State estimation method of pressure sensor based on belief rule base algorithm

Abstract

The invention provides a pressure sensor state estimation method based on a belief rule base algorithm, which comprises the following steps: under different temperature conditions, applying a pressure value to a pressure sensor, and recording current data and environment temperature data; converting the current data and the temperature data into characteristic values; constructing a corresponding evidence for each feature value; performing reliability evaluation on each evidence; based on a reliability evaluation result, distributing a reliability factor for each evidence; fusing a plurality of evidences into comprehensive evidences by using a fusion rule, and establishing a confidence rule base to perform preliminary evaluation test on the working pressure; setting an optimization objective function, and establishing an evidence reasoning optimization model; and applying the optimized belief rule base optimization model to an actual scene, and carrying out real-time calibration on the output of the pressure sensor. The method has remarkable advantages in the aspects of improving the calibration precision, enhancing the adaptability, realizing real-time calibration and the like, and has important significance in improving the performance of the pressure sensor and widening the application range of the pressure sensor.

Description

Technical Field

[0001] The invention relates to the technical field of marine diesel engines, and in particular to a state estimation method of a pressure sensor based on a confidence rule base algorithm. Background Art

[0002] Marine diesel engines are the main power units of ships. They provide powerful power output and can efficiently convert diesel fuel into power to meet the long-term and high-load operation requirements of ships. At the same time, they can work in harsh marine environments and maintain stable performance.

[0003] Diesel engines are an important power equipment, and pressure sensors play a key role in them. By monitoring and controlling pressure, they optimize combustion, improve combustion efficiency and performance, and help with fault diagnosis and repair. The temperature compensation capability of the pressure sensor suitable for the NSC2800 chip is limited. Although it has a certain temperature compensation function, its compensation range and effect are still subject to certain restrictions. In the case of large or rapid temperature changes, temperature compensation may not be performed in a timely and accurate manner, resulting in measurement errors. Summary of the invention

[0004] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a state estimation method for a pressure sensor based on a confidence rule base algorithm. The method collects the current of the measurement circuit and the temperature during measurement, extracts the characteristic values ​​of the current and temperature, extracts evidence from the current and temperature based on given initial input and output reference values, calculates the reliability of the evidence, establishes an objective function to optimize the model training inference parameters, and finally estimates the diesel engine working pressure by fusing the evidence, thereby achieving high-precision pressure detection of marine diesel engines.

[0005] The present invention achieves the above-mentioned purpose through the following technical solutions:

[0006] A state estimation method for a pressure sensor based on a confidence rule base algorithm, the method comprising the following steps:

[0007] Step S1, applying a series of pressure values ​​within a preset range to the pressure sensor under different temperature conditions, and recording corresponding current data and ambient temperature data;

[0008] Step S2, converting current data and temperature data into characteristic values;

[0009] Step S3, constructing corresponding evidence for each eigenvalue based on the extracted eigenvalues, including current eigenvalue evidence and temperature eigenvalue evidence;

[0010] Step S4, performing reliability assessment on each piece of evidence, i.e. determining its credibility in estimating the actual pressure; assigning a reliability factor to each piece of evidence based on the reliability assessment result;

[0011] Step S5, using fusion rules to fuse multiple pieces of evidence into comprehensive evidence, and establishing a confidence rule base to conduct a preliminary assessment test on work pressure;

[0012] Step S6, setting the optimization objective function and establishing the evidence reasoning optimization model;

[0013] Step S7, applying the optimized confidence rule base optimization model to the actual scene to calibrate the output of the pressure sensor in real time.

[0014] According to a state estimation method for a pressure sensor based on a confidence rule base algorithm provided by the present invention, in step S1, different temperatures C are collected respectively. i Under the condition of 0-4 bar, different current values ​​i of the pressure sensor are set, and K current reference levels and M temperature reference levels are set.

[0015] According to a state estimation method of a pressure sensor based on a confidence rule base algorithm provided by the present invention, in step S2, the collected current data and temperature data are converted into characteristic values ​​f1(t) and f2(t) by similarity normalization processing, the output pressure is recorded as P(t), and the reference value set of the pressure output result of the pressure sensor is set to D={D n |n=1,…,N}, represent f1(t), f2(t) and P(t) as the sample set S={[f1(t),f2(t),P(t)]|t=1,2,…T s};

[0016] Wherein, N is the number of result reference values ​​P(t) of the output pressure, and S=[f1(t), f2(t), P(t)] is a sample size.

