An engine fault detection method based on multi-sensor and spectral data fusion

Abstract

The present invention relates to an engine fault detection method based on multi-sensor and spectral data fusion, comprising: collecting first spectral data, reference temperature and reference pressure of the engine tail flame of a normal engine, extracting the ideal spectral characteristics of a target substance based on the first spectral data; collecting second spectral data of the engine tail flame of an engine to be detected, obtaining the actual spectral characteristics of the target substance based on the second spectral data; collecting the actual tail flame temperature and actual tail flame pressure of the engine tail flame of the engine to be detected, and correcting the actual spectral characteristics to obtain the actual spectral correction characteristics; obtaining dynamic characteristics for judging whether the engine to be detected is faulty based on the ideal spectral characteristics and the actual spectral correction characteristics; constructing a dynamic threshold range; judging whether the dynamic characteristics are abnormal based on the dynamic threshold range. The present invention can be flexibly and accurately used for prediction and timely alarm of engine faults, and can fully reduce the occurrence of accidents.

Description

Technical Field

[0001] The present invention relates to the technical field of engine detection, and in particular to an engine fault detection method based on multi-sensor and spectral data fusion. Background Art

[0002] When liquid rocket engines work for a long time under high temperature and high pressure, it is easy to cause ablation and corrosion of materials, especially during the combustion process, key components such as the engine nozzle and pipeline may experience abnormal wear. This wear not only affects the service life of the engine, but may also lead to unstable combustion and thrust reduction, and even cause failures in severe cases. Therefore, it is of great significance to monitor the composition of combustion products in the tail flame of liquid rocket engines in real time, especially to detect abnormal changes in material composition.

[0003] At present, ultraviolet and visible spectroscopy technology is widely used in engine wake monitoring. The characteristic spectral lines of ultraviolet and visible spectra can effectively analyze the composition and concentration of combustion products. The spectral characteristics of metal elements (such as Ni, Cr, Fe, etc.) in the wake can reflect the combustion state and wear degree of engine materials. Using spectroscopy to detect changes in the content of these elements can help determine whether abnormal wear or ablation occurs in the combustion chamber, nozzle, and pipeline.

[0004] Traditional engine tail flame monitoring methods mostly rely on data analysis of a single spectrum or sensor, which is easily affected by environmental interference and engine operating condition fluctuations, resulting in reduced reliability of detection results. Therefore, a new detection scheme is urgently needed to improve its detection accuracy and effectiveness. Summary of the invention

[0005] The object of the present invention is to provide an engine fault detection method based on multi-sensor and spectral data fusion.

[0006] To achieve the above-mentioned object of the invention, the present invention provides an engine fault detection method based on multi-sensor and spectral data fusion, comprising the following steps:

[0007] S1. Collecting the first spectral data of the engine tail flame of a normal engine, and extracting the ideal spectral characteristics of the target substance based on the first spectral data;

[0008] S2. collecting the second spectral data of the engine tail flame of the engine to be detected, and obtaining the actual spectral characteristics of the target substance based on the second spectral data;

[0009] S3. collecting the actual tail flame temperature and the actual tail flame pressure of the engine tail flame of the engine to be tested, and correcting the actual spectral characteristics based on the actual tail flame temperature and the actual tail flame pressure to obtain the actual spectral correction characteristics;

[0010] S4. Acquire dynamic features for determining whether the engine to be detected is faulty based on the ideal spectral features and the actual spectral correction features;

[0011] S5. Constructing a dynamic threshold range for determining whether the dynamic feature is abnormal;

[0012] S6. Determine whether the dynamic feature is abnormal based on the dynamic threshold range, wherein if the dynamic feature exceeds the dynamic threshold range, then the dynamic feature is abnormal, that is, it is determined that the engine to be detected has a fault; if the dynamic feature is within the dynamic threshold range, then the dynamic feature is not abnormal, that is, it is determined that the engine to be detected has no fault.

[0013] According to one aspect of the present invention, in step S1, in the step of extracting the ideal spectral characteristics of the target substance based on the first spectral data, the target substance is at least one metal element;

[0014] The ideal spectral feature is the spectral peak intensity of the target substance, which is extracted based on the following method:

[0015] A characteristic absorption wavelength in the atomic absorption spectrum of the target substance is obtained, and a corresponding spectrum peak is extracted from the first spectrum data based on the characteristic absorption wavelength to serve as the ideal spectrum feature.

