CN106908248A - The double regular empirical parameter automatic calibrating methods of weber burning of self-identifying list - Google Patents

Abstract

The purpose of the present invention is to provide an automatic calibration method for empirical parameters of self-identifying single and double Weber combustion rules. The present invention judges the number of Weber equations to be selected according to the test data of the burned fraction: when the number of Weber equations is 1, algebraic analysis is used to obtain the preliminary estimated value of the Weber parameter, and then the least square algorithm is used to obtain the final estimated value. When the number of Weber's equations is 2, the combustion phase separation point is obtained by using the combustion phase separation point determination method, and the burned fraction test data is divided into two parts according to the combustion phase separation point and processed accordingly. The preliminary estimated value of Weber parameter was obtained through analysis, and then the final estimated value was obtained by the least square algorithm. The invention can realize the self-identification of the number of Weber equations and automatically calibrate the parameters of the Weber equations, so that a zero-dimensional combustion model based on Weber's combustion rules can be quickly and accurately built, and the stability and optimality of the calibration results can be guaranteed.

Description

Automatic Calibration Method for Empirical Parameters of Self-identifying Single and Double Weber Combustion Rules

technical field

The invention relates to a method for calibrating combustion parameters of an internal combustion engine.

Background technique

In order to solve the increasingly serious environmental pollution problem, the international emission regulations are becoming more and more stringent, restricting the harmful emissions of the engine, so that the manufacturers regard the emission control of the engine as particularly important. The emission characteristics of diesel engines are closely related to the combustion process in the cylinder, so realizing the real-time control of the combustion process is of great significance to the emission control of the engine. With the rapid development of computer technology, computer simulation technology has a vigorous vitality. Through the abstract imitation of the real system, the system model is abstracted, and people conduct simulation experiments on such models on the computer, which saves scientific research and production. Costs are reduced, risks are reduced, and scientific research efficiency is improved. Then the reliability and accuracy of the system model directly determine the reliability and accuracy of the simulation results. In the field of internal combustion engines, the zero-dimensional combustion model based on the Weber (Wiebe) combustion rule is simple in form, less difficult to model, and has a certain simulation accuracy within a certain range of operating conditions. Based on this Weber (Wiebe) combustion model, many researchers have successfully predicted the in-cylinder pressure and temperature of direct injection, non-direct injection and two-stroke diesel engines. The empirical parameters of the Weber (Wiebe) combustion rule will directly affect the accuracy of the Weber (Wiebe) combustion model. Some literatures have studied how to calibrate the empirical parameters of the Weber (Wiebe) combustion model, such as algebraic analysis methods and least squares algorithms, but Both the algebraic analysis method and the least squares algorithm have advantages and disadvantages. The former has good stability and does not need to give an initial value, but it cannot guarantee the optimality of the calibration parameters; the latter can guarantee the local optimality of the calibration parameters, but the convergence and calibration results depend on the given initial value. Therefore, it is necessary to consider the combination of the two methods, so that the advantages and disadvantages of the two complement each other, and finally realize the rapid and accurate calibration of the empirical parameters of Weber's combustion rule. The research shows that the single-Wiebe combustion rule is only suitable for the combustion process simulation of one combustion phase or two combustion phases with slight mixing, but it cannot achieve a good simulation for the combustion process of the obvious mixture of two combustion phases. The double Weber (Wiebe) combustion rule can better simulate the combustion process of the two combustion phases mixed, but the calibration is difficult, so the selection of the number of single and double Wiebe equations needs to be a trade-off between calibration difficulty and accuracy compromise. For a given combustion process, how to automatically identify the number of Wiebe equations to be used and automatically calibrate to obtain the empirical parameters of the Wiebe equations is very critical.

Contents of the invention

The purpose of the present invention is to provide a self-identifying single and double Weber combustion rule empirical parameter automatic calibration method based on the burned fraction test data, which can quickly and accurately automatically calibrate parameters.