[0017] According to a state estimation method of a pressure sensor based on a confidence rule base algorithm provided by the present invention, in step S3, the data [f1(t), f2(t), P(t)] in the sample set S is converted into a form of confidence distribution using a given reference value;

[0018] Calculate the confidence distribution of all data sets in S (α i,j γ n ,α i,j+1 γ n ,α i,j γ n+1 ,α i,j+1 γ n+1 ), construct (f i(t), P(t)) reference value confidence distribution statistics table;

[0019] Among them, α i,j γ n Represents a sample pair (f i (t),P(t)) The input value corresponds to the reference value P(t) corresponds to the reference value D n The comprehensive similarity of

[0020] According to a state estimation method of a pressure sensor based on a confidence rule base algorithm provided by the present invention, (f i (t), P(t)) is normalized by likelihood to obtain the reference value Evidence, construct input f i (t) Evidence table.

[0021] According to a state estimation method of a pressure sensor based on a confidence rule base algorithm provided by the present invention, in step S4, the confidence distribution matrix is ​​transformed to obtain a pressure prediction model;

[0022] According to the input value f i The relative change value of (t) and the result value P(t) is used to calculate the evaluation factor reflecting the input characteristic signal and the output pressure value, thereby calculating the input information source f i (t) reliability factor;

[0023] According to a state estimation method for a pressure sensor based on a confidence rule base algorithm provided by the present invention, in step S5, the obtained input f i Based on the fusion results, a confidence rule base is established to conduct a preliminary evaluation test on the diesel engine working pressure C(t).

[0024] According to a state estimation method of a pressure sensor based on a confidence rule base algorithm provided by the present invention, in step S6, the mean square error between the diesel engine working pressure estimated by the confidence rule base and the actual pressure is set as the objective function to determine the optimized parameter set Clarify the conditions that need to be met for the optimization parameters and build an evidential reasoning optimization model based on mean square error;

[0025] Among them, w i represents the weight of evidence, and the other parameters are set as D1=l1, D N =l2, are the minimum and maximum values ​​of the input eigenvalues ​​f1(t) and f2(t), respectively, and l1 and l2 are the minimum and maximum values ​​of P(t), respectively.

[0026] According to a state estimation method of a pressure sensor based on a confidence rule base algorithm provided by the present invention, the constraints of the confidence rule base optimization model include: (a) the rule weight value interval is [0,1]; (b) the premise attribute weight value interval of each rule is [0,1]; (c) the confidence level shall not be greater than 1 or less than 0; (d) if the kth rule is complete, then the sum of all confidence levels in the output result is equal to 1; otherwise it is less than 1.

[0027] According to a state estimation method of a pressure sensor based on a confidence rule base algorithm provided by the present invention, the working pressure of a diesel engine is estimated according to the fusion result, including the following formula:

[0028]

[0029] It can be seen that compared with the prior art, the method provided by the present invention has the following beneficial effects:

[0030] 1. The present invention collects data under different temperature conditions. The method can take into account the influence of temperature on the performance of the pressure sensor, thereby enhancing the adaptability and accuracy of the calibration model, so that the calibrated sensor can maintain high accuracy under a wider range of environmental conditions.

[0031] 2. The present invention converts current and temperature data into characteristic values, constructs evidence based on these characteristic values, and then performs reliability assessment on the evidence. This method can accurately describe the relationship between the sensor output and the actual pressure. The introduction of the reliability factor further improves the accuracy of the estimation result and reduces the error.

[0032] 3. The present invention utilizes fusion rules to fuse multiple pieces of evidence (including evidence of current and temperature) into comprehensive evidence. This process fully considers the contribution of information from different sources to pressure estimation, effectively avoids the deviation that may be caused by a single information source, and improves the comprehensiveness and reliability of calibration.

[0033] 4. By setting the optimization objective function and establishing the evidence reasoning optimization model, the present invention can automatically adjust and optimize the calibration parameters to adapt to the needs of different sensors or different working environments, making the calibration process more flexible and efficient.

[0034] 5. The present invention applies the optimized confidence rule base optimization model to actual scenarios, which can realize real-time calibration of the pressure sensor output. This is particularly important for application scenarios that require continuous and accurate monitoring of pressure changes, such as industrial automation, aerospace and other fields.

