Method for analyzing reliability of avionics product by using multi-source interval data

Abstract

The invention relates to the technical field of data processing technologies, and discloses a method for analyzing the reliability of an avionics product by using multi-source interval data, which comprises the following steps: summarizing and classifying the multi-source interval data of a specified avionics product to obtain multiple groups of interval data, respectively substituting the interval data into an exponential distribution model, and constructing a corresponding quality function according to a unified recognition framework; acquiring a conflict coefficient and a credibility coefficient between quality functions corresponding to different groups of interval data, and determining a fusion weight coefficient of each quality function; obtaining weighted quality functions according to the fusion weight coefficient, fusing the weighted quality functions based on a D-S evidence theory to obtain a fusion quality function, and converting the fusion quality function according to a Pignistic probability conversion method; and obtaining the reliability of the avionics product at any moment according to a conversion result, and fitting to obtain a reliability curve about time change to execute reliability analysis.

Description

Technical Field

[0001] The present application relates to the technical field of data processing technology, and in particular to a method for analyzing the reliability of avionics products using multi-source interval data. Background Art

[0002] In recent years, my country's civil aircraft manufacturing industry has developed rapidly. Among them, avionics equipment is an important part of airborne equipment, and its safety has a significant impact on the overall safety of the aircraft. With the gradual localization of China's civil aircraft manufacturing, airborne avionics equipment is also gradually being localized.

[0003] Due to differences in equipment operating environments, tolerances in product design and production, as well as test observation methods and test measurement errors, the multi-source data obtained may not be accurate point data, but a range, that is, interval data. This is particularly evident in expert experience data. It is often difficult to make very accurate predictions based on engineering experience. In order to ensure the correctness of the data, only reasonable intervals can be obtained. During reliability testing and product service, when a product fails, it may not be observed immediately, and only interval data can be obtained. Therefore, interval data is an important part of basic safety data.

[0004] However, it is currently difficult to conduct accurate and reliable reliability analysis of domestic avionics equipment using traditional reliability analysis methods; therefore, data from different sources (experiments, simulations, expert experience, etc.) can be used for data fusion to conduct more accurate reliability analysis of domestic avionics equipment and obtain more reliable analysis results. Summary of the invention

[0005] In order to conduct more accurate reliability analysis on domestic avionics equipment and obtain more reliable analysis results, the present application provides a method for analyzing the reliability of avionics products using multi-source interval data, which adopts the following technical solutions: The method of analyzing the reliability of avionics products using multi-source interval data includes: After summarizing and classifying the multi-source interval data of the specified avionics product, multiple groups of interval data are obtained, which are respectively substituted into the exponential distribution model, and the corresponding quality function is constructed according to the unified recognition framework. ; Obtain the conflict coefficient and credibility coefficient between the quality functions corresponding to the interval data of different groups, and determine the fusion weight coefficient of each quality function ; Obtaining a weighted quality function according to the fusion weight coefficient, fusing the weighted quality function based on DS evidence theory to obtain a fusion quality function, and transforming the fusion quality function according to the Pignistic probability conversion method; Obtain the reliability of the avionics product at any time according to the conversion result, and obtain the reliability curve with respect to time variation by fitting to perform reliability analysis; in, is the group number of the interval data.

[0006] Optionally, the multi-source interval data includes expert experience data, product service data, test data, simulation data and manual data; The summary classification results include MTBF interval data and in Reliability interval data at the moment ; in, is the lower limit of the MTBF interval data, is the upper limit of the MTBF interval data, is the lower limit of the reliability interval data, It is the upper limit value of the reliability interval data.

[0007] Optionally, the method of substituting into the exponential distribution model and constructing a corresponding quality function according to a unified recognition framework includes: For group number Interval data of: Substitute the upper and lower limit values ​​of the interval data into the exponential distribution model to obtain the reliability interval data of the specified avionics product. The upper and lower limits of Construct the corresponding quality function according to the unified recognition framework ; in, is the lower limit of the reliability interval data, is the upper limit of the reliability interval data, , , , is the quality function The basic probability allocations to different subsets represent the operating states of the specified avionics products; The operating status includes empty set, working, failed, working or failed.