The purpose of the present invention is achieved like this:

The self-recognition method for automatic calibration of empirical parameters of single and double Weber combustion rules of the present invention is characterized in that:

(1) Carry out combustion test on diesel engine and collect test data sequence in is the crankshaft angle, x _b is and Corresponding burnt fraction, combustion fitting start angle Take it as the crankshaft angle corresponding to the burned fraction of 1%, and the combustion fitting end angle Take it as the crankshaft angle corresponding to the burned fraction of 99%;

(2) The test data sequence Linearization: Experimental data series based on fraction burned first calculated with preliminary estimate of with in is the corresponding crankshaft angle when the burnt fraction is zero. If the burnt fraction test data is always greater than zero, the crankshaft angle corresponding to the data starting point is taken as Linearizing the single Weber equation, let The implementation will test the data sequence The linearization of , the data sequence after linearization is

(3) Determine the number of Weber equations: defaults to right The data sequence is linearly fitted to obtain the fitting accuracy R ² , and set the single and double Weber recognition degree E. If R ² ≥ E, the number of Weber equations is recognized as 1; if R ² < E, the number of Weber equations is recognized as 2;

(4) For the determined number of Weber equations, the corresponding Weber parameter automatic calibration method is adopted to obtain the Weber equation parameter calibration result:

When the number of Weber's equations is identified as 1, according to the experimental value of the burned fraction, it is first obtained that with preliminary estimate of with in is the corresponding crankshaft angle when the burnt fraction is zero. If the burnt fraction test data is always greater than zero, the crankshaft angle corresponding to the data starting point is taken as Then for the data sequence Carry out linear fitting to obtain the fitting slope A, and m ₀ is calculated by m ₀ =A-1, where m ₀ is the initial value of the combustion index; then by Calculate the combustion efficiency factor a; As the equation to be fitted, m ₀ , with respectively as m, with The iterative initial value is obtained by fitting the non-linear least squares algorithm to obtain m, with the final estimate of

When the number of Weber's equations is identified as 2, according to the experimental value of the burnt fraction, first obtain with preliminary estimate of with in is the corresponding crankshaft angle when the burnt fraction is zero. If the burnt fraction test data is always greater than zero, the crankshaft angle corresponding to the data starting point is taken as Then for the linearized The data confirm the combustion phase separation point p, that is, to find a point p such that the points before and after this point with The comprehensive R ² accuracy of straight line fitting to the data reaches the maximum; according to the combustion phase separation point p, the burnt fraction test data sequence into two parts, namely with in x1 _b = [x _b (1), x _b (2), ..., x _b (p)], x2 _b = [x _b (p+1), x _b (p+2),..., x _b (n)]; α ₀ ＝x _b (p) is used as the initial value of the premixed combustion ratio, and x1 _b and x2 _b are normalized: make respectively realize the with linearization, yes with The two parts of the data series are linearly fitted respectively, and the fitting slopes A ₁ and A ₂ are obtained respectively, and m1 ₀ and m2 ₀ are respectively obtained from m1 ₀ =A ₁ -1, m2 ₀ =A ₂ -1, where m1 ₀ and m2 ₀ are the initial values of combustion index of premixed combustion and combustion index of diffusion combustion respectively; with α ₀ , m1 ₀ , m2 ₀ , with as α, m1, m2, with The iterative initial value of α, m1, m2, with the final estimate of

(5) Output the number of Weber equations and the corresponding Weber equation parameter sets, so as to complete the self-identification of single and double Weber combustion rules and automatically calibrate to obtain the empirical parameters of Weber combustion rules.

The present invention may also include:

1. The determination method of combustion phase separation point p is as follows:

Suppose the data separation point i will be the data sequence Divided into two parts, the part before the i-th data is The part after the i-th data is right with Carry out linear fitting respectively, the linear fitting precision of the two parts of data is R ² ₁ and R ² ₂ respectively, and the comprehensive precision R ² is: R ² (i)=[R ² ₁ ×i+R ² ₂ ×(ni) ]/n, where n is the total number of data, the data separation point i can be changed from 1 to n, and the comprehensive precision R ² is calculated in turn, and then the data separation point i when the comprehensive precision R ² reaches the maximum value is taken As the combustion phase separation point p.

The advantage of the present invention is: the present invention is according to Weber (Wiebe) combustion rule, relying on the test data of burned fraction, adopts original self-recognition odd and double Weber (Wiebe) algorithm, automatically recognizes the number of Weber (Wiebe) equations, and according to The number of identified Weber (Wiebe) equations adopts the corresponding automatic calibration algorithm of Wiebe parameters, and finally realizes the method of fast and accurate automatic calibration to obtain the empirical parameters of the Wiebe combustion rule. Self-recognition single and double Wiebe combustion rule empirical parameter automatic calibration method adopts original single and double Wiebe self-recognition algorithm to realize automatic recognition of the number of Wiebe equations, and combines algebraic analysis method and least square algorithm to make The advantages and disadvantages of the two complement each other to realize the automatic calibration of the empirical parameters of the Weber (Wiebe) combustion rule. This method automatically recognizes the number of Weber (Wiebe) equations. The convergence and stability of the parameter calibration are better, and the accuracy is higher. It is a research in the industry. It provides great convenience for personnel to select single and double Wiebe combustion rules and to calibrate empirical parameters of Wiebe combustion rules.