[0035] 6. The structure of the trust rule base of the present invention is clear, easy to understand and operate. With the addition of new data or new types of sensors, the rule base can be easily updated or expanded to adapt to new application scenarios or improve calibration accuracy. At the same time, the transparency of the rules also makes system maintenance and debugging more convenient.

[0036] 7. Since the present invention takes into account the influence of various conditions and factors and adopts strategies such as multi-source information fusion and reliability evaluation, the method shows strong robustness in the face of complex and changeable working environments and can maintain high calibration accuracy and stability.

[0037] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a first flow chart of an embodiment of a method for state estimation of a pressure sensor based on a confidence rule base algorithm of the present invention.

[0039] Figure 2 It is a second flow chart of a basic neural network and a physical information neural network in an embodiment of a state estimation method for a pressure sensor based on a confidence rule base algorithm of the present invention. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0042] See also Figure 1 and Figure 2 This embodiment provides a state estimation method for a pressure sensor based on a confidence rule base algorithm, the method comprising the following steps:

[0043] Step S1, applying a series of pressure values ​​within a preset range to the pressure sensor under different temperature conditions, and recording corresponding current data and ambient temperature data;

[0044] Step S2, converting current data and temperature data into characteristic values;

[0045] Step S3, constructing corresponding evidence for each eigenvalue based on the extracted eigenvalues, including current eigenvalue evidence and temperature eigenvalue evidence;

[0046] Step S4, performing reliability assessment on each piece of evidence, i.e. determining its credibility in estimating the actual pressure; assigning a reliability factor to each piece of evidence based on the reliability assessment result;

[0047] Step S5, using fusion rules to fuse multiple pieces of evidence into comprehensive evidence, and establishing a confidence rule base to conduct a preliminary assessment test on work pressure;

[0048] Step S6, setting the optimization objective function and establishing the evidence reasoning optimization model;

[0049] Step S7, applying the optimized confidence rule base optimization model to the actual scene to calibrate the output of the pressure sensor in real time.

[0050] In step S1, different temperatures C are collected respectively. i Under the present invention, different current values ​​i of the pressure sensor at 0-4 bar are set, and K current reference levels and M temperature reference levels are set, for example, 5 current reference levels and 3 temperature reference levels.

[0051] In step S2, the collected current data and temperature data are converted into characteristic values ​​f1(t) and f2(t) through similarity normalization processing, the output pressure is recorded as P(t), and the reference value set of the pressure output result of the pressure sensor is set to D = {D n |n=1,…,N}, represent f1(t), f2(t) and P(t) as the sample set S={[f1(t),f2(t),P(t)]|t=1,2,…T s};

[0052] Wherein, N is the number of result reference values ​​P(t) of the output pressure, and S=[f1(t), f2(t), P(t)] is a sample size.

[0053] In step S3, the data [f1(t), f2(t), P(t)] in the sample set S is converted into the form of reliability distribution using a given reference value;

[0054] Calculate the confidence distribution of all data sets in S (α i,j γ n ,α i,j+1 γ n ,α i,j γ n+1 ,αi,j+1 γ n+1 ), construct (f i (t), P(t)) reference value confidence distribution statistics table;

[0055] Among them, α i,j γ n Represents a sample pair (f i (t),P(t)) The input value corresponds to the reference value P(t) corresponds to the reference value D n The comprehensive similarity of

[0056] Yes (f i (t), P(t)) is normalized by likelihood to obtain the reference value Evidence, construct input f i (t) Evidence table.

[0057] In step S4, the confidence distribution matrix is ​​transformed to obtain a pressure prediction model;

[0058] According to the input value f i The relative change value of (t) and the result value P(t) is used to calculate the evaluation factor reflecting the input characteristic signal and the output pressure value, thereby calculating the input information source f i (t) reliability factor;

[0059] In step S5, using the obtained input f i Based on the fusion results, a confidence rule base is established to conduct a preliminary evaluation test on the diesel engine working pressure C(t).

[0060] In step S6, the mean square error between the diesel engine working pressure estimated by the confidence rule base and the actual pressure is set as the objective function to determine the optimized parameter set. Clarify the conditions that need to be met for the optimization parameters and build an evidential reasoning optimization model based on mean square error;

[0061] Among them, w i represents the weight of evidence, and the other parameters are set as D1=l1, D N =l2, are the minimum and maximum values ​​of the input eigenvalues ​​f1(t) and f2(t), respectively, and l1 and l2 are the minimum and maximum values ​​of P(t), respectively.