[0008] Optionally, the method of obtaining the conflict coefficients and credibility coefficients between the quality functions corresponding to the interval data of different groups and determining the fusion weight coefficients of the quality functions includes: The distance and similarity between two different quality functions are calculated using the Jouselme distance formula and the similarity coefficient formula respectively, which are used as conflict coefficients to measure the conflict between quality functions. The information entropy formula and standard deviation formula are used to measure the uncertainty of each quality function as a credibility coefficient to measure the credibility of the quality function itself. The product of the conflict coefficient and the credibility coefficient is normalized to obtain the fusion weight coefficient of each quality function.

[0009] Optionally, the method for obtaining the conflict coefficient includes: For quality function and the mass function Different arbitrary mass functions : The Jouselme distance formula and the similarity coefficient formula are used to calculate the quality function and any mass function The distance matrix between and the similarity matrix ; According to all the distance matrices and the similarity matrix Calculate the quality function The distance coefficient to all other mass functions and similarity coefficient ; Normalize all corresponding products of all distance coefficients and all similarity coefficients to obtain the conflict coefficient As support weight. Use the Josselme distance formula and similarity coefficient formula: in, and Different group numbers , The corresponding mass function; The D matrix is ​​defined as follows: ; According to the above formula, the distance matrix is ​​calculated and the similarity matrix ; Because it is generally believed that the smaller the conflict between a data and other data, the more reasonable the data is, so the following formula is used and , calculate the distance coefficient and similarity coefficient of each quality function with other quality functions, multiply them and normalize them to get the conflict coefficient as support weight.

[0010] Optionally, the method for obtaining the credibility coefficient includes: Obtain information entropy value through information entropy formula ; Get the standard deviation value using the standard deviation formula : The information entropy value and the standard deviation The reciprocals are taken respectively and then normalized, and the normalized results are multiplied and normalized again to obtain the credibility weight as the credibility weight.

[0011] in, To unify the recognition framework power set, A collection of propositions that form a unified recognition framework, Represents the subset in the recognition framework. A proposition, It is a state in the unified recognition framework power set The number of elements contained; and It is the upper and lower bounds of the corresponding interval data itself; Optionally, the method for acquiring a weighted quality function according to the fusion weight coefficient includes: according to , obtain the weighted quality function ; in, is the fusion weight of each quality function.

[0012] Optionally, the DS evidence theory includes: DS evidence theory fusion formula: in, is the support of the corresponding proposition in the fused quality function, is the conflict coefficient, and are the supports of the corresponding propositions in the quality function before fusion.

[0013] Optionally, the Pignistic probability conversion formula is used to distribute the basic probability of the non-single-focus element subset to the single-focus element subset, that is, the quality function is converted into a Bayesian quality function.

[0014] In summary, the present application includes at least one of the following beneficial technical effects: This application significantly improves the accuracy and comprehensiveness of avionics product reliability assessment through comprehensive processing and fusion analysis of multi-source interval data. First, through the classification modeling and quality function construction of multi-source interval data, the distribution characteristics of different data sources are effectively integrated, avoiding the limitations of single data source analysis. Secondly, the dynamic weight allocation mechanism based on the conflict coefficient and credibility coefficient can adaptively adjust the contribution of different quality functions, solving the fusion bias problem caused by evidence conflict in traditional methods. Through the fusion of DS evidence theory and Pignistic probability conversion, the reasonable transformation of uncertain information is achieved, so that the reliability analysis results not only retain the probabilistic characteristics of interval data, but also have clear engineering interpretability. The dynamic reliability curve finally generated intuitively reflects the reliability change trend of the product throughout its life cycle, and provides a highly reliable quantitative basis for fault prediction and maintenance decisions of avionics products. This application has stronger robustness in complex data environments, and is especially suitable for high-reliability product evaluation scenarios such as avionics equipment with multi-parameter coupling and diverse data sources. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of the reliability analysis method of avionics products in the present invention.

[0016] Figure 2 It is a schematic diagram of the fusion rule of the quality function in the present invention.

[0017] Figure 3 It is a curve showing the reliability of a certain avionics product in the present invention over time. DETAILED DESCRIPTION

[0018] Embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings.

[0019] In the description of this specification, the description with reference to the terms "certain embodiments", "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0020] The present application embodiment discloses a method for analyzing the reliability of avionics products using multi-source interval data, see Figure 1 ,include: After summarizing and classifying the multi-source interval data of the specified avionics product, multiple groups of interval data are obtained, which are respectively substituted into the exponential distribution model, and the corresponding quality function is constructed according to the unified recognition framework. ; Obtain the conflict coefficient and credibility coefficient between the quality functions corresponding to the interval data of different groups, and determine the fusion weight coefficient of each quality function ; Obtaining a weighted quality function according to the fusion weight coefficient, fusing the weighted quality function based on DS evidence theory to obtain a fusion quality function, and transforming the fusion quality function according to the Pignistic probability conversion method; Obtain the reliability of the avionics product at any time according to the conversion result, and obtain the reliability curve with respect to time variation by fitting to perform reliability analysis; in, is the group number of the interval data.