Description of drawings

Fig. 1 is a flowchart of the present invention.

detailed description

The present invention is described in more detail below in conjunction with accompanying drawing example:

Combined with Figure 1, the single and double Wiebe combustion rules commonly used in the zero-dimensional combustion modeling of internal combustion engines are shown in the parameter equations shown in equations (1) and (2) respectively.

In formula (1): x _b is the percentage of burned fuel; m is the combustion quality index; a is the combustion efficiency factor; — Instantaneous crankshaft angle; — combustion duration angle; - The starting point of combustion. where m, a, with and are empirical parameters to be calibrated. In formula (2): x _b is the percentage of burned fuel; m ₁ and m ₂ are the combustion quality indices of the first Wiebe and second Wiebe equations; a ₁ and a ₂ are the first Wiebe (Wiebe) and the combustion efficiency factor of the second Weber (Wiebe) equation; — Instantaneous crankshaft angle; with are the combustion duration angles of the first and second Wiebe equations, respectively; with are the starting points of combustion for the first and second Wiebe equations, respectively. where α, m ₁ , m ₂ , with is the empirical parameter to be calibrated.

Firstly, the ignition start point test value is used as the estimated value of the combustion start point by as burning duration The estimated value of the burned fraction test data is linearized according to the Weber (Wiebe) equation; then the processed test data is used for linear fitting, and the R2 accuracy of the linear fitting is used as a measure of the number of Weber ⁽ Wiebe) equations The standard of , taking the artificially set value E as the recognition degree of the number of Weber (Wiebe) equations, when R ² ≥ E, the number of Weber (Wiebe) equations is selected as 1, when R ² <E, the number of Weber (Wiebe) equations The number is selected as 2; according to the determined number of Weber equations, the corresponding automatic calibration method of Weber (Wiebe) parameters is adopted, and finally the calibrated empirical parameters of Weber (Wiebe) equations are obtained. When the number of Weber equations is 1, the algebraic analysis method is first used to obtain the preliminary estimated value of the parameters of the Weber (Wiebe) equation, and then it is used as the initial value of the iteration of the Weber (Wiebe) parameter, and the nonlinear least square algorithm is used to calibrate the Weber ( The final estimate of the Wiebe) parameters. When the number of Weber's equations is 2, first adopt the combustion phase separation point determination method proposed by the present invention to obtain the combustion phase separation point, and then divide the test data into two parts according to the combustion phase separation point, and carry out corresponding processing on the two parts of data respectively , and then use the algebraic analysis method to obtain the preliminary estimated value of the Wiebe equation for the two parts of the processed data, and finally use the nonlinear least squares algorithm to calibrate to obtain the final estimated value of the empirical parameters of the double Wiebe equation.

The automatic calibration method of empirical parameters of self-identifying single and double Weber (Wiebe) combustion rules, the calculation process is as follows:

Step 1: Import the measured burnt fraction test data sequence in is the crankshaft angle, x _b is and Corresponding burnt fraction, combustion fitting start angle Taken as the crankshaft angle corresponding to the burned fraction slightly greater than 0 (preferably 1%), the combustion fitting end angle Take the crank angle corresponding to the burnt fraction slightly less than 1 (preferably 99%), and extract between test data.

Step 2: Put the test data sequence linearization. According to burnt fraction test data series first calculated with preliminary estimate of with in is the corresponding crankshaft angle when the burnt fraction is zero (in order to improve the applicability and stability of the method, if the burnt fraction test data is always greater than zero, the crankshaft angle corresponding to the data starting point is taken as ). Linearizing the single Weber equation, let The implementation will test the data sequence The linearization of , the data sequence after linearization is

Step 3: Determine the number of Wiebe equations. defaults to right Perform linear fitting on the data sequence to obtain the fitting accuracy R ² , and set the single and double Wiebe recognition degree E (E is usually selected as a number between 0.99 and 1, preferably 0.995), if R ² ≥ E, The number of Wiebe equations is identified as 1; if R ² <E, the number of Wiebe equations is identified as 2.

