Special vehicle fault diagnosis method and system based on wide near infrared spectrum analysis

Abstract

The invention relates to the technical field of mode recognition, in particular to a special vehicle fault diagnosis method and system based on wide near infrared spectrum analysis. The invention discloses a special vehicle fault diagnosis method based on wide near infrared spectrum analysis. The special vehicle fault diagnosis method comprises the following steps: S1, collecting ultraviolet-visible-near infrared wide spectrum data of an oil sample; performing data layer fusion and spectrum preprocessing on the wide spectrum data to obtain final spectrum data; s2, on the basis of the final spectral data, screening characteristic wavelengths for oil state analysis by using a continuous projection algorithm; s3, determining an initial characteristic wavelength based on the characteristic wavelength; and determining an enhanced characteristic wavelength set and a weakened characteristic wavelength set according to the initial characteristic wavelength. According to the method, ultraviolet-visible-near infrared wide spectrum and time sequence dynamic characteristics are fused, key wavelengths are screened through a continuous projection algorithm, clustering and change point detection are combined, and oil state evolution mode recognition and fault early-stage accurate early warning are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of pattern recognition, in particular to a special vehicle fault diagnosis method and system based on wide near-infrared spectrum analysis. BACKGROUND

[0002] Wide near-infrared spectrum analysis technology realizes rapid and non-destructive qualitative and quantitative analysis of oil chemical composition and physical state by detecting the absorption and scattering characteristics of substances in the ultraviolet to near-infrared band. This technology has gradually been applied to the field of oil monitoring of special vehicles, and has shown significant advantages in improving detection efficiency and realizing equipment portability by establishing a quantitative relationship between spectral characteristics and key parameters of oil, providing a new technical approach for equipment condition monitoring.

[0003] However, existing diagnosis methods based on wide near-infrared spectrum are mostly limited to static index analysis, lacking deep mining of the dynamic evolution law of oil state. Traditional methods are difficult to effectively capture the time sequence correlation and collaborative change pattern between multiple characteristic parameters, and are not sensitive to the weak characteristic response of early faults, resulting in the risk of state misjudgment under complex working conditions. In particular, when dealing with the dynamic coupling relationship between spectral characteristic wavelengths, the existing technology has not established an effective pattern recognition and evolution trend prediction mechanism.

[0004] Therefore, there is an urgent need to provide a spectrum analysis method that can deeply analyze the dynamic characteristics of oil state, and improve the accuracy and early warning ability of special vehicle fault diagnosis by constructing a multi-parameter collaborative evolution model. SUMMARY

[0005] In order to overcome the shortcomings of insufficient dynamic pattern recognition ability, the present application provides a special vehicle fault diagnosis method and system based on wide near-infrared spectrum analysis.

[0006] The technical implementation scheme of the present application is a special vehicle fault diagnosis method based on wide near-infrared spectrum analysis, comprising the following steps: S1: Collecting ultraviolet-visible-near-infrared wide spectrum data of the oil sample; performing data layer fusion and spectrum pretreatment on the wide spectrum data to obtain final spectrum data; S2: Based on the final spectrum data, using a continuous projection algorithm to screen characteristic wavelengths for oil state analysis; S3: Determining an initial characteristic wavelength based on the characteristic wavelengths; determining a set of strengthened characteristic wavelengths and a set of weakened characteristic wavelengths according to the initial characteristic wavelength; S4: Determining a strengthening period according to the set of strengthened characteristic wavelengths; determining a weakening period according to the set of weakened characteristic wavelengths, and obtaining a mixed period according to the strengthening period and the weakening period; S5: constructing an oil state evolution mode library based on the initial characteristic wavelength and the mixing period and determining an oil state evolution mode; determining a transient change rate and a statistically significant inflection point based on the oil state evolution mode, and generating a special vehicle fault diagnosis conclusion according to the transient change rate and the statistically significant inflection point.

[0007] Preferably, the ultraviolet-visible-near-infrared wide spectrum data of the collected oil sample is acquired; the wide spectrum data is subjected to data layer fusion and spectrum pretreatment to obtain final spectrum data, including: Acquiring ultraviolet-visible-near-infrared continuous wide spectrum data of the oil sample in the wavelength range of 200 nm to 1700 nm; Using an interpolation algorithm to unify the ultraviolet-visible continuous wide spectrum data and the near-infrared continuous wide spectrum data to the same wavelength coordinate point; Using a transition function to perform smoothing processing on the spectral overlap region of the ultraviolet-visible continuous wide spectrum data and the near-infrared continuous wide spectrum data to obtain complete spectrum data; Using standard normal transformation to perform scattering correction on the complete spectrum data; Using Savitzky-Golay smoothing filtering to denoise the complete spectrum data subjected to scattering correction; Performing normalization processing on the complete spectrum data after denoising to obtain final spectrum data.

[0008] Preferably, based on the final spectrum data, a continuous projection algorithm is used to screen characteristic wavelengths for oil state analysis, including: Step 1, setting a characteristic wavelength number parameter N to be screened by the continuous projection algorithm, randomly selecting an initial wavelength in the full wavelength range of the final spectrum data, and adding the initial wavelength to a characteristic wavelength set; Step 2, taking the absorbance vector corresponding to the initial wavelength as an initial projection vector; Step 3, calculating the projection of the absorbance vector corresponding to each remaining wavelength on the current projection vector among the remaining wavelengths not selected into the characteristic wavelength set; Step 4, comparing the lengths of all the projections, and selecting the wavelength corresponding to the projection with the maximum length as a new round of characteristic wavelength, and adding the new round of characteristic wavelength to the characteristic wavelength set; Step 5, updating the projection vector, which is the orthogonal component of the absorbance vector corresponding to the new round of characteristic wavelength and the current projection vector; Step 6, if the number of wavelengths in the characteristic wavelength set does not reach the characteristic wavelength number parameter N, returning to step 3 for iteration; Step 7, if the number of wavelengths in the characteristic wavelength set reaches the characteristic wavelength number parameter N, terminate the iteration and output the characteristic wavelength set as the screening result.

[0009] Preferably, the initial characteristic wavelength is determined based on the characteristic wavelength, comprising: For each characteristic wavelength in the characteristic wavelength set, record the absorbance value corresponding to each characteristic wavelength in each oil detection in chronological order to form the absorbance time series data of each characteristic wavelength; The earliest characteristic wavelength in the characteristic wavelength set that deviates from the preset absorbance deviation threshold and is most relevant to the preset failure mode is taken as the initial characteristic wavelength.

