Fault analysis method for multivariable complex system

Abstract

The invention relates to the field of aircraft fuel oil, in particular to a fault analysis method for a multivariable complex system, which comprises the following steps of: 1, reading all test variable data of a ground simulation test of an aircraft fuel oil system; step 2, inputting fault variables in all test variable data of the ground simulation test of the aircraft fuel system, and determining a correlation coefficient threshold; step 3, performing correlation analysis on each fault variable; according to the method, aircraft fuel system ground simulation test data serves as a sample, aircraft fuel system ground simulation system fault and correlation analysis is achieved around fault variable data, correlation coefficients among fault variables are obtained through calculation, and fault detection is emphasized on strong correlation fault variables.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of aircraft fuel, in particular to a fault analysis method of a multivariate complex system. BACKGROUND

[0002] There are three analysis methods for the fault analysis of the current aircraft fuel system ground simulation test: one is experience analysis, this analysis method is simple, mainly relying on the experience of engineers to locate the problem, but the defects are very obvious, first of all, the engineers need to be familiar with the fuel system of the aircraft, which wastes a lot of energy and has great limitations; secondly, the aircraft fuel system is very complex, and experience analysis cannot accurately locate the problem, which may miss many influencing factors. The second is fault tree analysis, which is a top-down deductive failure analysis, which uses Boolean logic combination to analyze the initial failure of a complex system and the influence of events, but it cannot find all possible initial failures using fault tree analysis. The third is event tree analysis, which starts from the initial event (initiator) that does not want to appear, and according to the possible system events to a series of final results. This method uses a tree diagram called event tree to represent the logical relationship between the possible occurrence of an accident and the various causes of the accident, and through qualitative and quantitative analysis of the event tree, the main causes of the accident are found out, which provides a reliable basis for determining safety measures to achieve the purpose of guessing and preventing accidents. The disadvantage is that it may be influenced by subjective factors and needs to be compared and evaluated in many aspects. SUMMARY

[0003] The purpose of the present application is to solve the above-mentioned problems existing in the prior art, and to provide an aircraft fuel system ground simulation test correlation analysis method with wide application range, accurate analysis results and no influence of subjective factors.

[0004] The technical scheme of the present application: A fault analysis method of a multivariate complex system, comprising: Step 1, reading all test variable data of the aircraft fuel system ground simulation test; Step 2, inputting fault variables in all test variable data of the aircraft fuel system ground simulation test and determining a correlation coefficient threshold; Step 3, performing correlation analysis on each fault variable.

[0005] Further, step 1 is to establish a table name input port, import the data model of the aircraft fuel system ground simulation test, that is, import the table test data, remove the time column and variable row of the table, and store the row data.

[0006] Further, step 3 is to perform correlation analysis between multiple fault variables using the Pearson correlation coefficient model.

[0007] Further, step 3 specifically comprises: Step 3.1 Calculate the Pearson correlation coefficient between multiple fault variables; Step 3.2 Calculate the absolute value of the Pearson correlation coefficient calculated in step 3.1; Step 3.3 Compare the absolute value of the Pearson correlation coefficient obtained in step 3.2 with the correlation coefficient threshold, filter and output the fault variables whose absolute value is greater than or equal to the correlation coefficient threshold.

[0008] Further, step 3.1 calculates the Pearson correlation coefficient between any two fault variables X and Y by the following formula:

[0009] Wherein, the fault variable X is a data set containing n samples , the fault variable Y is a data set containing m samples , the Pearson correlation coefficient between the fault variables X and Y is , any sample in the fault variable X is , any sample in the fault variable Y is , the mean of all samples in the fault variable X is , the mean of all samples in the fault variable Y is ; and the Pearson correlation coefficient between all fault variables is calculated in the above manner.

[0010] Further,

[0011]

[0012] Wherein, n is the number of all samples in the fault variable X, and m is the number of all samples in the fault variable Y.

[0013] Further, in step 3.3, the fault variables whose absolute value of the Pearson correlation coefficient is not less than the correlation coefficient threshold are presented in tabular form.

[0014] Further, among all the fault variables whose absolute value of the Pearson correlation coefficient is not less than the correlation coefficient threshold, the correlation degree between the fault variables is determined.