[0021] This application significantly improves the accuracy and comprehensiveness of avionics product reliability assessment through comprehensive processing and fusion analysis of multi-source interval data. First, through the classification modeling and quality function construction of multi-source interval data, the distribution characteristics of different data sources are effectively integrated, avoiding the limitations of single data source analysis. Secondly, the dynamic weight allocation mechanism based on the conflict coefficient and credibility coefficient can adaptively adjust the contribution of different quality functions, solving the fusion bias problem caused by evidence conflict in traditional methods. Through the fusion of DS evidence theory and Pignistic probability conversion, the reasonable transformation of uncertain information is achieved, so that the reliability analysis results not only retain the probabilistic characteristics of interval data, but also have clear engineering interpretability. The dynamic reliability curve finally generated intuitively reflects the reliability change trend of the product throughout its life cycle, and provides a highly reliable quantitative basis for fault prediction and maintenance decisions of avionics products. This application has stronger robustness in complex data environments, and is especially suitable for high-reliability product evaluation scenarios such as avionics equipment with multi-parameter coupling and diverse data sources.

[0022] In this embodiment of the present application, the multi-source interval data includes expert experience data, product service data, test data, simulation data and manual data; the summary classification results include MTBF interval data and in Reliability interval data at the moment ;MTBF interval data is the mean time between failures interval data.

[0023] in, is the lower limit of the MTBF interval data, is the upper limit of the MTBF interval data, is the lower limit of the reliability interval data, It is the upper limit value of the reliability interval data.

[0024] In this embodiment, the MTBF interval data and reliability interval data of the LED lamp model CREEXPE2-R4 obtained by simulation are as follows: Interval data 1 is MTBF interval data (17520h, 26280h); Interval data 2 is MTBF interval data (21000h, 43000h); Interval data 3 is Reliability interval data at 20,000 hours (0.6065, 0.6703, 2000); Interval data 4 is the MTBF interval data of similar products (30000h, 40000h).

[0025] The interval data Substituting the upper and lower limit values ​​into the exponential distribution model, the upper and lower limit values ​​of the reliability interval data of the specified avionics product are obtained accordingly; is the lower limit of the reliability interval data, is the upper limit of the reliability interval data; when hour, ; when hour, ; when hour, , and the solution is , we can further get ; when hour, ; In this way, the reliability interval data corresponding to the 4 interval data can be obtained as follows: , , and ; After that, substitute it into the unified recognition framework to construct the corresponding quality function : in, , , , is the quality function The basic probability distribution of different subsets, i.e., propositions, respectively represents the operating status of the specified avionics product; the operating status includes an empty set, working, failure, working or failure; From the perspective of set theory, the unified identification framework Θ is a set consisting of all possible mutually exclusive and complete elements. For example, in this paper, the status of a product, the identification framework Θ is {working, failed}, these two subsets constitute a complete set (completeness), covering all possible results, there are no other possibilities, and the two elements "working" and "failed" are mutually exclusive (mutual exclusivity), and cannot appear at the same time.

[0026] The unified identification framework clearly defines all possible outcomes of the problem under consideration. By determining the identification framework, the scope of objects handled by the theory of evidence can be clarified, irrelevant information can be avoided from being taken into consideration, and a clear boundary can be provided for subsequent evidence analysis and fusion. The quality function is one of the core concepts of the theory of evidence. It is defined based on the identification framework. The quality function assigns basic probabilities to various subsets of the identification framework, which represent different propositions or hypotheses.

[0027] Please refer to the following Table 1 for details.

[0028] Table 1: Quality function results corresponding to interval data Then, the distance and similarity between two different quality functions are calculated using the Jousselme distance formula and the similarity coefficient formula, which are used as conflict coefficients to measure the conflict between quality functions. The information entropy formula and standard deviation formula are used to measure the uncertainty of each quality function as a credibility coefficient to measure the credibility of the quality function itself. The product of the conflict coefficient and the credibility coefficient is normalized to obtain the fusion weight coefficient of each quality function.