[0062] Repeat steps S3 to S5, and use the obtained confidence rule base to optimize the model to predict the pressure.

[0063] In this embodiment, the constraints of the confidence rule base optimization model include: (a) the rule weight value interval is [0, 1]; (b) the premise attribute weight value interval of each rule is [0, 1]; (c) the confidence level shall not be greater than 1 or less than 0; (d) if the kth rule is complete, then the sum of all confidence levels in the output result is equal to 1; otherwise, it is less than 1.

[0064] In this embodiment, the diesel engine working pressure is estimated according to the fusion result, including the following formula:

[0065]

[0066] In practical applications, the experiment in this embodiment uses a ceramic capacitor pressure sensor of NSC2860, the circuit allows a current of 4.000-20.000mA, the ammeter used to measure the current value of the sensor is a DM3058E ammeter produced by RIGOL, and the instrument used for pressure calibration is an intelligent digital pressure calibrator.

[0067] The embodiment of the present invention relates to a pressure calibration method of an NSC2860 chip ceramic capacitor pressure sensor based on a confidence rule base algorithm, the method comprising:

[0068] The current in the circuit when the sensor is connected at 0-4 bar at different temperatures is collected, and the collected data is divided into training set and test set according to x:y.

[0069] Among them, the three reference levels of setting temperature are C1=30°, C2=55°, C3=80°, and the five reference levels of current are i1=4mA, i2=8mA, i3=12mA, i4=16mA, i5=20mA.

[0070] The collected data is normalized to obtain the characteristic signals f1(t) and f2(t) of current and temperature. The output pressure is recorded as P(t), P(t)∈[l1,l2], and f1(t), f2(t) and P(t) are represented as a sample set S={[f1(t),f2(t),P(t)]|t=1,2,…T s}; Set the reference value set of the sensor output pressure result to D = {D n |n=1,…,N}, the reference value set of the input signal is

[0071] in, are the minimum and maximum values ​​of the input characteristic values ​​f1(t) and f2(t), l1 and l2 are the minimum and maximum values ​​of P(t), N is the number of reference values ​​P(t) of the output pressure, j is the number of reference values ​​P(t) of the output pressure, i is the input eigenvalue fi (t) is the number of

[0072] The data [f1(t), f2(t), P(t)] in the sample set S are converted into the form of reliability distribution using the given reference value to obtain the evidence table of the input value.

[0073] Calculate the input eigenvalue f i Matching reference value of (t) The similarity distribution of is expressed as formula (1)-(3):

[0074]

[0075] α i,j' =0 j'=1,...,J i ,j'≠j,j+1 (3)

[0076] Among them, α i,j Represents the input eigenvalue f i (t) Matching reference value Similarity

[0077] The calculated pressure output value P(t) matches the reference value D n The similarity is expressed as formula (4)-(5):

[0078] S(P(t))={(D n ,γ n )|n=1,...,N} (4)

[0079]

[0080] γ n' =0 n'=1,...,N,n'≠n,n+1 (6)

[0081] Among them, γ n It represents the relationship between P(t) and the reference value D n Confidence level;

[0082] Furthermore, the sample calculates the confidence distribution (α i,j γ n ,α i,j+1 γ n ,α i,j γ n+1 ,α i,j+1 γ n+1 ), where α i,j γ n Represents a sample pair (f i (t),P(t)) The input value corresponds to the reference value P(t) corresponds to the reference value D nThe comprehensive similarity of . According to the calculated confidence distribution, table (1) is constructed, where a n,j represents all f i (t) Corresponding reference value P(t) corresponds to the reference value D n The sample for (f i (t),P(t)) comprehensive similarity, Indicates that all P(t) corresponds to the reference value D n The sum of the comprehensive similarities of samples, represents f i (t) Corresponding reference value The sum of the comprehensive similarities of the samples, and

[0083] Table 1: (f i (t), P(t)) reference value confidence distribution statistics table

[0084]

[0085]

[0086] Furthermore, according to Table (1), the likelihood of the table is normalized by formula (7):

[0087]

[0088] Get reference value The evidence is formula (8):

[0089]

[0090] Furthermore, according to formulas (7) and (8), evidence table (2) is constructed.

[0091] Table 2: Input f i (t) Evidence table

[0092]

[0093] The reliability factor r of the evidence is calculated by the relative change between the input value and the result value. i , describing the input information f i (t) Ability to calculate output pressure P(t).