[0010] Preferably, the initial characteristic wavelength is determined based on the characteristic wavelength, comprising: If the absorbance time series data of the initial characteristic wavelength and the absorbance time series data of another characteristic wavelength in the characteristic wavelength set show a synergistic growth trend, the other characteristic wavelength is taken as the reinforcing characteristic wavelength of the initial characteristic wavelength, and the reinforcing characteristic wavelength set corresponding to the initial characteristic wavelength is constructed based on all characteristic wavelengths satisfying the condition; The synergistic growth trend refers to that the Pearson correlation coefficient value is greater than the preset positive correlation threshold, specifically, the Pearson correlation coefficient between the absorbance time series data of the initial characteristic wavelength and the absorbance time series data of all other characteristic wavelengths in the characteristic wavelength set is calculated. If the absorbance time series data of the initial characteristic wavelength and the absorbance time series data of another characteristic wavelength in the characteristic wavelength set show an antagonistic trend, the other characteristic wavelength is taken as the weakening characteristic wavelength of the initial characteristic wavelength, and the weakening characteristic wavelength set corresponding to the initial characteristic wavelength is constructed based on all characteristic wavelengths satisfying the condition; The antagonistic trend refers to that the Pearson correlation coefficient value is less than the preset negative correlation threshold, specifically, the Pearson correlation coefficient between the absorbance time series data of the initial characteristic wavelength and the absorbance time series data of all other characteristic wavelengths in the characteristic wavelength set is calculated.

[0011] Preferably, the reinforcing period is determined based on the reinforcing characteristic wavelength set, comprising: For each candidate characteristic wavelength in the reinforcing characteristic wavelength set, calculate the cross-correlation function between the absorbance time series data of the initial characteristic wavelength and the absorbance time series data of the candidate characteristic wavelength; In the cross-correlation function sequence, find the maximum value of the cross-correlation function and the time delay corresponding to the maximum value; If the maximum value of the cross-correlation function is greater than the preset positive correlation threshold value, and the time delay is greater than zero, the current candidate characteristic wavelength is determined as an effective reinforcement characteristic wavelength, and the time delay is recorded as a reinforcement period of the effective reinforcement characteristic wavelength relative to the initial characteristic wavelength.

[0012] Preferably, the determination of the weakening period according to the set of weakening characteristic wavelengths, and the obtaining of the mixed period according to the reinforcement period and the weakening period, comprise: For each candidate characteristic wavelength in the set of weakening characteristic wavelengths, a cross-correlation function between the absorbance time series data of the initial characteristic wavelength and the absorbance time series data of the candidate characteristic wavelength is calculated; In the sequence of cross-correlation functions, the minimum value of the cross-correlation function and the time delay corresponding to the minimum value are found; If the minimum value of the cross-correlation function is less than the preset negative correlation threshold value, and the time delay is greater than zero, the current candidate characteristic wavelength is determined as an effective weakening characteristic wavelength, and the time delay is recorded as a weakening period of the effective weakening characteristic wavelength relative to the initial characteristic wavelength; A set composed of the reinforcement periods corresponding to all effective reinforcement characteristic wavelengths and the weakening periods corresponding to all effective weakening characteristic wavelengths is defined as a mixed period corresponding to the initial characteristic wavelength.

[0013] Preferably, the construction of the oil condition evolution mode library and the determination of the oil condition evolution mode based on the initial characteristic wavelength and the mixed period comprise: According to all initial characteristic wavelengths and corresponding mixed periods, an oil condition evolution mode library is constructed using a clustering algorithm; Each clustering cluster in the oil condition evolution mode library represents a typical oil condition evolution mode, and each clustering cluster has a mapping relationship with at least one preset fault type; For real-time detection data, the initial characteristic wavelength and the mixed period of the real-time detection data are extracted to generate a real-time feature vector; The real-time feature vector is matched with the oil condition evolution mode library in terms of similarity to determine an oil condition evolution mode to which the real-time feature vector belongs, and a preliminary fault type inference is obtained.

[0014] Preferably, the determination of the instantaneous change rate and the statistically significant inflection point based on the oil condition evolution mode, and the generation of a special vehicle fault diagnosis conclusion according to the instantaneous change rate and the statistically significant inflection point, comprise: Based on the determined oil condition evolution mode, a sliding window linear fitting method is used to process time series data of key oil condition parameters to extract an instantaneous change rate of the key oil condition parameters; Based on the determined oil state evolution mode, a change point detection algorithm is applied to identify statistically significant inflection points of the key oil state parameters from the time series data of the key oil state parameters. If the instantaneous change rate of the key oil state parameter continues to exceed the historical baseline rate corresponding to the oil state evolution mode, a failure acceleration early warning is triggered. If a new statistically significant inflection point appears in the time series sequence of the key oil state parameter, a failure mechanism mutation alarm is triggered. The final fault diagnosis method of the key components of the special vehicle is output by integrating the oil state evolution mode, the instantaneous change rate state and the inflection point detection result.

[0015] The special vehicle fault diagnosis system based on wide near-infrared spectrum analysis comprises: A wide near-infrared spectrum acquisition module is used to acquire ultraviolet to near-infrared wide spectrum data of the oil sample, complete spectrum data fusion and preprocessing to obtain final spectrum data; A characteristic wavelength screening module is used to screen a characteristic wavelength set representing the oil state from the final spectrum data based on a continuous projection algorithm; A dynamic correlation analysis module is used to identify initial characteristic wavelengths through time series data, construct a set of strengthening and weakening characteristic wavelengths and calculate a mixing period; A fault diagnosis decision module is used to construct a state evolution mode library using the mixing period, and generate a fault diagnosis conclusion by combining the change rate and the inflection point detection.

[0016] Beneficial effects: The wide ultraviolet-visible-near-infrared spectrum data of the oil sample is collected, and data layer fusion and spectrum preprocessing are completed to obtain high-quality standard spectrum data. The continuous projection algorithm is used to automatically screen a characteristic wavelength set representing the oil state, effectively extracting key spectrum information and eliminating wavelength multiple collinearity interference. The initial characteristic wavelengths are identified based on time series data, and the sets of strengthening and weakening characteristic wavelengths and the mixing period are constructed through Pearson correlation coefficient and cross-correlation function analysis, thereby deeply quantifying the dynamic correlation of oil state evolution and the fault conduction time sequence rule. A clustering algorithm is used to construct an oil state evolution mode library, realizing intelligent clustering identification and accurate mapping of typical fault modes. Further, the instantaneous change rate of the key state parameter is extracted by a sliding window linear fitting method, and a change point detection algorithm is applied to identify statistically significant inflection points, realizing early warning of failure acceleration trend and mechanism mutation. Finally, while improving the accuracy and stability of the fault diagnosis of the special vehicle, the intelligent perception and prediction ability of the special vehicle for weak fault characteristics and multi-stage evolution modes under complex working conditions is significantly enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The flowchart of the special vehicle fault diagnosis method based on wide near-infrared spectrum analysis of the present application is shown in the figure. Figure 2 This is a structural diagram of the special vehicle fault diagnosis system based on wide near-infrared spectral analysis according to the present invention. Detailed Implementation

[0018] The present invention will be further described below with reference to specific embodiments. The illustrative embodiments and descriptions herein are used to explain the present invention, but are not intended to limit the present invention.