[0029] Specifically, the method for obtaining the conflict coefficient includes: For quality function and the mass function Different arbitrary mass functions : The Jouselme distance formula and the similarity coefficient formula are used to calculate the quality function and any mass function The distance matrix between and the similarity matrix ; According to all the distance matrices and the similarity matrix Calculate the quality function The distance coefficient to all other mass functions and similarity coefficient ; Normalize all corresponding products of all distance coefficients and all similarity coefficients to obtain the conflict coefficient as support weight.

[0030] Use the Jouselme distance formula and similarity coefficient formula: in, and Different group numbers , The corresponding mass function; The D matrix is ​​defined as follows: ; According to the above formula, taking the 20,000-hour regional data 3 as an example, the distance matrix is ​​calculated and the similarity matrix ; Because it is generally believed that the smaller the conflict between a data and other data, the more reasonable the data is, so the following formula is used and , calculate the distance coefficient between each mass function and the other mass functions and similarity coefficient , multiply each other and then normalize to get the conflict coefficient As the support weight, is the total number of interval data.

[0031] On the other hand, the method for obtaining the credibility coefficient includes: Obtain information entropy value through information entropy formula ; Get the standard deviation value using the standard deviation formula : The information entropy value and the standard deviation The reciprocals are taken respectively and then normalized, and the normalized results are multiplied and normalized again to obtain the credibility weight as the credibility weight.

[0032] Specific: in, To unify the recognition framework power set, A collection of propositions that unify the recognition framework. Represents the subset in the recognition framework. A proposition, It is a state in the unified recognition framework power set The number of elements contained; and It is the upper and lower bounds of the corresponding interval data itself; all interval data need to calculate the standard deviation, but because there are two types of interval data in the case, the two types of interval data need to be converted into the same type before calculation. In this article, the reliability interval data is converted into MTBF interval data before calculation.

[0033] Conversion process: Find the reliability interval data and , and then take the reciprocal to get the MTBF interval data.

[0034] Before calculation, the original interval data must be uniformly converted into the MTBF interval form and then the standard deviation is calculated. The uncertainty and standard deviation of each interval data are calculated using the above formula; Table 2: Information entropy of quality function and standard deviation of interval data It is generally believed that the smaller the uncertainty of a data is, the better the quality of the data is. Therefore, the calculated information entropy and standard deviation are normalized after taking the reciprocals respectively, and then the normalized results are multiplied and normalized again as the credibility weight.

[0035] Optionally, the method for acquiring a weighted quality function according to the fusion weight coefficient includes: according to , obtain the weighted quality function ; in, is the fusion weight of each quality function.

[0036] Then use the DS evidence theory fusion formula: , the weighted quality function is fused; Figure 2 The figure shows a schematic diagram of the fusion principle.

[0037] in, is the support of the corresponding proposition in the fused quality function, is the conflict coefficient, which plays a normalization role in the formula. and are the supports of the corresponding propositions in the quality function before fusion.

[0038] Using the Pignistic probability conversion formula: , the basic probability of the non-single-focus subset is assigned to the single-focus subset, that is, the quality function is converted into a Bayesian quality function. The result obtained by fusion according to the combination rule of evidence theory is still a quality function, which contains the basic probability distribution of each subset under the recognition framework, not a probability in the strict sense, so it is necessary to use Pignistic probability conversion to convert the basic probability distribution into probability, so as to facilitate the next step of reliability analysis.

[0039] After Pignistic probability transformation and , respectively represent the reliability and failure rate of the product; Mean time between failures of products subject to exponential distribution and characteristic life The failure rate The reciprocal of Among them, the characteristic life is the usage time corresponding to when the product reliability drops to 0.368.

[0040] Using the conversion result of the previous step, we can get the reliability of the avionics product at a certain moment. Taking different moments and using the above steps, we can get the reliability of the product at different moments. If we take a large number of moments and repeat the above steps, we can get a large amount of reliability data of the product, and then use curve fitting to get the curve of the product's reliability over time.

[0041] like Figure 3 As shown in , according to the change of the reliability curve, it can be analyzed that the median life of the LED lamp is 25871h, the characteristic life is 28651h, and its reliability at some times is shown in Table 3 below.

[0042] Table 3: Reliability of LED lights at certain times Through the reliability curve of the product, we can obtain the time corresponding to the reliability of 0.368, which is the MTBF of the product.