[0094] In which, the input value f is defined i The calculation formulas for the relative change value of (t) and the result value P(t) are as follows:

[0095]

[0096]

[0097] Furthermore, an evaluation factor reflecting the input characteristic signal and the output pressure value can be obtained, which is expressed as formula (11):

[0098]

[0099] Furthermore, according to the obtained evaluation factor, the input information source f is calculated by formula (12): i (t) Reliability:

[0100]

[0101] Using the evidence table 2 and reliability factors obtained in the above steps, a rule base is established, and the weights are calculated by integrating the evidence to estimate the diesel engine working pressure, reflecting the temperature and current f i The linear relationship between (t) and pressure.

[0102] Among them, the kth rule in the confidence rule base can be expressed as follows:

[0103]

[0104] Furthermore, for the input f i (t), the evidence corresponding to each set of input and will be activated, input value f i The final proof of (t) can be obtained by and Weighted addition gives formulas (13) and (14):

[0105] e i ={(D n ,p n,i ),n=1,...,N} (13)

[0106]

[0107] Furthermore, the evidence e1 and e2 of f1(t) and f2(t) are obtained by formula (13) and formula (14), and the initial evidence weight w is set i =r i , using the evidence reasoning rules for fusion, expressed as formulas (15) and (16):

[0108] O(f i (t))={(D n ,p n,e(2) ),n=1,...,N} (15)

[0109]

[0110] Furthermore, the diesel engine working pressure is estimated based on the inference fusion result, which is expressed as formula (17):

[0111]

[0112] A confidence rule base parameter optimization model is constructed based on mean square error.

[0113] The training set As the input of the confidence rule base constructed in the above step, the inferred estimated pressure is generated, where w i represents the weight of evidence, and the other parameters are set as D1=l1, D N =l2,

[0114] Furthermore, the mean square error between the diesel engine working pressure estimated by the confidence rule base and the actual pressure is set as the objective function, which is expressed as formulas (18)-(21):

[0115]

[0116] 0≤w i ≤1,i=1,2 (19)

[0117]

[0118] D2 <D3<…<D N-1 (twenty one)

[0119] The optimization process based on the training samples, that is, by adjusting the values ​​of the confidence rule base parameters, makes the value of the objective function reach the minimum, so the optimal parameter value of the model can be obtained; the optimization can be achieved through the nonlinear optimization function Fmincon in the Matlab optimization toolbox.

[0120] In this step, formulas (18)-(21) represent the conditions that the optimization parameters must satisfy.

[0121] Finally, the rule base with the best index parameter value is the optimized confidence rule base. After the optimized parameters are determined, the root mean square error of the model can be tested through the test set, and the pressure can be predicted based on the optimized confidence rule base.

[0122] In summary, by collecting data under different temperature conditions, the present invention can take into account the impact of temperature on the performance of the pressure sensor, thereby enhancing the adaptability and accuracy of the calibration model, so that the calibrated sensor can maintain high accuracy under a wider range of environmental conditions.

[0123] The present invention converts current and temperature data into characteristic values, constructs evidence based on these characteristic values, and then performs reliability assessment on the evidence. The method can accurately describe the relationship between sensor output and true pressure. The introduction of reliability factors further improves the accuracy of the estimation results and reduces errors.

[0124] The present invention utilizes fusion rules to fuse multiple pieces of evidence (including evidence of current and temperature) into comprehensive evidence. This process fully considers the contribution of information from different sources to pressure estimation, effectively avoids the deviation that may be caused by a single information source, and improves the comprehensiveness and reliability of calibration.

[0125] The present invention sets an optimization objective function and establishes an evidential reasoning optimization model. The method can automatically adjust and optimize calibration parameters to adapt to the needs of different sensors or different working environments, making the calibration process more flexible and efficient.

[0126] The present invention applies the optimized confidence rule base optimization model to actual scenarios, which can realize real-time calibration of the pressure sensor output. This is particularly important for application scenarios that require continuous and accurate monitoring of pressure changes, such as industrial automation, aerospace and other fields.

[0127] The structure of the trust rule base of the present invention is clear, easy to understand and operate. With the addition of new data or new types of sensors, the rule base can be easily updated or expanded to adapt to new application scenarios or improve calibration accuracy. At the same time, the transparency of the rules also makes system maintenance and debugging more convenient.