[0019] Example 1: A fault diagnosis method for special vehicles based on wide near-infrared spectral analysis, such as... Figure 1 As shown, it includes the following steps: S1: Collect UV-Vis-NIR broadband spectral data of the oil sample; perform data layer fusion and spectral preprocessing on the broadband spectral data to obtain the final spectral data, including: The data layer fusion includes wavelength alignment and overlap region optimization; the spectral preprocessing includes scattering correction, denoising, and normalization. Collect continuous broadband spectral data of oil samples in the ultraviolet-visible-near-infrared wavelength range of 200 nm to 1700 nm; Interpolation algorithms were used to unify ultraviolet-visible continuous broadband data and near-infrared continuous broadband data to the same wavelength coordinate points; The spectral overlap region between the ultraviolet-visible continuous broadband spectral data and the near-infrared continuous broadband spectral data is smoothed using a transition function to obtain complete spectral data; The complete spectral data were scattering corrected using a standard normal transform. The complete spectral data after scattering correction was denoised using Savitzky-Golay smoothing filtering; The denoised complete spectral data is normalized to obtain the final spectral data.

[0020] It should be noted that in the monitoring of oil conditions in special vehicles, traditional spectral analysis methods often suffer from insufficient feature extraction stability due to inadequate fusion of multi-source spectral data and noise interference. To address this issue, this scheme first uses a UV-Vis-NIR composite optical sensing system to acquire the full-band transmission spectrum of oil samples from 200–1700 nm. Since the spectra obtained from different waveband sensors exhibit wavelength scaling differences and response shifts, interpolation algorithms (such as linear interpolation or cubic spline interpolation) are needed to unify the wavelength coordinates of the UV-Vis and NIR spectra. Furthermore, a transition function is used for weighted fusion in the overlapping spectral region to construct a continuous broad spectral curve. The transition function is a linear weighting function, i.e., in the overlapping region […]. , Within this range, the weight of the ultraviolet-visible spectrum decreases linearly from 1 to 0, while the weight of the near-infrared spectrum increases linearly from 0 to 1.

[0021] To further eliminate the scattering effect caused by suspended particles in oil and random noise of the instrument, a standard normal transformation is used to correct the spectral baseline, and then combined with Savitzky-Golay filtering to smooth high-frequency fluctuations. Finally, through normalization (such as maximum normalization method) processing to unify the signal amplitude, the standardized spectral data for subsequent feature learning is formed. This preprocessing procedure significantly improves the spectral quality, and lays a data foundation for subsequent realization of high-robustness oil condition analysis and fault diagnosis.

[0022] S2: Based on the final spectral data, using a continuous projection algorithm to screen feature wavelengths for oil condition analysis, comprising: Step 1, setting the number of feature wavelengths N to be screened by the continuous projection algorithm, randomly selecting an initial wavelength in the full wavelength range of the final spectral data, and adding the initial wavelength to the feature wavelength set; Step 2, taking the absorbance vector corresponding to the initial wavelength as the initial projection vector; Here, the 'absorbance vector' refers to a vector composed of the absorbance values measured at all wavelength points of an oil sample, and the dimension is equal to the total number of wavelength points.

[0023] Step 3, among the remaining wavelengths not selected into the feature wavelength set, calculating the projection of the absorbance vector corresponding to each remaining wavelength on the current projection vector; Step 4, comparing the lengths of all the projections, and selecting the wavelength corresponding to the projection with the longest length as the feature wavelength of the new round, and adding the feature wavelength of the new round to the feature wavelength set; Step 5, updating the projection vector, which is the orthogonal component of the absorbance vector corresponding to the feature wavelength of the new round and the current projection vector; Step 6, if the number of wavelengths in the feature wavelength set does not reach the feature wavelength number parameter N, return to step 3 for iteration; Step 7, if the number of wavelengths in the feature wavelength set reaches the feature wavelength number parameter N, terminate the iteration and output the feature wavelength set as the screening result.

[0024] It should be noted that in the field of spectral analysis, the traditional oil detection method often relies on expert experience for characteristic wavelength selection, and there is multiple collinearity interference between wavelengths, which leads to insufficient reliability of the analysis process. This step realizes the autonomous optimization and screening of characteristic wavelengths through the continuous projection algorithm. Specifically, the number of characteristic wavelengths N needs to be determined comprehensively according to the resolution of the spectrometer and the number of characteristic wavelengths to be screened and the complexity of the final diagnostic rule, and is usually 5-20 key wavelengths. When the diagnostic system runs for a period of time and accumulates enough new data, the cross-validation process is periodically re-executed (such as every quarter or every year) to verify or update the optimal N value, ensuring that the selected characteristic wavelength set can adapt to the slow changes in the data. The full wavelength range refers to the complete data interval of the 200-1700nm spectrometer, which covers the characteristic absorption bands of water, oxidation products and wear particle components in the oil.

[0025] During algorithm initialization, an initial wavelength is randomly selected to eliminate human selection bias, and the absorbance vector corresponding to this wavelength is used as the initial projection vector. In each iteration, the projection modulus of the remaining wavelength absorbance vector and the current projection vector is calculated, and the wavelength corresponding to the maximum modulus is selected into the characteristic wavelength set, as it carries the most spectral information that is not represented by the existing characteristics. Subsequently, the projection vector is updated through the Gram-Schmidt orthogonalization method to ensure that subsequent selection is always performed in a space orthogonal to the selected characteristic wavelengths. The iteration is terminated when the size of the characteristic wavelength set reaches the characteristic wavelength number parameter N, at which time the obtained characteristic wavelength set has both maximum information content and minimum redundancy.