[0016] According to one aspect of the present invention, in step S2, in the step of obtaining the actual spectral feature of the target substance based on the second spectral data, the actual spectral feature is the spectral peak intensity of the target substance, which is extracted based on the following method:

[0017] A characteristic absorption wavelength in the atomic absorption spectrum of the target substance is obtained, and a corresponding spectrum peak is extracted from the second spectrum data based on the characteristic absorption wavelength to serve as the actual spectrum feature.

[0018] According to one aspect of the present invention, in step S3, in the step of correcting the actual spectrum feature based on the actual tail flame temperature and the actual tail flame pressure to obtain an actual spectrum correction feature, the actual spectrum correction feature is expressed as:

[0019] ;

[0020] ;

[0021] in, represents the actual spectral correction characteristics, represents the actual spectral characteristics, represents the correction factor, Indicates the actual tail flame temperature, Indicates the actual tail flame pressure, is the sensitivity coefficient of temperature to spectral intensity, is the sensitivity coefficient of pressure to spectral intensity, is the reference temperature, is the base pressure.

[0022] According to one aspect of the present invention, in step S1, in the step of collecting the reference temperature and reference pressure of the engine tail flame of a normal engine, the reference temperature and the reference pressure are collected based on the nozzle of the normal engine;

[0023] In step S3, in the step of collecting the actual tail flame temperature and the actual tail flame pressure of the engine tail flame of the engine to be tested, the actual tail flame temperature and the actual tail flame pressure are collected based on the nozzle of the engine to be tested.

[0024] According to one aspect of the present invention, in step S4, in the step of acquiring dynamic features for determining whether the engine to be detected is faulty based on the ideal spectral features and the actual spectral correction features, the dynamic features include: spectral feature differences and / or spectral feature ratio differences of the target substance;

[0025] The spectral feature differences are obtained based on the same metal element and are expressed as:

[0026] ;

[0027] in, Represents the actual spectral characteristics; represents the actual spectral correction characteristics;

[0028] The spectral feature ratio difference is obtained based on different metal elements, which is expressed as:

[0029] ;

[0030] in, Represents the ratio of the actual spectral correction characteristics between different metal elements, Represents the ratio of the ideal spectral characteristics between different metal elements. represents the ideal spectral characteristics, the subscript and They represent different metal elements respectively.

[0031] According to one aspect of the present invention, in step S5, the step of constructing a dynamic threshold range for determining whether the dynamic feature is abnormal includes:

[0032] S51. Obtaining the ideal spectral characteristics of a preset time, and calculating the initial mean and initial standard deviation of the ideal spectral characteristics;

[0033] S52. constructing a dynamic mean value about the initial mean value and a dynamic standard deviation about the initial standard deviation based on an exponentially weighted moving average formula;

[0034] S53. Construct the dynamic threshold range based on the dynamic mean and the dynamic standard deviation.

[0035] According to one aspect of the present invention, in step S51, in the step of calculating the initial mean and initial standard deviation of the ideal spectral feature, the initial mean is expressed as:

[0036] ;

[0037] The initial standard deviation is expressed as:

[0038] ;

[0039] in, represents the number of data points of the ideal spectral feature collected within a preset time, represents the ideal spectral characteristics collected, where the subscript The flag bit representing the ideal spectral feature, i.e. An ideal spectral feature;

[0040] In step S52, in the step of respectively constructing a dynamic mean value about the initial mean value and a dynamic standard deviation about the initial standard deviation based on an exponentially weighted moving average formula, the dynamic mean value is expressed as:

[0041] ;

[0042] in, represents the weighting coefficient, and its value ranges from 0.1 to 0.3. represents the ideal spectral characteristics newly generated at present, Represents the standard deviation of the previous moment, and its initial value is the initial mean ;

[0043] The dynamic standard deviation is expressed as:

[0044] ;

[0045] in, represents the weighting coefficient, and its value ranges from 0.1 to 0.3. represents the ideal spectral characteristics newly generated at present, Represents the standard deviation of the previous moment, and its initial value is the initial mean ;

[0046] According to one aspect of the present invention, in step S53, in the step of constructing the dynamic threshold range based on the dynamic mean and the dynamic standard deviation, the dynamic threshold range includes: a first dynamic threshold range for judging whether the spectral feature difference is abnormal and / or a second dynamic threshold range for judging whether the spectral feature ratio difference is abnormal;

[0047] The first dynamic threshold range is expressed as:

[0048] ;

[0049] in, represents the lower limit of the first dynamic threshold range, represents the upper limit of the first dynamic threshold range;

[0050] The second dynamic threshold range is expressed as:

[0051] ;

[0052] in, represents the lower limit of the second dynamic threshold range, Indicates the upper limit of the second dynamic threshold range.