Step 4: According to the number of Weber equations determined in step 3, the corresponding Weber parameter automatic calibration method is adopted to obtain the Weber equation parameter calibration result. When the number of Wiebe equations is identified as 1, according to the experimental value of the burnt fraction, firstly get with preliminary estimate of with in is the corresponding crankshaft angle when the burnt fraction is zero (in order to improve the applicability and stability of the method, if the burnt fraction test data is always greater than zero, the crankshaft angle corresponding to the data starting point is taken as ); then the data sequence Carry out linear fitting to obtain the fitting slope A, and m ₀ is calculated by m ₀ =A-1, where m ₀ is the initial value of the combustion index; then by Calculate the combustion efficiency factor a, which is preferably a fixed value of 4.605, corresponding to the combustion duration of 0% to 99% of the burned fraction; with the formula (1) in claim 1 as the equation to be fitted, with m ₀ , with respectively as m, with The iterative initial value is obtained by fitting the non-linear least squares algorithm to obtain m, with final estimate of . When the number of Wiebe equations is identified as 2, according to the experimental value of the burnt fraction, firstly get with preliminary estimate of with in is the corresponding crankshaft angle when the burnt fraction is zero (in order to improve the applicability and stability of the method, if the burnt fraction test data is always greater than zero, the crankshaft angle corresponding to the data starting point is taken as ); then the linearized The data confirm the combustion phase separation point p, that is, to find a point p such that the points before and after this point with The comprehensive R ² accuracy of straight line fitting to the data reaches the maximum; according to the combustion phase separation point p, the burnt fraction test data sequence into two parts, namely with in x1 _b = [x _b (1), x _b (2), ..., x _b (p)], x2 _b =[x _b (p+1), x _b (p+2), . . . , x _b (n)]. α ₀ ＝x _b (p) is used as the initial value of the premixed combustion ratio, and x1 _b and x2 _b are normalized: make respectively realize the with linearization, yes with The two parts of the data series are linearly fitted respectively, and the fitting slopes A ₁ and A ₂ are obtained respectively, and m1 ₀ and m2 ₀ are respectively obtained from m1 ₀ =A ₁ -1, m2 ₀ =A ₂ -1, where m1 ₀ and m2 ₀ are the initial value of the combustion index of premixed combustion and the initial value of the combustion index of diffusion combustion; both a1 and a2 are preferably a fixed value of 4.605 (not limited to 4.605), corresponding to the combustion duration of 0% to 99% of the burned fraction, As a premixed combustion and diffusion combustion efficiency factor. With α ₀ , m1 ₀ , m2 ₀ , with as α, m1, m2, with The iterative initial value of α, m1, m2, with final estimate of .

Step 5: Output the number of Weber equations and the corresponding parameter set of Weber equations.

So far, according to the measured test data, the empirical parameters of the Wiebe combustion rules can be obtained by self-identifying the single and double Wiebe combustion rules and automatically calibrating.

The principle of the determination method of the combustion phase separation point p described in step four is as follows:

Combustion phase separation point p. Suppose the data separation point i will be the data sequence Divided into two parts, the part before the i-th data is The part after the i-th data is right with Carry out linear fitting respectively, the linear fitting precision of the two parts of data is R ² ₁ and R ² ₂ respectively, the comprehensive precision R ² is defined as shown in the following formula, where n is the total number of data, the data separation point i can be determined by 1~n are changed, and the comprehensive precision R ² is calculated respectively in turn, and then the data separation point i when the comprehensive precision R ² reaches the maximum value is taken as the combustion phase separation point p.

The specific principle of the automatic calibration method of empirical parameters of self-identifying single and double Weber (Wiebe) combustion rules is as follows:

The single Weber equation is shown in formula (1).

for Determined by the corresponding crankshaft angle when x _b is 0; for It is determined by the crank angle period corresponding to x _b being 0-0.99.

Usually 50% of the heat release corresponds to the crankshaft angle As the combustion center, it corresponds to formula (1), that is:

Organized to get:

Transform formula (5) into:

It can be known from formula (6) that m can be calculated by calculating the slope of G and H.

Assuming that the starting point of combustion fitting is x _bs , the corresponding crankshaft angle is The combustion fitting end point is x _bc , and the corresponding crankshaft angle is From formula (1), it can be concluded that:

When the combustion duration is selected as the crank angle period corresponding to the burned fraction of 0-99%,

It can be seen from formula (7) that for different combustion durations and m, a will be different. The present invention a is preferably 4.605.

The single Weber combustion rule of formula (1) is transformed into:

It can be seen from formula (8) that the size of m can represent the speed of burning speed.