[0026] For example, starting from the water feature peak at about 1450nm, an optimized set containing the oxidation peak at about 1700nm and the fuel peak at about 1200nm is selected through six rounds of orthogonal projection iteration. The feature optimization is automatically completed through mathematical mechanism, effectively overcoming the subjectivity and collinearity defects of traditional manual screening, and laying a foundation for subsequent construction of high-precision oil condition evolution pattern library and fault diagnosis system.

[0027] To illustrate the determination method of the key parameter N, this example takes the engine oil of a certain type of special vehicle as an example. 300 sets of oil sample spectral data and water content reference values covering different running times, healthy and abnormal states are collected. The data set is divided into a training set (210 sets) and a validation set (90 sets) according to a ratio of 7:3.

[0028] On the training set, let the characteristic wavelength number parameter N range from 3 to 25. For each N value, execute SPA to select a subset of characteristic wavelengths, and based on this subset, establish a water content partial least squares (PLSR) prediction function on the training set to assist in evaluation. Subsequently, apply the PLSR function to the validation set to calculate the root mean square error (RMSE) of the predicted values, thereby evaluating the prediction performance of the characteristic wavelength subset under different N values.

[0029] The verification set RMSE under different N values is recorded, and the change rule is as follows: when N∈[3,10], the RMSE value decreases significantly and rapidly with the increase of N, indicating that increasing the characteristic wavelength can effectively capture more key information. When N∈[11,15], the RMSE decreases slowly. When N>15, the RMSE no longer decreases or even fluctuates, indicating that continuing to increase the number of wavelengths will introduce redundancy.

[0030] Based on the above rule, on the RMSE-N relationship curve, N=12 is located in the inflection point region of the curve from “rapid decline” to “tend to be flat”. At this point, the RMSE has decreased to a low level (0.076%), while avoiding redundancy. Therefore, in this application scenario, N=12 is selected as the optimal characteristic wavelength number parameter. This method balances the prediction performance and complexity, ensuring that the selected characteristic wavelength set not only guarantees the core information, but also has good robustness and practicality.

[0031] It should be noted that the moisture content PLSR prediction function used to determine the parameter N in this example is a single-index, temporary auxiliary tool, and its purpose is only to optimize the feature selection process through cross-validation. The auxiliary function itself is not the content claimed by the present application, and it is also different from the oil liquid state evolution pattern library finally constructed for fault diagnosis.

[0032] S3-1: determining an initial characteristic wavelength based on the characteristic wavelength, comprising: For each characteristic wavelength in the characteristic wavelength set, record the absorbance value corresponding to each characteristic wavelength in each oil detection in time sequence to form the absorbance time sequence data of each characteristic wavelength; Take the characteristic wavelength in the characteristic wavelength set that first deviates from the preset absorbance deviation threshold and is most related to the preset fault mode as the initial characteristic wavelength.

[0033] It should be noted that, in order to realize early identification and tracing of faults, the present scheme captures the dynamic evolution process of oil state by constructing absorbance time sequence data. The preset absorbance deviation threshold is determined according to the statistical distribution of historical data under the normal running state of the equipment, and is usually taken as ±3 times the standard deviation range of the mean absorbance of healthy samples. The wavelength that first breaks through the threshold and matches the spectral characteristics of the preset fault mode is set as the initial characteristic wavelength, because it not only represents the time sequence starting point of fault occurrence, but also has clear fault indication characteristics. Taking the cooling liquid leakage fault as an example, when the water characteristic peak absorbance at about 1450 nm first exceeds the threshold in the third detection, and has the highest matching degree with the spectral database of the cooling liquid pollution mode, the wavelength is determined as the initial characteristic wavelength, which is used as the reference anchor point for subsequent correlation feature mining and fault evolution analysis.

[0034] S3-2: determining a set of reinforcing characteristic wavelengths and a set of weakening characteristic wavelengths according to the initial characteristic wavelength, comprising: If the absorbance time series data of the initial characteristic wavelength and the absorbance time series data of another characteristic wavelength in the set of characteristic wavelengths show a synergistic growth trend, the other characteristic wavelength is taken as the reinforcing characteristic wavelength of the initial characteristic wavelength, and the set of reinforcing characteristic wavelengths corresponding to the initial characteristic wavelength is constructed based on all characteristic wavelengths satisfying this condition; The synergistic growth trend refers to that the Pearson correlation coefficient value is greater than a preset positive correlation threshold, specifically, the Pearson correlation coefficient between the absorbance time series data of the initial characteristic wavelength and the absorbance time series data of all other characteristic wavelengths in the set of characteristic wavelengths is calculated; If the absorbance time series data of the initial characteristic wavelength and the absorbance time series data of another characteristic wavelength in the set of characteristic wavelengths show an antagonistic trend, the other characteristic wavelength is taken as the weakening characteristic wavelength of the initial characteristic wavelength, and the set of weakening characteristic wavelengths corresponding to the initial characteristic wavelength is constructed based on all characteristic wavelengths satisfying this condition; The antagonistic trend refers to that the Pearson correlation coefficient value is less than a preset negative correlation threshold, specifically, the Pearson correlation coefficient between the absorbance time series data of the initial characteristic wavelength and the absorbance time series data of all other characteristic wavelengths in the set of characteristic wavelengths is calculated.

[0035] It should be noted that in the monitoring of special vehicle oil, the fault evolution is essentially a dynamic process formed by the interaction of multiple chemical components. The limitation of traditional diagnosis method is that only single parameter threshold is concerned, and the conduction mechanism of fault in oil system cannot be revealed. The present scheme aims to establish a complete path map of fault propagation by constructing the set of reinforcing and weakening characteristic wavelengths.

[0036] The Pearson correlation coefficient is selected as the quantitative tool based on its sensitive characteristics to linear relationship, which can accurately capture the synchronicity of absorbance changes at different wavelengths. This method is used to represent two basic interaction modes between oil parameters: when the absorbance of two characteristic wavelengths shows the same change, a synergistic effect is formed, such as the synchronous enhancement of water intrusion and metal corrosion; when the change is in the opposite direction, an antagonistic effect is formed, such as the complementary relationship between additive consumption and pollutant increase.

[0037] The threshold setting adopts a statistical inference method based on historical data. By analyzing the distribution characteristics of the correlation coefficients between the characteristic wavelengths in the normal operation state of the equipment, the statistical significance boundary is determined combined with hypothesis testing. For example: "By calculating the Pearson correlation coefficients between all feature wavelengths in a large amount of historical healthy data, a correlation coefficient distribution is formed, and a certain quantile (such as 97.5% quantile as the positive threshold and 2.5% quantile as the negative threshold) of the distribution is taken as the threshold, and the statistical significance level corresponds to p<0.05. The positive correlation threshold is used to identify the synergistic growth mode, and the negative correlation threshold is used to capture the antagonistic trend. The determination of the two thresholds completely depends on the statistical characteristics of a large amount of historical data, which ensures that normal fluctuations and abnormal correlations can be effectively distinguished.