[0053] According to one aspect of the present invention, in step S6, if the dynamic feature exceeds the dynamic threshold range, then the dynamic feature is abnormal, that is, in the step of determining that the engine to be detected has a fault, if the number of times the dynamic feature exceeds the dynamic threshold range exceeds a preset threshold, an alarm is triggered, and the alarm signal is at least one of sound, image, and text;

[0054] The metal element of the target substance is at least one of nickel, chromium and iron.

[0055] According to one solution of the present invention, the present invention realizes high-precision and real-time fault diagnosis by analyzing the spectral characteristics in the engine tail flame and combining the temperature and pressure sensor data correction. It can be used flexibly and accurately for the prediction and timely alarm of engine faults, which can fully reduce the occurrence of accidents.

[0056] According to one scheme of the present invention, the present invention is suitable for measuring the tail flame of a liquid rocket engine, can effectively detect the ultraviolet and visible spectral characteristics of the engine wake, can fully improve the sensitivity and accuracy of detecting metal elements in the tail flame, especially in a complex combustion environment, and ensures accurate monitoring of trace elements.

[0057] According to one scheme of the present invention, the present invention can identify the composition and concentration of the combustion products by analyzing the ultraviolet and visible light characteristic spectral line sequence of the wake. Detecting key metal elements (such as Ni, Cr, Fe, etc.) in the combustion process helps to understand the chemical composition and reaction efficiency of the combustion reaction in real time, thereby optimizing the combustion conditions and ensuring the stable operation of the engine. In addition, by extracting the characteristic spectral lines of these elements, the concentration of each metal element can be quantitatively analyzed. This multi-element monitoring capability effectively improves the overall control of the combustion state and wake composition, and provides strong support for the combustion optimization and wake pollution monitoring of rocket engines.

[0058] According to one solution of the present invention, by analyzing the changes in characteristic spectral lines, the present invention can accurately determine whether abnormal ablation or corrosion occurs in key components such as pipelines and nozzles in liquid rocket engines. Early detection and alarm of possible material ablation and structural damage can be performed, thereby achieving early prevention of faults, extending the life of engine components, and reducing maintenance frequency and costs.

[0059] According to one solution of the present invention, the present invention introduces a dynamic threshold adjustment solution, which combines multi-sensor data such as temperature and pressure to adjust the detection threshold of each metal element feature in real time, ensuring that the system adapts to complex combustion environments and working conditions. In particular, through dynamic adaptive adjustment, the interference of environmental noise and working condition changes on the detection results is effectively reduced, and the stability and robustness of the detection are improved.

[0060] According to a scheme of the present invention, this scheme has data storage and feedback optimization functions, and can dynamically adjust the corresponding detection threshold range for data storage analysis and the introduction of currently acquired data, thereby improving the accuracy and reliability of fault diagnosis and establishing a long-term detection and maintenance mechanism for liquid rocket engines.

[0061] According to a solution of the present invention, based on the introduction of real-time data, a real-time feedback mechanism is provided, which can continuously improve the accuracy and robustness of fault detection.

[0062] According to one solution of the present invention, the present invention can integrate multiple sensor data (such as temperature and pressure) and combine with high-precision ultraviolet and visible spectrum detection technology to achieve accurate monitoring of the engine operating status and timely identify abnormal changes in materials.

[0063] According to one scheme of the present invention, based on ultraviolet spectrum detection and multi-sensor data fusion technology, the present invention detects the ultraviolet and visible light characteristic spectra of the liquid rocket engine wake, analyzes the metal element composition and concentration in the combustion products, thereby accurately judging the ablation of key components and realizing real-time monitoring of the engine health status and early fault warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is a flowchart schematically showing the steps of an engine fault detection method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0065] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments. The embodiments cannot be described one by one here, but the embodiments of the present invention are not therefore limited to the following embodiments.