[0015] Further, the correlation degree determination rules are as follows: The fault variables X and Y are extremely strongly correlated. The fault variables X and Y are strongly correlated. , the fault variables X, Y are weakly correlated; , the fault variables X, Y are weakly correlated; , the fault variables X, Y are weakly correlated; , the fault variables X, Y are weakly correlated.

[0016] The beneficial effects of the present application: the present application provides a fault analysis method of a multivariate complex system, taking the ground simulation test data of an aircraft fuel system as a sample, and focusing on the fault variable data, the fault and correlation analysis of the ground simulation system of the aircraft fuel system is realized, the correlation coefficient between each fault variable is calculated, and the fault detection of the strongly correlated fault variable is emphasized. The correlation strength between each variable can be directly observed from the data analysis result graph, and the correlation analysis result is obtained after screening. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The correlation analysis method of the multivariate complex system provided by the present application; Figure 2 The correlation analysis flowchart of the multivariate complex system provided by the present application; Figure 3 The correlation analysis program running result graph of the multivariate complex system provided by the present application; Figure 4 The man-machine interaction interface of the correlation analysis of the multivariate complex system provided by the present application; Figure 5 The output result graph of the correlation analysis of the multivariate complex system provided by the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0019] The present application provides a fault analysis method of a multivariate complex system, comprising: Step 1, read all test variable data of the aircraft fuel system ground simulation test, step 1 is to design a multivariate complex system correlation analysis program, and establish a table name input port, import the aircraft fuel system ground simulation test data model, that is, import the table test data, remove the time column and variable row of the table, and store the row data.

[0020] The imported test data model is mainly imported stored table test data, and the stored data dimensions are different, mainly including calibrated flow, pressure and other physical quantity parameters, and the test data storage amount reaches the order of one hundred million. Taking the aircraft fuel system ground simulation system and its pressure as an example, the pressure mainly includes pipeline pressure, pump pressure, radiator inlet and outlet pressure, etc. The various physical quantity parameters of the aircraft fuel system ground simulation system are connected with each other, coupled with each other, and influence each other.

[0021] Step 2, input the fault variables in all test variable data of the aircraft fuel system ground simulation test, and determine the correlation coefficient threshold; the fault variable name is input in the table name input port, the variable name in the table has been searched, and the search function is convenient and accurate. In addition, the multi-variable complex system correlation analysis program also has a correlation coefficient setting port, and the correlation coefficient is selected and input. The correlation coefficient input in this step is the correlation coefficient threshold for subsequent comparison with the absolute value of the Pearson correlation coefficient. The correlation coefficient setting port can adjust the correlation strength range, and can expand or narrow the fault troubleshooting range.

[0022] Step 3, correlation analysis is performed on each fault variable. The correlation analysis method in the application uses the Pearson correlation coefficient analysis method. This method measures the ratio of the product of the covariance between two variables and their standard deviations to represent the linear relationship between them. Step 3 specifically includes: Step 3.1, calculate the Pearson correlation coefficient between multiple fault variables; wherein the Pearson correlation coefficient between two fault variables X and Y is obtained through the following process: (1) first calculate the mean value of all samples in the fault variable X and the mean value of all samples in the fault variable Y :

[0023]

[0024] Wherein, n is the number of all samples in the fault variable X, and m is the number of all samples in the fault variable Y.

[0025] (2) then calculate the difference between each sample of the fault variable X and the mean value , the difference between each sample of the fault variable Y and the mean value :

[0026]

[0027] (3) calculate the product of the difference and sum:

[0028] (4) Calculate the product of the standard deviation:

[0029] (5) Calculate the Pearson correlation coefficient between any two failure variables X, Y:

[0030] Wherein, the failure variable X is a data set containing n samples , the failure variable Y is a data set containing m samples , the Pearson correlation coefficient between the failure variables X, Y is , the failure variable X is any sample , the failure variable Y is any sample , the mean of all samples in the failure variable X is , the mean of all samples in the failure variable Y is ; and the Pearson correlation coefficient between all failure variables is calculated in the above manner.