[0038] This double set analysis mechanism promotes fault diagnosis from a single point to a system level, which can not only identify directly related fault features, but also find indirect influence derived effects, providing a more comprehensive analysis dimension for special vehicle fault warning.

[0039] S4-1: Determine the reinforcement period according to the reinforcement feature wavelength set, including: For each candidate feature wavelength in the reinforcement feature wavelength set, calculate the cross-correlation function between the absorbance time series data of the initial feature wavelength and the absorbance time series data of the candidate feature wavelength; In the cross-correlation function sequence, find the maximum value of the cross-correlation function and the time delay corresponding to the maximum value; If the maximum value of the cross-correlation function is greater than the preset positive correlation threshold, and the time delay is greater than zero, the current candidate feature wavelength is determined as an effective reinforcement feature wavelength, and the time delay is recorded as the reinforcement period of the effective reinforcement feature wavelength relative to the initial feature wavelength.

[0040] It should be noted that in view of the defect that the traditional oil diagnosis method is insufficient to describe the fault time evolution law, the cross-correlation function analysis technology is adopted to establish the dynamic correlation relationship between the feature wavelengths. This method calculates the sliding correlation coefficient of the initial feature wavelength and the candidate wavelength on the absorbance time series data point by point, and constructs a complete cross-correlation function sequence. This step is based on the maximum correlation principle in signal processing theory, and by finding the maximum value and the corresponding time delay in the cross-correlation function sequence, the most significant same direction change mode of the two feature wavelengths is captured. The preset positive / negative correlation threshold is set in the same way as the correlation threshold setting method described in the specific implementation manner S3-2.

[0041] The constraint condition that the maximum value of the cross-correlation function needs to exceed a preset positive correlation threshold is used to ensure the statistical significance of the feature correlation, and the requirement that the time delay is strictly greater than zero is to follow the timing logic of fault conduction, ensuring that the change of the candidate feature is indeed caused by the initial feature. When both conditions are met, the candidate feature wavelength is assigned the identity of an effective reinforcement feature wavelength, and its time delay is used as a reinforcement period for subsequent construction of an oil condition evolution mode library. In engineering practice, the quantitative description of this timing relationship provides a key basis for predictive maintenance, for example, by observing the delay period of metal corrosion features relative to moisture intrusion features, the development process of engine cooling system failures can be accurately calculated, thereby realizing the transition from passive repair to active warning.

[0042] S4-2: Determine the weakening period according to the set of weakening feature wavelengths, and obtain the mixed period according to the reinforcement period and the weakening period, comprising: For each candidate feature wavelength in the set of weakening feature wavelengths, calculate the cross-correlation function between the absorbance time series data of the initial feature wavelength and the absorbance time series data of the candidate feature wavelength; In the cross-correlation function sequence, find the minimum value of the cross-correlation function and the time delay corresponding to the minimum value; If the minimum value of the cross-correlation function is less than a preset negative correlation threshold and the time delay is greater than zero, the current candidate feature wavelength is determined as an effective weakening feature wavelength, and the time delay is recorded as the weakening period of the effective weakening feature wavelength relative to the initial feature wavelength; The set formed by the reinforcement periods corresponding to all effective reinforcement feature wavelengths and the weakening periods corresponding to all effective weakening feature wavelengths is defined as the mixed period corresponding to the initial feature wavelength.

[0043] It should be noted that in view of the current situation that traditional diagnostic methods are insufficient in analyzing negative timing relationships between features, the present scheme uses cross-correlation analysis technology to quantify the antagonistic effect. This method constructs a complete cross-correlation function sequence by calculating the correlation coefficient distribution of the initial feature and the candidate feature within a sliding time window. The identification of the minimum value in the sequence is based on the negative correlation maximization principle, and the corresponding time delay represents the degree of reverse synchronization of the changes of the two feature wavelengths.

[0044] The minimum value is required to be lower than the preset negative correlation threshold to ensure the statistical significance of the antagonistic relationship; at the same time, the time delay is required to be greater than zero to meet the causal timing logic of fault conduction. The determination of the effective weakening characteristic wavelength and the record of its weakening period provide key parameters for understanding the inhibition mechanism caused by faults. Compared with the reinforcement period reflecting synergistic deterioration, the weakening period reveals the compensatory response mechanism within the system. By constructing a mixed period set containing reinforcement and weakening periods, a complete timing description of the fault propagation process is formed, and this dual-mode period integration strategy significantly improves the accuracy of fault evolution trend prediction, providing more timing feature basis for equipment maintenance decision-making.

[0045] Based on the above 300 sets of healthy oil sample data, the following steps are performed to determine the key threshold: Absorbance deviation threshold: for each of the 12 characteristic wavelengths screened by SPA, the mean (μ) and standard deviation (σ) of the absorbance of all healthy samples are calculated. The preset absorbance deviation threshold is set to μ±3σ. For example, for the 1450nm water feature peak, μ=0.15 and σ=0.02 are calculated, and the deviation threshold is [0.09, 0.21].

[0046] Correlation threshold: the Pearson correlation coefficient between all 12 characteristic wavelengths in the healthy samples is calculated, and a total of 66 correlation coefficient values are obtained. Draw a histogram of the distribution, and take the 97.5% quantile (0.78) of the distribution as the preset positive correlation threshold, and take the 2.5% quantile (-0.75) as the preset negative correlation threshold. This setting statistically means that the probability of stronger correlation under healthy state is less than 5%.

[0047] Historical baseline rate: when constructing and applying the oil condition evolution pattern library for diagnosis, for each identified oil condition evolution pattern (such as "cooling fluid leakage pattern"), samples belonging to the pattern and in the "healthy" or "early" stage are extracted from the historical database (for example, the first 50 detection data). Calculate the instantaneous change rate of these samples' key parameters (such as iron content) to form a "baseline rate set". The historical baseline rate warning threshold of this pattern is set to the mean + 2 times the standard deviation of the set.

[0048] In the present application, "lasting more than" is defined as the instantaneous change rate exceeding the historical baseline rate for 3 or more consecutive detection periods.