[0066] like Figure 1 As shown, according to one embodiment of the present invention, an engine fault detection method based on multi-sensor and spectral data fusion of the present invention comprises the following steps:

[0067] S1. Collecting the first spectral data, reference temperature and reference pressure of the engine tail flame of a normal engine, and extracting the ideal spectral characteristics of the target substance based on the first spectral data;

[0068] S2. collecting the second spectral data of the engine tail flame of the engine to be detected, and obtaining the actual spectral characteristics of the target substance based on the second spectral data;

[0069] S3. Collecting the actual tail flame temperature and the actual tail flame pressure of the engine tail flame of the engine to be tested, and correcting the actual spectral characteristics based on the actual tail flame temperature and the actual tail flame pressure to obtain the actual spectral correction characteristics;

[0070] S4. Acquire dynamic features for determining whether the engine to be detected is faulty based on the ideal spectral features and the actual spectral correction features;

[0071] S5. Constructing a dynamic threshold range for determining whether a dynamic feature is abnormal;

[0072] S6. Determine whether the dynamic feature is abnormal based on the dynamic threshold range, wherein if the dynamic feature exceeds the dynamic threshold range, the dynamic feature is abnormal, that is, it is determined that the engine to be detected has a fault; if the dynamic feature is within the dynamic threshold range, the dynamic feature is not abnormal, that is, it is determined that the engine to be detected has no fault.

[0073] According to an embodiment of the present invention, in step S1, in the step of collecting the first spectrum data of the engine tail flame of a normal engine, an ultraviolet-visible spectrum detector is used to collect data of the engine tail flame, wherein the detection spectrum range of the ultraviolet-visible spectrum detector is 300 to 800 nm. Thus, the collected first spectral data includes the intensity information of the tail flame spectrum at different wavelengths, covering the spectral characteristics of various metal elements. In this embodiment, the collected first spectral data can be stored in a corresponding database in real time, and the spectral data measurement time, engine tail flame temperature, engine tail flame pressure and other related information can be stored accordingly during the storage process, so that the subsequent fault diagnosis can be combined with data of multiple dimensions for analysis. In this embodiment, in the process of collecting the first spectral data, the method of collecting once every 0.1s can be adopted to realize the continuous acquisition of the first spectral data from the beginning to the end of the ignition.

[0074] In this embodiment, in step S1, in the step of collecting the reference temperature and reference pressure of the engine tail flame of the normal engine, a temperature sensor and a pressure sensor can be used to continuously collect the corresponding temperature and pressure (or pressure) as the reference temperature and reference pressure; wherein, before ignition, the pressure sensor and the temperature sensor have been installed at the normal engine nozzle (i.e., the engine nozzle outlet), so that the data of the reference temperature and reference pressure can be obtained from the beginning to the end of the ignition. In this embodiment, the collection position is the normal engine nozzle (i.e., the engine nozzle outlet).

[0075] According to an embodiment of the present invention, in step S1, in the step of extracting the ideal spectral characteristics of the target substance based on the first spectral data, the target substance is at least one metal element; in this embodiment, the metal element of the target substance is at least one of nickel (Ni), chromium (Cr), and iron (Fe). In this embodiment, the metal elements of the target substance are nickel (Ni), chromium (Cr), and iron (Fe) to achieve the characteristic analysis of the engine tail flame. Therefore, the ideal spectral characteristic is the spectral peak intensity of the target substance, which is extracted based on the following method:

[0076] The characteristic absorption wavelength in the atomic absorption spectrum of the target substance is obtained, and the corresponding spectrum peak is extracted from the first spectrum data based on the characteristic absorption wavelength as the ideal spectrum feature. Specifically, for the collected first spectrum data, by analyzing the spectrum data within a specific wavelength range, the spectrum peak intensity of the selected nickel (Ni), chromium (Cr), and iron (Fe) is calculated to obtain the corresponding ideal spectrum features, which are:

[0077] Nickel (Ni) spectral peak intensity: Extract the spectral peak at the characteristic absorption wavelength of 341nm to obtain the ideal spectral characteristics of the nickel (Ni) element .

[0078] Chromium (Cr) spectral peak intensity: Extract the spectral peak at the characteristic absorption wavelength of 358nm to obtain the ideal spectral characteristics of the chromium (Cr) element .

[0079] Iron (Fe) spectral peak intensity: Extract the spectral peak at the characteristic absorption wavelength of 248nm to obtain the ideal spectral characteristics of the iron (Fe) element .

[0080] According to one embodiment of the present invention, in step S2, in the step of collecting the second spectral data of the engine tail flame of the engine to be detected, the time for collecting the second spectral data of the engine tail flame of the engine to be detected can be flexibly set, for example, the collection is performed when the engine tail flame is burning for 0.5s or 2s, and then in subsequent calculations, the first spectral data of the corresponding time can be extracted based on the collection time of the second spectral data to perform calculations, which will not be repeated here.