[0031] Step 3.2 Calculate the absolute value of the Pearson correlation coefficient calculated in step 3.1. Step 3.3 Compare the absolute value of the Pearson correlation coefficient obtained in step 3.2 with the correlation coefficient threshold value, filter and output the failure variables whose absolute value is greater than or equal to the correlation coefficient threshold value. In step 3.3, the failure variables whose absolute value of the Pearson correlation coefficient is greater than or equal to the correlation coefficient threshold value are filtered out and presented in table form, so that the distribution of variables from strong to weak correlation can be directly observed. Among all the failure variables whose absolute value of the Pearson correlation coefficient is greater than or equal to the correlation coefficient threshold value, the correlation degree between the failure variables is determined. The correlation degree judgment rules are as follows: The failure variables X, Y are extremely strongly correlated. The failure variables X, Y are strongly correlated. The failure variables X, Y are moderately correlated. The failure variables X, Y are weakly correlated. The failure variables X, Y are extremely weakly correlated. When

[0032] Referring to Figure 2 , Figure 3The present application firstly carries out correlation analysis modeling by MATLAB, takes historical data of an airplane fuel system ground simulation system as samples, realizes fault and correlation analysis of the airplane fuel system ground simulation system around abnormal data variables, calculates correlation coefficients between variables, and focuses on fault detection of strongly correlated variables.

[0033] Referring to Figure 4 , Figure 5 The correlation analysis program of the multivariable complex system designed by the present application provides input ports and output displays of a user interface, including: a data table import port, which can index a directory; a self-defined correlation coefficient value input port; a variable name input port, which can search; the user interface is output displayed in a table form, only displays variable names, and the results are sorted from strong to weak correlation.

[0034] The above merely describes specific embodiments of the present application, and detailed descriptions are provided, and the part not described in detail is conventional technology. However, the protection scope of the present application is not limited to this, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. The protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A fault analysis method for multivariable complex systems, characterized in that, include: Step 1: Read all test variable data from the ground simulation test of the aircraft fuel system; Step 2: Input the fault variables from all test variable data of the aircraft fuel system ground simulation test, and determine the correlation coefficient threshold; Step 3: Perform correlation analysis on each fault variable.

2. The fault analysis method for multivariable complex systems according to claim 1, characterized in that, Step 1 involves creating a table name input port and importing the aircraft fuel system ground simulation test data model, which means importing the table test data and storing the data after removing the time column and variable row from the table.

3. The fault analysis method for multivariable complex systems according to claim 2, characterized in that, Step 3 involves using the Pearson correlation coefficient model to perform correlation analysis among multiple fault variables.

4. The fault analysis method for multivariable complex systems according to claim 3, characterized in that, Step 3 specifically includes: Step 3.1 Calculate the Pearson correlation coefficient among multiple fault variables; Step 3.2 Calculate the absolute value of the Pearson correlation coefficient obtained in Step 3.1; Step 3.3 Compare the absolute value of the Pearson correlation coefficient obtained in Step 3.2 with the correlation coefficient threshold, and filter and output the fault variables whose absolute value is not less than the correlation coefficient threshold.

5. The fault analysis method for multivariable complex systems according to claim 4, characterized in that, Step 3.1 Calculate the Pearson correlation coefficient between any two fault variables X and Y using the following formula: Wherein, the fault variable X contains n samples The dataset contains m samples, where the fault variable Y is a subset of the dataset. The dataset, The Pearson correlation coefficient between fault variables X and Y. Let X be any sample from the fault variable X. For any sample in the fault variable Y, Let X be the mean of all samples in the fault variable X. Let Y be the mean of all samples in the fault variable Y; then, using the above method, calculate the Pearson correlation coefficient among all fault variables.

6. The fault analysis method for multivariable complex systems according to claim 5, characterized in that, Where n is the total number of samples in fault variable X, and m is the total number of samples in fault variable Y.

7. The fault analysis method for multivariable complex systems according to claim 6, characterized in that, In step 3.3, the fault variables whose absolute Pearson correlation coefficient is not less than the correlation coefficient threshold are presented in tabular form.

8. The fault analysis method for multivariable complex systems according to claim 7, characterized in that, Among all the selected fault variables whose absolute Pearson correlation coefficient is not less than the correlation coefficient threshold, the degree of correlation between the fault variables is determined.

9. The fault analysis method for multivariable complex systems according to claim 8, characterized in that, The rules for determining relevance are as follows: The fault variables X and Y are extremely strongly correlated; The fault variables X and Y are strongly correlated; The fault variables X and Y are moderately correlated. The fault variables X and Y are weakly correlated; The fault variables X and Y are extremely weakly correlated; When = 0, the fault variables X and Y are unrelated.

Images