[0049] S5-1: Constructing an oil condition evolution pattern library based on the initial characteristic wavelengths and mixed periods and determining the oil condition evolution pattern, comprising: According to all initial characteristic wavelengths and corresponding mixed periods, an oil condition evolution pattern library is constructed using clustering algorithm; Each cluster in the oil state evolution pattern library represents a typical oil state evolution pattern, and each cluster is mapped to at least one preset fault type; For real-time detection data, the initial characteristic wavelength and the mixing period of the real-time detection data are extracted to generate a real-time feature vector; The real-time feature vector is matched with the oil state evolution pattern library to determine the oil state evolution pattern to which the real-time feature vector belongs, and a preliminary fault type inference is obtained.

[0050] It should be noted that, in order to realize intelligent diagnosis from oil spectrum data to fault type, the present scheme constructs a complete oil state evolution pattern recognition system. In view of the defects of traditional methods relying on expert experience and lacking of systematic classification system, the automatic identification of faults is realized through the pattern library construction and matching mechanism.

[0051] The specific process of constructing the pattern library is as follows: first, a feature matrix is constructed based on historical data: the initial characteristic wavelength value of each sample is taken as the first dimension, and the period values in the mixing period are arranged in a fixed order as the subsequent dimensions, to form a standardized feature vector. Then, the K-means clustering algorithm is used to determine the optimal number of clusters k through the elbow rule to realize automatic grouping of the feature matrix. Each cluster represents a fault pattern with similar evolution law, such as a cluster with 1720nm water peak as the main feature and containing specific period combination corresponding to the cooling liquid leakage mode. Finally, through statistical analysis, the fault type with the highest frequency in each cluster is taken as the preset label of the cluster, and the construction of the fault knowledge graph is completed.

[0052] In the real-time diagnosis execution process, for real-time detection data, the initial characteristic wavelength value is first extracted, and the period values in the mixing period are arranged in the same order to generate a feature vector consistent with the dimension of the historical data. Then, the Euclidean distance formula is used to calculate the distance between the real-time feature vector and the cluster centers in the pattern library, and the most matched state evolution pattern is determined by the minimum distance criterion. For example, when a real-time sample with initial characteristic wavelength of 1720nm and mixing period of [8, 12, 5] is detected, the distance between it and each cluster center is calculated, and it is found that the distance with the "cooling liquid leakage" pattern cluster center is the smallest, so the fault type is inferred.

[0053] To specifically illustrate the construction of the pattern library, the present example is based on 500 groups of historical data containing explicit fault labels (covering cooling liquid leakage, fuel dilution, normal wear). The "initial characteristic wavelength value" and "mixing period" sequence are extracted for each group of data to form a feature vector.

[0054] Taking the typical failure mode of "coolant leakage accompanied by initial rust" as an example: initial characteristic wavelength: among the 12 wavelengths screened out by SPA, 1450 nm (water hydroxyl absorption peak) is identified as the initial characteristic wavelength of this mode. Mixing period: through cross-correlation analysis, the following effective correlation wavelengths and periods are found: reinforcing characteristic wavelength: 500 nm (iron element absorption peak), with a reinforcing period of 5 detection periods relative to the initial characteristic wavelength. Weakening characteristic wavelength: 1150 nm (absorption peak of a certain antioxidant), with a weakening period of 8 detection periods relative to the initial characteristic wavelength. Feature vector construction: the feature vector of this mode is defined as [initial wavelength value, reinforcing period 1, weakening period 1,...]. In this example, a representative feature vector is: [0.35, 5, 8]. Where: 0.35 is the normalized absorbance value of the initial characteristic wavelength (1450 nm). 5 is the reinforcing period of the reinforcing characteristic wavelength (500 nm). 8 is the weakening period of the weakening characteristic wavelength (1150 nm). Use the K-means clustering algorithm to cluster the feature vectors of all 500 samples. The elbow rule is used to determine the optimal number of clusters k = 6. The center of one of the clusters is calculated as [0.32, 5.2, 7.8], and 85% of the samples in this cluster are labeled as "coolant leakage", so the cluster is mapped to the "coolant leakage accompanied by initial rust" failure type, and its cluster center [0.32, 5.2, 7.8] is stored in the pattern library.

[0055] Real-time diagnosis example: In a real-time detection, the extracted feature vector is =[0.34, 5, 8]. The Euclidean distance between it and the "coolant leakage" cluster center [0.32, 5.2, 7.8] in the pattern library is 0.28, and the distance between it and the cluster centers of other failure modes is greater than 1.5. Therefore, it is determined that the real-time sample belongs to the "coolant leakage accompanied by initial rust" mode.

[0056] This pattern recognition-based diagnosis mechanism realizes automatic determination of failure types by quantifying feature similarity, ensuring the repeatability of the diagnosis process and significantly improving the diagnosis efficiency, providing a reliable intelligent diagnosis solution for special vehicle oil systems.

[0057] S5-2: determining the instantaneous change rate and statistically significant inflection point based on the oil state evolution pattern, and generating a special vehicle fault diagnosis conclusion according to the instantaneous change rate and statistically significant inflection point, including: Based on the determined oil state evolution pattern, the time series data of the key oil state parameters is processed using the sliding window linear fitting method to extract the instantaneous change rate of the key oil state parameters; The key oil condition parameter is based on the oil condition evolution mode, and the characteristic wavelength absorbance time series data or the oil condition indicators (such as water content, iron element concentration) calculated therefrom.

[0058] Based on the determined oil condition evolution mode, a change point detection algorithm is applied to identify the statistically significant inflection point of the key oil condition parameter from the time series data of the key oil condition parameter. If the instantaneous change rate of the key oil condition parameter continuously exceeds the historical reference rate corresponding to the oil condition evolution mode, a failure acceleration warning is triggered. If a new statistically significant inflection point appears in the time series sequence of the key oil condition parameter, a failure mechanism mutation alarm is triggered. The final fault diagnosis method of the key component of the special vehicle is output by integrating the oil condition evolution mode, the instantaneous change rate state and the inflection point detection result.

[0059] It should be noted that in the intelligent diagnosis of oil condition, the traditional method often only confirms the fault after it occurs due to the lack of quantitative tracking of the dynamic evolution process of the fault, and cannot realize early warning. This step introduces two dynamic indicators, instantaneous change rate and statistically significant inflection point, to capture the key process from quantitative change to qualitative change of the fault, thereby realizing predictive diagnosis.

[0060] The instantaneous change rate is used to quantify the speed of fault development. The specific operation is: based on the determined oil condition evolution mode (such as "coolant leakage mode"), a sliding window linear fitting method is used to process the time series data of the key parameter (such as iron content). This method slides a fixed length data window on the time axis, performs linear regression on the data points in the window, and takes the slope of the obtained fitting straight line as the instantaneous change rate at the center time of the window. This rate value reflects the average change speed of the parameter per unit time.

