A valve multi-parameter combined fault early warning method for engine test

Abstract

The application discloses a valve multi-parameter joint fault early warning method for engine test, and aims at solving the technical problems of high cost, low operation efficiency and difficulty in finding hidden faults in the system caused by too large data packet of valve fault early warning collection, which needs large storage device during engine test. Specifically, the method comprises the following steps: step 1, establishing a fault model library, wherein the fault model library comprises a plurality of fault rules, and the fault rules are composed of known faults of the valve and corresponding fault features; step 2, collecting all parameters of the valve; step 3, according to the parameter fragment interception method, at least one of the following methods is used to characterize all collected parameters of the valve and extract fault features: mode one, 3D data analysis method combining time domain and frequency domain; mode two, data fusion method based on sound signal interception window and interception density; and step 4, finding corresponding known faults in the fault model library according to the fault features and performing early warning.

Description

Technical Field

[0001] The present invention relates to a valve fault early warning method, and in particular to a multi-parameter combined fault early warning method for a valve used in an engine test. Background Art

[0002] Liquid rocket engine testing, due to the high cost of engine testing itself, places extremely high demands on test system reliability. Valve failures, particularly in the main supply pipeline, often lead to test failure. This is particularly true in the testing of new-generation liquid oxygen / kerosene and liquid hydrogen / liquid oxygen engines, where valves operate in low temperatures, high pressures, and vibrating environments. Valve failures can be very common, resulting in serious consequences and compromising the reliability of the test system.

[0003] In order to improve the reliability of the test system, it is necessary to provide fault warning for the key valves in the test. The traditional valve fault warning method usually uses multiple sensors to collect multiple parameters of each valve during the test, and then performs simple calculations on the collected parameters. The fault type and fault location are diagnosed based on the calculation results.

[0004] Existing fault warning methods have the following two main problems:

[0005] First, the collected data includes all data in the entire working state of the valve, and the resulting data packet is large. This not only requires a larger storage device, but also puts greater pressure on the system calculation, reduces the efficiency of fault diagnosis, and increases the cost of fault diagnosis. In order to reduce the size of the data packet, the data collection density is sometimes reduced, but this also reduces the amount of data at key points, making fault location difficult.

[0006] Second, the fusion method of the collected data is simple, and parameter fusion is only performed based on the time correspondence, which makes some hidden faults difficult to be discovered, resulting in low fault location accuracy.

[0007] Therefore, a simple and effective valve failure early warning method is urgently needed to improve the reliability of the test system. Summary of the Invention

[0008] The present invention provides a multi-parameter joint fault warning method for valves used in engine testing to solve the technical problems in engine testing that the data packets collected by existing valve fault warning methods are too large, requiring large storage devices, resulting in high fault warning costs, low computing efficiency, and difficulty in discovering hidden faults in the system.

[0009] In order to achieve the above-mentioned object, the present invention provides a multi-parameter combined fault warning method for an engine test valve, which is special in that it includes the following steps:

[0010] Step 1, establishing a fault model library, the fault model library comprising a plurality of fault rules, the fault rules consisting of known faults of the valve and corresponding fault features thereof;

[0011] Step 2, collecting all parameters of the valve;

[0012] Step 3, according to the parameter segment interception method, at least one of the following ways is used to characterize all the collected parameters of the valve, and the fault features are extracted:

[0013] Method one, 3D data analysis method combining time domain and frequency domain;

[0014] Method two, data fusion method based on sound signal interception window and interception density;

[0015] Step 4, according to the fault features, find the corresponding known faults in the fault model library and give a warning.

[0016] Further, method one specifically includes:

[0017] A1, for the time-amplitude curve in the collected all parameters of the valve, intercepting in different frequency domains, calculating all the slopes of the intercepted curves, and forming the slope change table of different frequency domains respectively;

[0018] A2, comparing the slopes of different frequency domains at the same time, placing the most relevant parameters in the 3D coordinate system to form a 3D data fusion atlas based on frequency domain and time domain; in the 3D coordinate system, the X-axis is time, the Y-axis is frequency domain, and the Z-axis is the amplitude of each parameter;

[0019] A3, according to the known amplitude curve of different faults for analysis and processing, filtering irrelevant parameters in the 3D data fusion atlas, and taking relevant parameters as fault features.