[0081] Furthermore, in step S2, in the step of obtaining the actual spectral feature of the target substance based on the second spectral data, the actual spectral feature is the spectral peak intensity of the target substance, which is extracted based on the following method:

[0082] The characteristic absorption wavelength in the atomic absorption spectrum of the target substance is obtained, and the corresponding spectrum peak is extracted from the second spectrum data based on the characteristic absorption wavelength as the actual spectrum feature. Specifically, for the collected second spectrum data, by analyzing the spectrum data within a specific wavelength range, the spectrum peak intensity of the selected nickel (Ni), chromium (Cr), and iron (Fe) is calculated to obtain the corresponding actual spectrum features, which are:

[0083] Nickel (Ni) spectral peak intensity: Extract the spectral peak at the characteristic absorption wavelength of 341nm to obtain the actual spectral characteristics of the nickel (Ni) element .

[0084] Chromium (Cr) spectral peak intensity: Extract the spectral peak at the characteristic absorption wavelength of 358nm to obtain the actual spectral characteristics of the chromium (Cr) element .

[0085] Iron (Fe) spectral peak intensity: Extract the spectral peak at the characteristic absorption wavelength of 248nm to obtain the actual spectral characteristics of the iron (Fe) element .

[0086] According to an embodiment of the present invention, in step S3, in the step of collecting the actual tail flame temperature and the actual tail flame pressure of the engine tail flame to be detected, a temperature sensor and a pressure sensor are used to collect the corresponding actual tail flame temperature and the actual tail flame pressure respectively. In this embodiment, the nozzle of the engine to be detected (i.e., the outlet of the engine nozzle) is measured by the temperature sensor and the pressure sensor to collect the corresponding temperature and pressure data (which can also be pressure data). In this embodiment, the arrangement position of the temperature sensor and the pressure sensor at the nozzle of the engine to be detected is consistent with the arrangement position when the reference temperature and the reference pressure are collected in step S1. In this embodiment, the time for collecting the actual tail flame temperature and the actual tail flame pressure can be flexibly set, for example, the engine tail flame is selected to be collected when it is 0.5s or 2s burning, wherein, since the collection of the reference temperature and the reference pressure is continuous, in the subsequent calculation, the corresponding reference temperature and the reference pressure can be extracted based on the collection time of the actual tail flame temperature and the actual tail flame pressure, which will not be repeated here.

[0087] Further, in step S3, in the step of correcting the actual spectrum feature based on the actual tail flame temperature and the actual tail flame pressure to obtain the actual spectrum correction feature, the actual spectrum correction feature is expressed as:

[0088] ;

[0089] ;

[0090] in, represents the actual spectral correction characteristics, represents the actual spectral characteristics, represents the correction factor, Indicates the actual tail flame temperature, Indicates the actual tail flame pressure, is the sensitivity coefficient of temperature to spectral intensity, is the sensitivity coefficient of pressure to spectral intensity, is the reference temperature, is the base pressure.

[0091] In this embodiment, the actual spectrum characteristics of the three metal elements (nickel (Ni), chromium (Cr), and iron (Fe)) collected can be corrected based on the above actual spectrum correction characteristic formula. ， ， The calibration is performed separately to obtain the corresponding actual spectral correction characteristics.

[0092] According to an embodiment of the present invention, in step S4, in the step of obtaining dynamic features for determining whether the engine to be detected is faulty based on the ideal spectral features and the actual spectral correction features, the dynamic features include: spectral feature differences and / or spectral feature ratio differences of the target substance; in this embodiment, the spectral feature differences are obtained based on the same metal element, which is expressed as:

[0093] ;

[0094] in, Represents the actual spectral characteristics; represents the actual spectral correction characteristics;

[0095] Therefore, for different metal elements, their spectral characteristics can be expressed as follows:

[0096] Differences in nickel (Ni) spectral characteristics (i.e., differences in peak intensity):

[0097] ;

[0098] Differences in chromium (Cr) spectral characteristics (i.e., differences in peak intensity):

[0099] ;

[0100] Differences in the spectral characteristics of iron (Fe) (i.e., differences in peak intensity):

[0101] ;

[0102] Furthermore, the spectral feature ratio difference based on different metal elements is obtained, which is expressed as:

[0103] ;