[0043] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for analyzing the reliability of avionics products using multi-source interval data, characterized in that: include: After summarizing and classifying the multi-source interval data of the specified avionics product, multiple groups of interval data are obtained, which are respectively substituted into the exponential distribution model, and the corresponding quality function is constructed according to the unified recognition framework. ; Obtain the conflict coefficient and credibility coefficient between the quality functions corresponding to the interval data of different groups, and determine the fusion weight coefficient of each quality function ; Obtaining a weighted quality function according to the fusion weight coefficient, fusing the weighted quality function based on DS evidence theory to obtain a fusion quality function, and transforming the fusion quality function according to the Pignistic probability conversion method; Obtain the reliability of the avionics product at any time according to the conversion result, and obtain the reliability curve with respect to time variation by fitting to perform reliability analysis; in, is the group number of the interval data.

2. The method for analyzing the reliability of avionics products using multi-source interval data according to claim 1, characterized in that: The multi-source interval data includes expert experience data, product service data, test data, simulation data and manual data; The summary classification results include MTBF interval data and in Reliability interval data at the moment ; in, is the lower limit of the MTBF interval data, is the upper limit of the MTBF interval data, is the lower limit of the reliability interval data, It is the upper limit value of the reliability interval data.

3. The method for analyzing the reliability of avionics products using multi-source interval data according to claim 1, characterized in that: The method of substituting into the exponential distribution model and constructing the corresponding quality function according to the unified recognition framework includes: For group number Interval data of: Substituting the upper and lower limit values ​​of the interval data into the exponential distribution model, and correspondingly obtaining the upper and lower limit values ​​of the reliability interval data of the specified avionics product; Construct the corresponding quality function according to the unified recognition framework ; in, is the lower limit of the reliability interval data, is the upper limit of the reliability interval data, , , , is the quality function The basic probability allocations to different subsets represent the operating states of the specified avionics products; The operating status includes empty set, working, failed, working or failed.

4. The method for analyzing the reliability of avionics products using multi-source interval data according to claim 1, characterized in that: The method of obtaining the conflict coefficients and credibility coefficients between the quality functions corresponding to the interval data of different groups and determining the fusion weight coefficients of the quality functions includes: The distance and similarity between two different quality functions are calculated using the Jouselme distance formula and the similarity coefficient formula respectively, which are used as conflict coefficients to measure the conflict between quality functions. The information entropy formula and standard deviation formula are used to measure the uncertainty of each quality function as a credibility coefficient to measure the credibility of the quality function itself. The product of the conflict coefficient and the credibility coefficient is normalized to obtain the fusion weight coefficient of each quality function.

5. The method for analyzing the reliability of avionics products using multi-source interval data according to claim 4 is characterized in that: The method for obtaining the conflict coefficient includes: For quality function and the mass function Different arbitrary mass functions : The Jouselme distance formula and the similarity coefficient formula are used to calculate the quality function and any mass function The distance matrix between and the similarity matrix ; According to all the distance matrices and the similarity matrix Calculate the quality function The distance coefficient to all other mass functions and similarity coefficient ; Normalize all corresponding products of all distance coefficients and all similarity coefficients to obtain the conflict coefficient as support weight.

6. The method for analyzing the reliability of avionics products using multi-source interval data according to claim 4, characterized in that: The method for obtaining the credibility coefficient includes: Obtain information entropy value through information entropy formula ; Get the standard deviation value using the standard deviation formula : The information entropy value and the standard deviation The reciprocals are taken respectively and then normalized, and the normalized results are multiplied and normalized again to obtain the credibility weight as the credibility weight.

7. The method for analyzing the reliability of avionics products using multi-source interval data according to claim 1, characterized in that: The method for obtaining a weighted quality function according to the fusion weight coefficient includes: according to , obtain the weighted quality function ; in, is the fusion weight of each quality function.

8. The method for analyzing the reliability of avionics products using multi-source interval data according to claim 7, characterized in that: The DS-based evidence theory includes: DS evidence theory fusion formula: in, is the support of the corresponding proposition in the fused quality function, is the conflict coefficient, and are the supports of the corresponding propositions in the quality function before fusion.

9. The method for analyzing the reliability of avionics products using multi-source interval data according to claim 8, characterized in that: The method for converting the fusion quality function according to the Pignistic probability conversion method includes: The Pignistic probability conversion formula is used to distribute the basic probability of the non-single-focus element subset to the single-focus element subset, and the quality function is converted into a Bayesian quality function.

Images