[0020] Further, method two specifically includes:

[0021] B1, collecting the sound data when the valve vibrates, and establishing a time-sound intensity curve;

[0022] B2, intercepting the position where the sound intensity value in the sound parameter axis appears a peak, and recording the start and end time of the peak;

[0023] B3, according to the start and end time of the peak, intercepting the remaining parameters of the valve during the period when the peak sound appears, and obtaining the remaining parameters during the period when the peak sound appears;

[0024] B4, according to the sound intensity value, data screening is performed on the remaining parameters during the period when the peak sound appears, and the screening principle is: the higher the sound intensity value, the greater the screening density;

[0025] B5. Correlate the filtered data with the peak of the time-sound intensity curve when the fault occurs to extract the fault characteristics.

[0026] Furthermore, the data screening is specifically as follows:

[0027] When the sound intensity value is less than 40DB, data screening is performed in the form of 1 data point / s;

[0028] When the sound intensity value is 40DB≤<50DB, data screening is performed in the form of 10 data points / s;

[0029] When 50DB≤sound intensity value≤60DB, data screening is performed in the form of 50 data points / s.

[0030] Furthermore, step 1 is specifically as follows:

[0031] 1.1. Establish fault rules based on the known fault parameters of the diagnosis object and the corresponding fault characteristics. Multiple fault rules constitute a fault model library;

[0032] 1.2. Improve the fault rules based on the mathematical model and diagnosis results;

[0033] 1.3. Update the fault model library through learning and training of improved fault rules.

[0034] Furthermore, the remaining parameters in step B3 at least include pressure parameters, temperature parameters, strain parameters, vibration parameters and sound parameters.

[0035] Beneficial effects of the present invention:

[0036] 1. The present invention establishes a valve fault model library, and establishes multiple parameter segment interception methods based on multiple common valve parameters. By establishing a 3D data analysis method combining time domain and frequency domain, the effective fusion of multiple parameters is achieved, and only effective parameters are intercepted, thereby improving the accuracy of fault diagnosis, reducing the size of the data acquisition package, reducing storage costs, and improving computing efficiency.

[0037] 2. This invention also establishes a data fusion method based on the interception window and interception density of sound signals, which improves the effectiveness of collected data and the data density of fault time. It can better complete fault warning and fault location determination, meeting the requirements of high-reliability valve multi-parameter joint fault warning.

[0038] 3. The present invention combines multiple parameters for joint fault diagnosis. Combined with the distribution of vibration data in the time domain and frequency domain, it can discover hidden faults, make the fault diagnosis results more accurate, and can quickly and accurately locate faults.

[0039] 4. The present invention can continuously perform artificial intelligence training on the fault model library through mathematical models and data cases, thereby continuously enriching the fault model library to improve the accuracy of fault diagnosis results.

[0040] 5. The early warning method provided by the present invention can effectively increase the data collection density in the fault area, reduce the amount of invalid data in the non-fault area, and improve the utilization efficiency of data.

[0041] 6. The present invention combines the frequency domain and the time domain to analyze and process the collected data. Compared with the existing technology that analyzes and processes the collected data only based on the time relationship, the present invention is more refined and accurate, thereby improving the accuracy of the fault diagnosis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a time-intensity curve intercept provided by an embodiment of the present invention;

[0043] Figure 2 This is a data decomposition diagram based on the frequency domain and time domain provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0045] Principle of the Invention: The present invention combines a variety of valve parameters to quickly and effectively diagnose valve faults. This is different from the traditional fault diagnosis mode, in which the collection of general data or feature points is only based on the time axis. Although the time axis can achieve continuity from the parameters themselves, the valve movement is often nonlinear, so the overlap between the collected data and the fault occurrence process on the time axis is insufficient. The present invention adopts a method that combines the time axis fault occurrence and the sound axis fault to establish a fault correlation map, which is highly targeted and improves the reliability of fault diagnosis. The details are as follows:

[0046] A multi-parameter combined fault early warning method for an engine test valve includes the following steps:

[0047] Step 1: Establish a fault model library, wherein the fault model library includes multiple fault rules, and the fault rules are composed of known fault parameters of the valve and their corresponding fault characteristics;

[0048] 1.1. Establish fault rules based on the known fault parameters of the diagnosis object and the corresponding fault characteristics. Multiple fault rules constitute a fault model library;

[0049] Based on the actual characteristics and usage of the valve, fault rules are established, primarily for major faults such as internal leakage and fracture that cause the valve to fail to open or close properly during use. The fault model library can be constructed based on fault rules and fault characteristics under the fault diagnosis reasoning model. Known symptoms of the diagnostic object are stored in the fault rule library. The acquired valve parameters are matched against the fault rule library to determine whether they meet the rules. The most likely fault cause for the selected fault characteristic is then determined, including causes such as cylinder internal leakage, valve internal leakage, and valve sticking.

[0050] 1.2. Improve the fault rules based on the mathematical model and diagnosis results;

[0051] 1.3. Update the fault model library through learning and training of improved fault rules.

[0052] When the fault rules are enriched to a certain extent, a fault model library for fault reasoning and diagnosis can be established based on machine learning methods trained with mathematical models and data cases. The knowledge can be stored in the fault model library in the form of fault case data, and the fault model library can be improved through learning and training.

[0053] Step 2: Collect all valve parameters. Read data: Read various valve parameters (primarily including pressure, temperature, strain, vibration, and sound data). Display information: After the data is read, the right side of the front-end display first displays data acquisition-related information, including the number of monitored channels, physical quantity, sampling rate, etc. Analysis settings: Based on the information displayed for each channel in the previous step, select the corresponding data display frequency, upper and lower limits, and useless data filtering criteria for each channel.

[0054] Step 3: Based on the parameter segment extraction method, characterize all collected valve parameters using at least one of the following methods to extract fault features:

[0055] Method 1: 3D data analysis method combining time domain and frequency domain;

[0056] Specifically include:

[0057] A1. For the time-amplitude curves of all valve parameters collected, intercept them in different frequency domains, calculate all the slopes of the intercepted curves, and form slope change tables for different frequency domains respectively;

[0058] For example, a pressure change data is analyzed, a change curve in the pressure change data is intercepted, and a change slope calculation is performed on its change parameter, and a slope change table is formed through calculation.

[0059] The vibration data collected is intercepted in different frequency domains, and the slopes of the change curves formed by the changes in the intercepted data are calculated to form a slope change table for each frequency segment.

[0060] A2. See Figure 2 , compare the slopes of different frequency domains at the same time, place the parameters with the strongest correlation in a 3D coordinate system, and form a 3D data fusion map based on the frequency domain and time domain; in the 3D coordinate system, the X axis is time, the Y axis is frequency domain, and the Z axis is the amplitude of each parameter;

[0061] That is, the slope change tables at each time are compared, their correlation is analyzed, and the parameter is placed in the corresponding frequency coordinate system based on the principle of the strongest correlation.

[0062] A3. Analyze and process the known amplitude curves of different faults, filter out irrelevant parameters in the 3D data fusion map, and use relevant parameters as fault features.

[0063] If a fault is discovered during testing, the 3D data curve at the time of the fault is recorded, its relevant parameters are analyzed, and irrelevant parameters are filtered out. This creates a 3D data fusion map based on the frequency and time domains corresponding to the fault pattern. Through continuous accumulation and enrichment of the map, a rich data foundation is provided for fault diagnosis and machine intelligence learning.

[0064] Method 2: Data fusion method based on sound signal interception window and interception density;

[0065] Specifically include:

[0066] B1. Collect the sound data of the valve during vibration and establish a time-sound intensity curve;

[0067] B2. See Figure 1 , intercepting the position where the sound intensity value in the sound parameter axis has a peak, and recording the start and end time of the peak;

[0068] B3. intercepting the remaining parameters of the valve according to the start and end time of the peak, and obtaining the remaining parameters during the period when the peak sound occurs;

[0069] B4. Filter the remaining parameters during the period of the peak sound according to the sound intensity value. The filtering principle is: the higher the sound intensity value, the greater the filtering density;

[0070] The data screening is specifically as follows:

[0071] When the sound intensity value is less than 40DB, data screening is performed in the form of 1 data point / s;

[0072] When the sound intensity value is 40DB≤<50DB, data screening is performed in the form of 10 data points / s;

[0073] When 50DB≤sound intensity value≤60DB, data screening is performed in the form of 50 data points / s.