[0104] in, Represents the ratio of the actual spectral correction characteristics between different metal elements, Represents the ratio of the ideal spectral characteristics between different metal elements. represents the ideal spectral characteristics, the subscript and Respectively represent different metal elements;

[0105] Therefore, for different metal elements, the differences in their spectral characteristic ratios can be expressed as:

[0106] The difference in the spectral characteristics of nickel (Ni) and chromium (Cr) is (difference in intensity ratio):

[0107] ;

[0108] The difference in the spectral characteristics of nickel (Ni) and iron (Fe) is (difference in intensity ratio):

[0109] ;

[0110] The difference in the spectral characteristics of chromium (Cr) and iron (Fe) is (difference in intensity ratio):

[0111] .

[0112] According to an embodiment of the present invention, in step S5, the step of constructing a dynamic threshold range for determining whether a dynamic feature is abnormal includes:

[0113] S51. Obtain the ideal spectral characteristics of the preset time, and calculate the initial mean and initial standard deviation of the ideal spectral characteristics; in this embodiment, the initial mean is expressed as:

[0114] ;

[0115] The initial standard deviation is expressed as:

[0116] ;

[0117] in, The number of data points representing the desired spectral signature collected within a preset time. represents the ideal spectral characteristics of the collection, where the subscript The flag bit representing the ideal spectral characteristics, i.e. An ideal spectral feature;

[0118] S52. Based on the exponentially weighted moving average formula, a dynamic mean value about the initial mean value and a dynamic standard deviation about the initial standard deviation are respectively constructed; in this embodiment, the dynamic mean value is expressed as:

[0119] ;

[0120] in, represents the weighting coefficient, and its value ranges from 0.1 to 0.3. represents the ideal spectral characteristics newly generated at present, Represents the standard deviation of the previous moment, and its initial value is the initial mean ;

[0121] The dynamic standard deviation is expressed as:

[0122] ;

[0123] in, represents the weighting coefficient, and its value ranges from 0.1 to 0.3. represents the ideal spectral characteristics newly generated at present, Represents the standard deviation of the previous moment, and its initial value is the initial mean ;

[0124] S53. A dynamic threshold range is constructed based on the dynamic mean and the dynamic standard deviation; in this embodiment, the dynamic threshold range includes: a first dynamic threshold range for determining whether the spectral feature difference is abnormal and / or a second dynamic threshold range for determining whether the spectral feature ratio difference is abnormal; wherein the first dynamic threshold range is expressed as:

[0125] ;

[0126] in, represents the lower limit of the first dynamic threshold range, represents the upper limit of the first dynamic threshold range;

[0127] The second dynamic threshold range is expressed as:

[0128] ;

[0129] in, represents the lower limit of the second dynamic threshold range, Indicates the upper limit of the second dynamic threshold range.

[0130] Through the above settings, based on the characteristics of the exponentially weighted moving average formula, it can give more weight to the latest data, thus, and It will dynamically reflect the latest status of the system and can quickly adapt to data changes, thereby setting upper and lower limits more accurately.

[0131] According to an embodiment of the present invention, in step S6, in the step of determining whether the dynamic feature is abnormal based on the dynamic threshold range, the spectral feature difference ( , , ) are judged separately, thus, the spectral characteristic differences of each metal element can be judged separately ( , , ) is abnormal, thereby achieving real-time detection of the working status of the structural components containing the corresponding metal elements. Further, the spectral feature ratio difference ( , , ) are judged separately, thus, the spectral characteristic ratio difference of each metal element can be judged separately ( , , ) is abnormal, thereby achieving real-time detection of the working status of the structural components containing the corresponding metal elements. Therefore, the engine status can be comprehensively judged by combining the spectral feature difference and the spectral feature ratio difference, thereby achieving the accuracy of the overall judgment of the engine status, effectively ensuring the detection accuracy of the present invention.

[0132] Furthermore, in step S6, if the dynamic feature exceeds the dynamic threshold range, then there is an abnormality in the dynamic feature, that is, in the step of determining whether there is a fault in the engine to be detected, if the number of times the dynamic feature exceeds the dynamic threshold range exceeds the preset threshold, then an alarm is triggered, and the alarm signal is at least one of sound, image, and text; in this embodiment, the preset threshold can be set to 3 times, 4 times, etc.