[0128] Since the present invention takes into account the influence of various conditions and factors and adopts strategies such as multi-source information fusion and reliability evaluation, the method shows strong robustness in the face of complex and changeable working environments and can maintain high calibration accuracy and stability.

[0129] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0130] The above-mentioned embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by technicians in this field on the basis of the present invention shall fall within the scope of protection required by the present invention.

Claims

1. A state estimation method for a pressure sensor based on a confidence rule base algorithm, characterized in that: The method comprises the following steps: Step S1, applying a series of pressure values ​​within a preset range to the pressure sensor under different temperature conditions, and recording corresponding current data and ambient temperature data; Step S2, converting current data and temperature data into characteristic values; Step S3, constructing corresponding evidence for each eigenvalue based on the extracted eigenvalues, including current eigenvalue evidence and temperature eigenvalue evidence; Step S4, performing reliability assessment on each piece of evidence, i.e. determining its credibility in estimating the actual pressure; assigning a reliability factor to each piece of evidence based on the reliability assessment result; Step S5, using fusion rules to fuse multiple pieces of evidence into comprehensive evidence, and establishing a confidence rule base to conduct a preliminary assessment test on work pressure; Step S6, setting the optimization objective function and establishing the evidence reasoning optimization model; Step S7, applying the optimized confidence rule base optimization model to the actual scene to calibrate the output of the pressure sensor in real time.

2. The method according to claim 1, characterized in that: In step S1, different temperatures C are collected respectively. i Under the condition of 0-4 bar, different current values ​​i of the pressure sensor are set, and K current reference levels and M temperature reference levels are set.

3. The method according to claim 1, characterized in that: In step S2, the collected current data and temperature data are converted into characteristic values ​​f1(t) and f2(t) through similarity normalization processing, the output pressure is recorded as P(t), and the reference value set of the pressure output result of the pressure sensor is set to D = {D n |n=1,…,N}, represent f1(t), f2(t) and P(t) as the sample set S={[f1(t),f2(t),P(t)]|t=1,2,…T s }; Wherein, N is the number of result reference values ​​P(t) of the output pressure, and S=[f1(t), f2(t), P(t)] is a sample size.

4. The method according to claim 3, characterized in that: In step S3, the data [f1(t), f2(t), P(t)] in the sample set S is converted into the form of reliability distribution using a given reference value; Calculate the confidence distribution of all data sets in S (α i,j γ n ,α i,j+1 γ n ,α i,j γ n+1 ,α i,j+1 γ n+1 ), construct (f i (t), P(t)) reference value confidence distribution statistics table; Among them, α i,j γ n Represents a sample pair (f i (t),P(t)) The input value corresponds to the reference value P(t) corresponds to the reference value D n The comprehensive similarity.

5. The method according to claim 4, characterized in that: Yes (f i (t), P(t)) is normalized by likelihood, and the reference value A is obtained. j i Evidence, construct input f i (t) Evidence table.

6. The method according to claim 5, characterized in that: In step S4, the confidence distribution matrix is ​​transformed to obtain a pressure prediction model; According to the input value f i The relative change value of (t) and the result value P(t) is used to calculate the evaluation factor reflecting the input characteristic signal and the output pressure value, thereby calculating the input information source f i (t) reliability factor.

7. The method according to claim 6, characterized in that: In step S5, using the obtained input f i Based on the fusion results, a confidence rule base is established to conduct a preliminary evaluation test on the diesel engine working pressure C(t).

8. The method according to claim 7, characterized in that: In step S6, the mean square error between the diesel engine working pressure estimated by the confidence rule base and the actual pressure is set as the objective function to determine the optimized parameter set. Clarify the conditions that need to be met for the optimization parameters and build an evidential reasoning optimization model based on mean square error; Among them, w i represents the weight of evidence, and the other parameters are set as D1=l1, D N =l2, are the minimum and maximum values ​​of the input eigenvalues ​​f1(t) and f2(t), respectively, and l1 and l2 are the minimum and maximum values ​​of P(t), respectively.

9. The method according to claim 8, characterized in that: The constraints of the confidence rule base optimization model include: (a) the rule weight value interval is [0, 1]; (b) the premise attribute weight value interval of each rule is [0, 1]; (c) the confidence level shall not be greater than 1 or less than 0; (d) if the kth rule is complete, then the sum of all confidence levels in the output result is equal to 1; otherwise it is less than 1.

10. The method according to claim 8, characterized in that: The diesel engine operating pressure is estimated based on the fusion results, including the following formula:

Images