For the combustion process with single combustion phase or slight dual combustion phase blending, the linear relationship between K and G is better because the burning speed is not much different in the whole combustion process, and the ^R2 accuracy of linear fitting is used for K and G For the combustion process with obvious ^double combustion phase blending, the linear relationship between K and G is poor due to the large difference in combustion velocity throughout the combustion process, and the R2 accuracy of linear fitting for K and G is relatively low Low. From the above analysis, it can be seen that the R ² precision level of linear fitting of K and G can characterize the severity of double combustion phase blending, E is the identification degree of the number of Weber equations, and R ² ≥ E is selected as the identity of a single Weber equation, and R ² <E is used as the double Weber equation identification to realize the single and double Weber self-identification function. In the present invention, E is preferably 0.995. The mathematical expression of the single and double Weber equation self-identification algorithm is as follows:

In the formula, Num is the number of Weber equations.

For direct injection diesel engines, in addition to the diffusion combustion mode, there is always a premixed combustion mode more or less in the combustion process. In order to facilitate the parameter calibration of the double Weber equation, it is very important to find the separation point of the two combustion modes of premixed combustion and diffusion combustion . This software takes an original approach to determine the combustion mode separation point, the validation algorithm is described below.

It can be seen from formula (8) that the size of m can reflect the speed of the combustion process. For the premixed combustion mode, the combustion velocity is faster, so m is smaller; for the diffusion combustion mode, the combustion velocity is slower, so m is larger. ^The separation point of the combustion mode is to find a point so that the comprehensive R2 accuracy of the G and K data before and after this point is respectively fitted with a straight line to maximize. Assuming that the R ² precision of linear fitting for G and K before this point (the pth data point) is R ² ₁ , and the R ² precision of linear fitting for G and K after this point is R ² ₂ , define The combined R ² (p) accuracy is:

Then, when R ² (p) is taken as the maximum value, the corresponding p is the combustion mode separation point to be found.

Divide the burnt fraction test data into two parts according to the data separation point p, x1 _b = x _b (1:p), x2 _b = x _b (p+1:end), α ₀ ＝x _b (p), and process x1 _b and x2 _b : Algebraic analysis is carried out on the processed two sets of data respectively, and m, with The estimated value of is used as the iterative initial value of the least squares algorithm.

The least squares algorithm is commonly used in nonlinear equation fitting problems, and its theory is described below.

Given n pairs of independent variable and dependent variable test data ( _xi , y _i ), the parameter set to be determined is β, and the fitting equation selected is p(x, β), therefore, the sum of squares of errors is:

The least squares algorithm is to obtain a set of β such that S(β) is the smallest.

The preferred least squares algorithm of the present invention is the Levenberg-Marquardt algorithm. The initial value of the parameter set β to be calibrated needs to be given at the beginning of the algorithm calculation. Afterwards, at each iteration step β is replaced by the newly estimated value β+δ. To determine δ, the equation p( _xi ,β+δ) is estimated linearly:

p( _xi ,β+δ)=p( _xi ,β)+J _i δ (12)

In the formula is the gradient relative to β. When S(β) reaches the minimum, the gradient of S(β) to δ becomes 0. According to formula (12), the first-order estimate of formula (11) is as follows:

Expressed in vector form as:

S(β+δ)≈||yf(β)-Jδ|| ² (14)

Equation (14) is derived with respect to J, and the derivative function is set to zero, we can get:

(J ^T J)δ=J ^T [yf(β)] (15)

The Levenberg-Marquardt algorithm improves formula (15) and becomes the following formula:

(J ^T J+λI)δ=J ^T [yf(β)] (16)

In the formula, I is the identity matrix, and λ is the damping coefficient, which is used to adjust the step size of each iteration. When λ is zero, formula (16) degenerates into formula (15), which is a Gauss-Newton algorithm; when λ is a larger value, formula (16) degenerates into a gradient descent algorithm.

In order to improve the convergence speed of formula (16), replace I with J ^T J for formula (16), and the final Levenberg-Marquardt algorithm is shown in the following formula:

(J ^T J+λdiag(J ^T J))δ=J ^T [yf(β)] (17)

Since the least squares algorithm needs to give the initial value of the parameters to be calibrated at the beginning of the calculation, the convergence and iterative calculation time of the least squares algorithm are strongly dependent on the initial value, and the final result cannot guarantee the global optimality. Only local optimality can be guaranteed, so it is very important to give a reasonable initial value. The present invention adopts the Weber empirical parameters calculated by the algebraic analysis method according to the test data as the initial value of the least squares algorithm, and then performs further iterative calculations, so that the optimality of the calibration results can be guaranteed, and the minimum value can be greatly improved. The convergence of the square method can reduce the iterative calculation time.