[0133] The present invention also provides an engine fault detection system based on spectrum detection and multi-sensor data engine fault detection method, comprising:

[0134] The database module is used to store the first spectrum data and the second spectrum data collected by the ultraviolet-visible spectrum detector. The first spectrum data includes the intensity information of the spectrum in the tail flame of a normal engine at different wavelengths, so as to cover the spectrum characteristics of multiple metal elements. The second spectrum data includes the intensity information of the spectrum in the tail flame of the engine to be tested at different wavelengths, so as to cover the spectrum characteristics of multiple metal elements.

[0135] The spectral data acquisition module, based on the ultraviolet-visible spectrum detector, obtains the spectral data of the engine exhaust flame in real time.

[0136] The sensor data acquisition module uses temperature sensors and pressure sensors to monitor environmental data such as temperature and pressure when the engine is running, so as to correct the actual spectral characteristics and improve the accuracy of fault detection.

[0137] A data processing module extracts ideal spectral characteristics of a target substance based on the first spectral data, obtains actual spectral characteristics of the target substance based on the second spectral data, corrects the actual spectral characteristics based on the actual tail flame temperature and the actual tail flame pressure to obtain actual spectral correction characteristics, and obtains dynamic characteristics for determining whether an engine to be detected is faulty based on the ideal spectral characteristics and the actual spectral correction characteristics.

[0138] The dynamic threshold adjustment and state discrimination module is used to construct a dynamic threshold range for judging whether the dynamic characteristics are abnormal, and judge whether the dynamic characteristics are abnormal based on the dynamic threshold range. If the dynamic characteristics exceed the dynamic threshold range, the dynamic characteristics are abnormal, that is, it is judged that there is a fault in the engine to be detected; if the dynamic characteristics are within the dynamic threshold range, the dynamic characteristics are not abnormal, that is, it is judged that there is no fault in the engine to be detected.

[0139] The alarm module triggers a fault alarm if there is a fault in the engine to be detected.

[0140] The above contents are merely examples of specific solutions of the present invention. For devices and structures not described in detail therein, it should be understood that they can be implemented by adopting general devices and general methods available in the art.

[0141] The above is only one solution of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An engine fault detection method based on multi-sensor and spectral data fusion, characterized in that: The following steps are involved: S1. Collecting the first spectral data, reference temperature and reference pressure of the engine tail flame of a normal engine, and extracting the ideal spectral characteristics of the target substance based on the first spectral data; S2. collecting the second spectral data of the engine tail flame of the engine to be detected, and obtaining the actual spectral characteristics of the target substance based on the second spectral data; S3. collecting the actual tail flame temperature and the actual tail flame pressure of the engine tail flame of the engine to be tested, and correcting the actual spectral characteristics based on the actual tail flame temperature and the actual tail flame pressure to obtain the actual spectral correction characteristics; S4. Acquire dynamic features for determining whether the engine to be detected is faulty based on the ideal spectral features and the actual spectral correction features; S5. Constructing a dynamic threshold range for determining whether the dynamic feature is abnormal; S6. Determine whether the dynamic feature is abnormal based on the dynamic threshold range, wherein if the dynamic feature exceeds the dynamic threshold range, then the dynamic feature is abnormal, that is, it is determined that the engine to be detected has a fault; if the dynamic feature is within the dynamic threshold range, then the dynamic feature is not abnormal, that is, it is determined that the engine to be detected has no fault.

2. The engine fault detection method based on multi-sensor and spectral data fusion according to claim 1 is characterized in that: In step S1, in the step of extracting the ideal spectral characteristics of the target substance based on the first spectral data, the target substance is at least one metal element; The ideal spectral feature is the spectral peak intensity of the target substance, which is extracted based on the following method: A characteristic absorption wavelength in the atomic absorption spectrum of the target substance is obtained, and a corresponding spectrum peak is extracted from the first spectrum data based on the characteristic absorption wavelength to serve as the ideal spectrum feature.

3. The engine fault detection method based on multi-sensor and spectral data fusion according to claim 2 is characterized in that: In step S2, in the step of acquiring the actual spectral feature of the target substance based on the second spectral data, the actual spectral feature is the spectral peak intensity of the target substance, which is extracted based on the following method: A characteristic absorption wavelength in the atomic absorption spectrum of the target substance is obtained, and a corresponding spectrum peak is extracted from the second spectrum data based on the characteristic absorption wavelength to serve as the actual spectrum feature.