[0074] This method can effectively increase the density of fault-prone areas and reduce the amount of invalid data in non-faulty areas, thereby improving data utilization efficiency.

[0075] B5. Correlate the filtered data with the peak of the time-sound intensity curve when the fault occurs to extract the fault characteristics.

[0076] Through the correlation analysis between fault modes and parameters based on the time axis and parameters based on the sound generation axis, the sound sensitivity parameters are located. The maximum value can be well combined with these related parameters through the sound change amplitude to improve the reliability of fault diagnosis. Finally, the fault diagnosis analysis results can be displayed to make the diagnosis results visual.

[0077] Step 4: Search for the corresponding fault in the fault model library according to the fault characteristics and issue an early warning.

[0078] When a fault occurs, a correlation is established with the corresponding graph. As the data becomes richer, an extended correlation analysis of possible faults is performed to establish a fault warning method.

[0079] Through the description of the above implementation methods, technical personnel in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0080] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions within the technical scope disclosed by the present invention shall be covered by the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A multi-parameter combined fault warning method for valves used in engine testing, characterized in that: The following steps are involved: Step 1: Establish a fault model library, wherein the fault model library includes multiple fault rules, and the fault rules are composed of known faults of the valve and their corresponding fault characteristics; Step 2: Collect all valve parameters; Step 3: Based on the parameter segment extraction method, characterize all the collected valve parameters using at least one of the following methods to extract fault features: Method 1: 3D data analysis method combining time domain and frequency domain; specifically including: A1. For the time-amplitude curves of all valve parameters collected, intercept them in different frequency domains, calculate all the slopes of the intercepted curves, and form slope change tables for different frequency domains respectively; A2. Compare the slopes of different frequency domains at the same time and place the parameters with the strongest correlation in a 3D coordinate system to form a 3D data fusion map based on the frequency domain and time domain; in the 3D coordinate system, the X-axis is time, the Y-axis is frequency domain, and the Z-axis is the amplitude of each parameter; A3. Analyze and process the amplitude curves of different known faults, filter out irrelevant parameters in the 3D data fusion map, and use relevant parameters as fault features; Method 2: Data fusion method based on sound signal interception window and interception density; specifically including: B1. Collect the sound data of the valve during vibration and establish a time-sound intensity curve; B2. intercepting the position where the sound intensity value in the sound parameter axis has a peak, and recording the start and end time of the peak; B3. intercepting the remaining parameters of the valve according to the start and end time of the peak, and obtaining the remaining parameters during the period when the peak sound occurs; B4. Filter the remaining parameters during the period of the peak sound according to the sound intensity value. The filtering principle is: the higher the sound intensity value, the greater the filtering density; B5. Correlate the filtered data with the peak of the time-sound intensity curve when the fault occurs to extract the fault characteristics; Step 4: Search for corresponding known faults in the fault model library according to the fault characteristics and issue an early warning.

2. The multi-parameter combined fault warning method for engine test valves according to claim 1, characterized in that: The data screening is specifically as follows: When the sound intensity value is less than 40DB, data screening is performed in the form of 1 data point / s; When the sound intensity value is 40DB≤<50DB, data screening is performed in the form of 10 data points / s; When 50DB≤sound intensity value≤60DB, data screening is performed in the form of 50 data points / s.

3. The multi-parameter combined fault warning method for engine test valves according to claim 1 or 2, characterized in that: Step 1 is as follows: 1.

1. Establish fault rules based on the known fault parameters of the diagnosis object and the corresponding fault characteristics. Multiple fault rules constitute a fault model library; 1.

2. Improve the fault rules based on the mathematical model and diagnosis results; 1.

3. Update the fault model library through learning and training of improved fault rules.

4. The multi-parameter combined fault warning method for engine test valves according to claim 3, characterized in that: The remaining parameters in step B3 at least include pressure parameters, temperature parameters, strain parameters, vibration parameters and sound parameters.

Images