[0061] The statistically significant inflection point is used to identify the stage transition of fault development. The specific operation is: a change point detection algorithm (such as a method based on likelihood ratio test) is applied to scan the time series data of the same key parameter, and when the algorithm identifies the position where the statistical characteristics (such as mean, variance) of the data sequence have a significant mutation, it is marked as a statistically significant inflection point. This inflection point marks the entry of the fault into a new and more serious stage.

[0062] The triggering logic of the warning and alarm is completely based on the comparison with the historical reference: The triggering logic of the failure acceleration warning is as follows: first, calculate the instantaneous change rate of the key parameter. When the rate continuously exceeds the historical reference rate corresponding to the oil condition evolution mode, it indicates that the wear is accelerating abnormally, and the warning is triggered immediately.

[0063] It should be noted that the failure acceleration early warning mechanism relies on establishing a quantitative and comparative change rate benchmark for each "oil state evolution pattern". The method for determining the historical benchmark rate is as follows: define the benchmark data subset: from the "oil state evolution pattern library" that has been constructed, for a certain specific pattern (for example, "cooling liquid leakage pattern"), all sample data historically judged as this pattern but in its early stage of failure development are extracted and backtracked. These early stage samples constitute the "in-pattern health / early benchmark data set" for calculating the benchmark. Calculate the instantaneous change rate: for each sample in this benchmark data set, according to its position in the time series, the instantaneous change rate of its key oil state parameter (for example, iron content) is calculated using the sliding window linear fitting method. In this way, a "sample distribution of instantaneous change rate values" representing the normal evolution speed of the pattern is obtained. Set the statistical early warning threshold: statistical analysis is performed on all the instantaneous change rate values calculated above, and the mean and standard deviation are calculated. The historical benchmark rate early warning threshold is finally set to mean + 2 x standard deviation. This threshold statistically means that if a new sample has a change rate exceeding this value, the probability that it does not belong to the early normal evolution speed distribution of the pattern is as high as 95%, thus triggering an early warning. In addition, the "continuous exceedance" is defined in the present application as: in the last M consecutive detection periods (for example, M = 3), the calculated instantaneous change rate of the key parameter is higher than the historical benchmark rate early warning threshold of the corresponding pattern.

[0064] Failure mechanism mutation alarm: when a new statistically significant inflection point is identified in the time series of a key parameter (such as viscosity), it indicates that the physical and chemical state of the oil has undergone a qualitative change, thus triggering a mechanism mutation alarm.

[0065] The final diagnosis method is a fusion decision of the above-mentioned multi-source information. For example: Pattern recognition: it is determined that the current state is "cooling liquid leakage pattern".

[0066] Dynamic analysis: the "continuous exceedance of iron content instantaneous change rate over the historical benchmark rate of the pattern" is detected, triggering a failure acceleration early warning; at the same time, a new statistically significant inflection point is detected in the "oil viscosity" data, triggering a failure mechanism mutation alarm.

[0067] Comprehensive diagnosis: "combining" the basic pattern of "cooling liquid leakage", the abnormal wear trend indicated by the failure acceleration early warning, and the lubrication performance change signal indicated by the failure mechanism mutation alarm, the final output of the accurate diagnosis conclusion is: "there is abnormal wear in the engine caused by cooling liquid leakage, and the lubrication performance has begun to deteriorate, immediate repair is recommended".

[0068] Example 2: on the basis of example 1, a special vehicle failure diagnosis system based on wide near-infrared spectroscopy analysis, such asFigure 2 As shown, comprising: A wide near-infrared spectrum acquisition module is configured to acquire oil sample ultraviolet to near-infrared wide spectrum data, complete spectrum data fusion and preprocessing to obtain final spectrum data; A characteristic wavelength screening module is configured to screen a characteristic wavelength set representing oil liquid state from the final spectrum data based on a continuous projection algorithm; A dynamic correlation analysis module is configured to identify initial characteristic wavelengths through time series data, construct a set of strengthened and weakened characteristic wavelengths and calculate a mixing period; A fault diagnosis decision module is configured to construct a state evolution mode library using the mixing period, and generate a fault diagnosis conclusion in combination with a change rate and an inflection point detection.

[0069] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for diagnosing a fault of a special vehicle based on wide near infrared spectrum analysis, characterized in that, The method comprises the following steps: S1: collecting ultraviolet-visible-near infrared wide spectrum data of the oil sample; performing data layer fusion and spectrum pretreatment on the wide spectrum data to obtain final spectrum data; S2: screening characteristic wavelengths for oil liquid state analysis based on the final spectrum data by using a continuous projection algorithm; S3: determining an initial characteristic wavelength based on the characteristic wavelengths; determining a set of strengthened characteristic wavelengths and a set of weakened characteristic wavelengths according to the initial characteristic wavelength; S4: determining a strengthened period according to the set of strengthened characteristic wavelengths; determining a weakened period according to the set of weakened characteristic wavelengths, and obtaining a mixed period according to the strengthened period and the weakened period; S5: constructing an oil liquid state evolution mode library and determining an oil liquid state evolution mode based on the initial characteristic wavelength and the mixed period; determining an instantaneous change rate and a statistically significant inflection point based on the oil liquid state evolution mode, and generating a special vehicle fault diagnosis conclusion according to the instantaneous change rate and the statistically significant inflection point.

