in,

[0077] Fak is the adjustment factor in a traditional closed-loop PID controller for fuel cell combustion. This factor is an integral term based on the difference between the actual and target air-fuel mixture concentrations, fed back from the oxygen sensor voltage. The adjustment ratio of the integral term can be calibrated, for example, to 0.001. The integral time should be set as small as possible, or it can be set to adjust instantly. A larger derivative time setting is recommended, generally around 10% adjustment after a certain driving distance. The richness or leanness of the engine mixture will act on the LOLIMOT neural network through synaptic weighting factors, thus determining the learning speed and value of the synaptic parameter weighting factors. The larger the difference between the actual and target air-fuel mixture concentrations, the faster Fak updates, and the larger its value.

[0078] When starting the engine, use Z out The base fuel injection quantity is corrected, and the corrected value is used as the fuel injection quantity for engine startup. Increment value Z out The value is either positive or negative, and the Z value is used. out Correcting the base injection quantity means adding the positive or negative value to the base injection quantity. That is, when Z... outWhen the injection is positive, the amount of fuel injected is increased based on the base injection quantity; when the injection is negative, the amount of fuel injected is decreased based on the base injection quantity. This achieves control over the amount of fuel injected during startup. This control method breaks away from the linear interpolation method in traditional software, introduces a simple artificial intelligence algorithm, takes into account multi-dimensional sensor inputs, defines the engine's operating conditions more finely, and achieves more precise control.

[0079] In one embodiment, the air-fuel mixture concentration is not entered into closed-loop control when the engine starts, Z out The output value is the learned value calculated in the previous stage of engine operation. After the engine starts and enters the closed-loop control of the air-fuel mixture concentration, the system continues to calculate Z. out The updated value.

[0080] In one embodiment, self-learning requires certain conditions to be met:

[0081] 1. Whether self-learning is enabled can be calibrated. When the self-learning module is enabled, the system can enter self-learning mode.

[0082] 2. The air-fuel mixture concentration is in closed-loop regulation mode;

[0083] 3. The self-learning value is within the set maximum and minimum range;

[0084] 4. The vehicle is not in the carbon canister flushing state;

[0085] 5. The vehicle is in the catalytic converter heating or catalytic converter working state.

[0086] When the above conditions are met, the vehicle begins self-learning.

[0087] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process may be rearranged without departing from the scope of this disclosure. The appended method claims provide elements of various steps in an exemplary order and are not intended to limit the scope to the specific order or hierarchy described.

[0088] In the above detailed description, various features are combined together in a single embodiment to simplify this disclosure. This approach to disclosure should not be construed as reflecting an intention that embodiments of the claimed subject matter require more features than are explicitly stated in each claim. Rather, as reflected in the appended claims, the invention is presented with fewer features than all of the features of the single disclosed embodiment. Therefore, the appended claims are hereby explicitly incorporated into the detailed description, wherein each claim stands alone as a preferred embodiment of the invention.

[0089] The disclosed embodiments have been described above to enable any person skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be applied to other embodiments without departing from the spirit and scope of this disclosure. Therefore, this disclosure is not limited to the embodiments given herein, but is consistent with the broadest scope of the principles and novel features disclosed in this application.

[0090] The foregoing description includes examples of one or more embodiments. It is certainly impossible to describe all possible combinations of components or methods in order to describe the above embodiments, but those skilled in the art will recognize that further combinations and arrangements of the various embodiments are possible. Therefore, the embodiments described herein are intended to cover all such changes, modifications, and variations that fall within the scope of the appended claims. Furthermore, the term "comprising" as used in the specification or claims is interpreted in a manner similar to the term "including," as it is used as a conjunction in the claims. Additionally, the use of any term "or" in the specification of the claims is intended to mean "non-exclusive or."

[0091] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for controlling the amount of fuel injection during engine startup, characterized in that, include An M-dimensional space is defined by M relevant parameters of the engine's fuel injection quantity, and K self-learning intervals of neurons are established based on the M-dimensional space, with each self-learning interval corresponding to a neuron. Calculate the probability Z of the engine operating within each neuron interval. NORM [k], where k is any natural number from 1 to K; The incremental value Z of the base injection quantity is calculated and output based on a pre-built neural network model. out ,in, In the summation calculation, the value of m takes values ​​sequentially from the interval of the first relevant parameter to the interval of the Mth relevant parameter, W k [1] is the weighting factor of the synaptic parameter corresponding to the first relevant parameter, W norm (m) is a standardized constant value, and x(m) is the input value of m under different values; Weighting factor W for synaptic parameters k (m) performs iterative calculations, and after each iteration, the new W is generated. k (m) Re-input the neural network model to Z out Update; When starting the engine, use Z out The base fuel injection quantity is corrected, and the corrected value is used as the fuel injection quantity for engine startup.

2. The engine start-up fuel injection quantity control method according to claim 1, characterized in that: The relevant parameters include engine coolant temperature T, engine speed N, and engine load R, and the defined M-dimensional space is a three-dimensional space.

3. The engine start-up fuel injection quantity control method according to claim 2, characterized in that: The K self-learning intervals are intervals between various typical operating conditions in three-dimensional space. The typical operating conditions are coordinate points (T, N, R) formed in three-dimensional space with the values ​​of engine coolant temperature T, engine speed N, and engine load R as coordinates. The engine coolant temperature T, engine speed N, and engine load R each include multiple values. The unit of engine coolant temperature T is °C, the unit of engine speed N is rmp, and the unit of engine load R is N·m.

4. The engine start-up fuel injection quantity control method according to claim 3, characterized in that: The typical operating points include (90, 1200, 25), (90, 5000, 50), (90, 2500, 50), (90, 5250, 90), (90, 2000, 200), (90, 5500, 200), (50, 1200, 25), (60, 2250, 60), (50, 5000, 30), (60, 750, 100), (60, 5250, 90), (30, 1200, 25), (30, 2250, 60), (30, 5000, 30), (20, 750, 100).

5. The engine start-up fuel injection quantity control method according to claim 2, characterized in that, The probability Z of the engine operating within each neuron interval is... NORM [k] is calculated using the following formula: Where x(m) follows a normal distribution, C k [m] represents the amplitude at the center of the normal distribution, σ k [m] represents the variance of the normal distribution.

6. The engine start-up fuel injection quantity control method according to claim 1, characterized in that: The sum of probabilities that the engine operates within each neuron interval 7. The engine start-up fuel injection quantity control method according to claim 1, characterized in that, The synaptic parameter weighting factor W k (m) Iteratively calculate according to the following formula: Fak is the adjustment factor for a traditional closed-loop PID controller for fuel systems.

8. The engine start-up fuel injection quantity control method according to claim 1, characterized in that, The incremental value Z out The value is either positive or negative, and the Z value is used. out Correcting the base fuel injection quantity means adding the positive or negative value to the base fuel injection quantity.

Images