4. The engine fault detection method based on multi-sensor and spectral data fusion according to claim 3 is characterized in that: In step S3, in the step of correcting the actual spectrum feature based on the actual tail flame temperature and the actual tail flame pressure to obtain an actual spectrum correction feature, the actual spectrum correction feature is expressed as: in, represents the actual spectral correction characteristics, represents the actual spectral characteristics, represents the correction factor, Indicates the actual tail flame temperature, Indicates the actual tail flame pressure, is the sensitivity coefficient of temperature to spectral intensity, is the sensitivity coefficient of pressure to spectral intensity, is the reference temperature, is the base pressure.

5. The engine fault detection method based on multi-sensor and spectral data fusion according to claim 4 is characterized in that: In step S1, in the step of collecting the reference temperature and reference pressure of the engine tail flame of a normal engine, the reference temperature and the reference pressure are collected based on the nozzle of the normal engine; In step S3, in the step of collecting the actual tail flame temperature and the actual tail flame pressure of the engine tail flame of the engine to be tested, the actual tail flame temperature and the actual tail flame pressure are collected based on the nozzle of the engine to be tested.

6. The engine fault detection method based on multi-sensor and spectral data fusion according to claim 5 is characterized in that: In step S4, in the step of acquiring dynamic features for judging whether the engine to be detected is faulty based on the ideal spectrum features and the actual spectrum correction features, the dynamic features include: spectrum feature differences and / or spectrum feature ratio differences of the target substance; The spectral feature ratio difference is obtained based on different metal elements, which is expressed as: in, Represents the ratio of the actual spectral correction characteristics between different metal elements, Represents the ratio of the ideal spectral characteristics between different metal elements. represents the ideal spectral characteristics, the subscript and They represent different metal elements respectively.

7. The engine fault detection method based on multi-sensor and spectral data fusion according to claim 6 is characterized in that: In step S5, the step of constructing a dynamic threshold range for determining whether the dynamic feature is abnormal includes: S51. Obtaining the ideal spectral characteristics of a preset time, and calculating the initial mean and initial standard deviation of the ideal spectral characteristics; S52. constructing a dynamic mean value about the initial mean value and a dynamic standard deviation about the initial standard deviation based on an exponentially weighted moving average formula; S53. Construct the dynamic threshold range based on the dynamic mean and the dynamic standard deviation.

8. The engine fault detection method based on multi-sensor and spectral data fusion according to claim 7 is characterized in that: In step S51, in the step of calculating the initial mean and initial standard deviation of the ideal spectral feature, the initial mean is expressed as: The initial standard deviation is expressed as: in, represents the number of data points of the ideal spectral feature collected within a preset time, represents the ideal spectral characteristics collected, where the subscript The flag bit representing the ideal spectral feature, i.e. An ideal spectral feature; In step S52, in the step of respectively constructing a dynamic mean value about the initial mean value and a dynamic standard deviation about the initial standard deviation based on an exponentially weighted moving average formula, the dynamic mean value is expressed as: in, represents the weighting coefficient, and its value ranges from 0.1 to 0.

3. represents the ideal spectral characteristics newly generated at present, Represents the standard deviation of the previous moment, and its initial value is the initial mean ; The dynamic standard deviation is expressed as: in, represents the weighting coefficient, and its value ranges from 0.1 to 0.

3. represents the ideal spectral characteristics newly generated at present, Represents the standard deviation of the previous moment, and its initial value is the initial mean .

9. The engine fault detection method based on multi-sensor and spectral data fusion according to claim 8, characterized in that: In step S53, in the step of constructing the dynamic threshold range based on the dynamic mean and the dynamic standard deviation, the dynamic threshold range includes: a first dynamic threshold range for judging whether the spectral feature difference is abnormal and / or a second dynamic threshold range for judging whether the spectral feature ratio difference is abnormal; The first dynamic threshold range is expressed as: in, represents the lower limit of the first dynamic threshold range, represents the upper limit of the first dynamic threshold range; The second dynamic threshold range is expressed as: in, represents the lower limit of the second dynamic threshold range, Indicates the upper limit of the second dynamic threshold range.

10. The engine fault detection method based on multi-sensor and spectral data fusion according to claim 9, characterized in that: In step S6, if the dynamic feature exceeds the dynamic threshold range, the dynamic feature is abnormal, that is, in the step of determining whether the engine to be detected has a fault, if the number of times the dynamic feature exceeds the dynamic threshold range exceeds a preset threshold, an alarm is triggered, and the alarm signal is at least one of sound, image, and text; The metal element of the target substance is at least one of nickel, chromium and iron.

Images