Claims (2)

1. An automatic calibration method for empirical parameters of self-identifying single and double Weber combustion rules, which is characterized by: (1) Carry out combustion test on diesel engine and collect test data sequence in is the crankshaft angle, x _b is and Corresponding burnt fraction, combustion fitting start angle Take it as the crankshaft angle corresponding to the burned fraction of 1%, and the combustion fitting end angle Take it as the crankshaft angle corresponding to the burned fraction of 99%; (2) The test data sequence Linearization: Experimental data series based on fraction burned first calculated with preliminary estimate of with in is the corresponding crankshaft angle when the burnt fraction is zero. If the burnt fraction test data is always greater than zero, the crankshaft angle corresponding to the data starting point is taken as Linearizing the single Weber equation, let The implementation will test the data sequence The linearization of , the data sequence after linearization is (3) Determine the number of Weber equations: defaults to right The data sequence is linearly fitted to obtain the fitting accuracy R ² , and set the single and double Weber recognition degree E. If R ² ≥ E, the number of Weber equations is recognized as 1; if R ² < E, the number of Weber equations is recognized as 2; (4) For the determined number of Weber equations, the corresponding Weber parameter automatic calibration method is adopted to obtain the Weber equation parameter calibration result: When the number of Weber's equations is identified as 1, according to the experimental value of the burned fraction, it is first obtained that with preliminary estimate of with in is the corresponding crankshaft angle when the burnt fraction is zero. If the burnt fraction test data is always greater than zero, the crankshaft angle corresponding to the data starting point is taken as Then for the data sequence Carry out linear fitting to obtain the fitting slope A, and m ₀ is calculated by m ₀ =A-1, where m ₀ is the initial value of the combustion index; then by Calculate the combustion efficiency factor a; As the equation to be fitted, m ₀ , with respectively as m, with The iterative initial value is obtained by fitting the non-linear least squares algorithm to obtain m, with the final estimate of When the number of Weber's equations is identified as 2, according to the experimental value of the burnt fraction, first obtain with preliminary estimate of with in is the corresponding crankshaft angle when the burnt fraction is zero. If the burnt fraction test data is always greater than zero, the crankshaft angle corresponding to the data starting point is taken as Then for the linearized The data confirm the combustion phase separation point p, that is, to find a point p such that the points before and after this point with The comprehensive R ² accuracy of straight line fitting to the data reaches the maximum; according to the combustion phase separation point p, the burnt fraction test data sequence into two parts, namely with in x1 _b = [x _b (1), x _b (2), ..., x _b (p)], x2 _b = [x _b (p+1), x _b (p+2),..., x _b (n)]; α ₀ ＝x _b (p) is used as the initial value of the premixed combustion ratio, and x1 _b and x2 _b are normalized: make respectively realize the with linearization, yes with The two parts of the data series are linearly fitted respectively, and the fitting slopes A ₁ and A ₂ are obtained respectively, and m1 ₀ and m2 ₀ are respectively obtained from m1 ₀ =A ₁ -1, m2 ₀ =A ₂ -1, where m1 ₀ and m2 ₀ are the initial values of combustion index of premixed combustion and combustion index of diffusion combustion respectively; with α ₀ , m1 ₀ , m2 ₀ , with as α, m1, m ₂ , with The iterative initial value of α, m1, m2, with the final estimate of (5) Output the number of Weber equations and the corresponding Weber equation parameter sets, so as to complete the self-identification of single and double Weber combustion rules and automatically calibrate to obtain the empirical parameters of Weber combustion rules. 2. The self-identifying single and double Weber combustion rule empirical parameter automatic calibration method according to claim 1 is characterized in that: The determination method of combustion phase separation point p is as follows: Suppose the data separation point i will be the data sequence Divided into two parts, the part before the i-th data is The part after the i-th data is right with Carry out linear fitting respectively, the linear fitting precision of the two parts of data is R ² ₁ and R ² ₂ respectively, and the comprehensive precision R ² is: R ² (i)=[R ² ₁ ×i+R ² ₂ ×(ni) ]/n, where n is the total number of data, the data separation point i can be changed from 1 to n, and the comprehensive precision R ² is calculated in turn, and then the data separation point i when the comprehensive precision R ² reaches the maximum value is taken As the combustion phase separation point p.