2. The method for diagnosing a fault of a special vehicle based on wide near infrared spectrum analysis according to claim 1, characterized in that, The method comprises the following steps: S1: collecting ultraviolet-visible-near infrared wide spectrum data of the oil sample; performing data layer fusion and spectrum pretreatment on the wide spectrum data to obtain final spectrum data, which comprises the following steps: collecting ultraviolet-visible-near infrared continuous wide spectrum data of the oil sample in a wavelength range of 200 nm to 1700 nm; unifying the ultraviolet-visible continuous wide spectrum data and the near infrared continuous wide spectrum data to the same wavelength coordinate point by using an interpolation algorithm; performing smoothing processing on the spectral overlap region of the ultraviolet-visible continuous wide spectrum data and the near infrared continuous wide spectrum data by using a transition function to obtain complete spectrum data; performing scattering correction on the complete spectrum data by using a standard normal transformation; performing denoising on the complete spectrum data subjected to the scattering correction by using a Savitzky-Golay smoothing filter; 3. The method for diagnosing a fault of a special vehicle based on wide near infrared spectrum analysis according to claim 1, characterized in that, performing normalization processing on the complete spectrum data subjected to the denoising to obtain final spectrum data. The method comprises the following steps: Step 1: setting a characteristic wavelength number parameter N to be screened by the continuous projection algorithm, randomly selecting an initial wavelength in the full wavelength range of the final spectrum data, and adding the initial wavelength to a characteristic wavelength set; Step 2: taking the absorbance vector corresponding to the initial wavelength as an initial projection vector; Step 3: calculating the projection of the absorbance vector corresponding to each remaining wavelength on the current projection vector in the remaining wavelengths that are not selected into the characteristic wavelength set; Step 4: comparing the lengths of all the projections, selecting the wavelength corresponding to the projection with the largest length as a new round of characteristic wavelength, and adding the new round of characteristic wavelength to the characteristic wavelength set; Step 5: updating the projection vector, wherein the updated projection vector is the orthogonal component of the absorbance vector corresponding to the new round of characteristic wavelength and the current projection vector; Step 6: if the number of wavelengths in the characteristic wavelength set does not reach the characteristic wavelength number parameter N, returning to Step 3 for iteration; Step 7: if the number of wavelengths in the characteristic wavelength set reaches the characteristic wavelength number parameter N, terminating the iteration and outputting the characteristic wavelength set as a screening result.

4. The method for diagnosing a fault of a special vehicle based on wide near infrared spectrum analysis according to claim 1, characterized in that, The determining the initial characteristic wavelength based on the characteristic wavelength includes: For each characteristic wavelength in the characteristic wavelength set, record the absorbance value corresponding to each characteristic wavelength in each oil detection in time sequence to form the absorbance time sequence data of each characteristic wavelength; Take the characteristic wavelength in the characteristic wavelength set that first appears the absorbance value deviating from the preset absorbance deviation threshold and is most relevant to the preset failure mode as the initial characteristic wavelength.

5. The method for diagnosing a fault of a special vehicle based on wide near infrared spectrum analysis according to claim 1, characterized in that, The determining the strengthened characteristic wavelength set and the weakened characteristic wavelength set according to the initial characteristic wavelength includes: If the absorbance time sequence data of the initial characteristic wavelength and the absorbance time sequence data of another characteristic wavelength in the characteristic wavelength set present a synergistic growth trend, take the another characteristic wavelength as the strengthened characteristic wavelength of the initial characteristic wavelength, and construct the strengthened characteristic wavelength set corresponding to the initial characteristic wavelength based on all the characteristic wavelengths satisfying the condition; The synergistic growth trend refers to that the Pearson correlation coefficient value is greater than a preset positive correlation threshold, and specifically, the Pearson correlation coefficient between the absorbance time sequence data of the initial characteristic wavelength and the absorbance time sequence data of all other characteristic wavelengths in the characteristic wavelength set is calculated; If the absorbance time sequence data of the initial characteristic wavelength and the absorbance time sequence data of another characteristic wavelength in the characteristic wavelength set present an antagonistic trend, take the another characteristic wavelength as the weakened characteristic wavelength of the initial characteristic wavelength, and construct the weakened characteristic wavelength set corresponding to the initial characteristic wavelength based on all the characteristic wavelengths satisfying the condition; The antagonistic trend refers to that the Pearson correlation coefficient value is less than a preset negative correlation threshold, and specifically, the Pearson correlation coefficient between the absorbance time sequence data of the initial characteristic wavelength and the absorbance time sequence data of all other characteristic wavelengths in the characteristic wavelength set is calculated.

6. The method for diagnosing a fault of a special vehicle based on wide near infrared spectrum analysis according to claim 1, characterized in that, The determining the strengthened period according to the strengthened characteristic wavelength set includes: For each candidate characteristic wavelength in the strengthened characteristic wavelength set, calculate the cross-correlation function between the absorbance time sequence data of the initial characteristic wavelength and the absorbance time sequence data of the candidate characteristic wavelength; In the cross-correlation function sequence, find the maximum value of the cross-correlation function and the time delay corresponding to the maximum value; If the maximum value of the cross-correlation function is greater than a preset positive correlation threshold and the time delay is greater than zero, determine the current candidate characteristic wavelength as an effective strengthened characteristic wavelength, and record the time delay as the strengthened period of the effective strengthened characteristic wavelength relative to the initial characteristic wavelength.

7. The method for diagnosing a fault of a special vehicle based on wide near infrared spectrum analysis according to claim 1, characterized in that, The determining the weakened period according to the weakened characteristic wavelength set, and obtaining the mixed period according to the strengthened period and the weakened period includes: For each candidate characteristic wavelength in the weakened characteristic wavelength set, calculate the cross-correlation function between the absorbance time sequence data of the initial characteristic wavelength and the absorbance time sequence data of the candidate characteristic wavelength; In the cross-correlation function sequence, find the minimum value of the cross-correlation function and the time delay corresponding to the minimum value; If the minimum value of the cross-correlation function is less than the preset negative correlation threshold, and the time delay is greater than zero, the current candidate characteristic wavelength is determined as an effective weakening characteristic wavelength, and the time delay is recorded as a weakening period of the effective weakening characteristic wavelength relative to the initial characteristic wavelength; A set composed of all the strengthening periods corresponding to the effective strengthening characteristic wavelengths and all the weakening periods corresponding to the effective weakening characteristic wavelengths is defined as a mixed period corresponding to the initial characteristic wavelength.

8. The method for diagnosing a fault of a special vehicle based on wide near infrared spectrum analysis according to claim 1, characterized in that, The method comprises the following steps of: constructing an oil state evolution mode library based on the initial characteristic wavelength and the mixed period, and determining an oil state evolution mode; constructing an oil state evolution mode library based on all the initial characteristic wavelengths and the corresponding mixed periods by using a clustering algorithm; Each clustering cluster in the oil state evolution mode library represents a typical oil state evolution mode, and each clustering cluster has a mapping relationship with at least one preset fault type; For real-time detection data, the initial characteristic wavelength and the mixed period of the real-time detection data are extracted to generate a real-time feature vector; 9. The method for diagnosing a fault of a special vehicle based on wide near infrared spectrum analysis according to claim 1, characterized in that, The real-time feature vector is matched with the oil state evolution mode library to determine the oil state evolution mode to which the real-time feature vector belongs, and a preliminary fault type inference is obtained. The method comprises the following steps of: based on the determined oil state evolution mode, using a sliding window linear fitting method to process time series data of a key oil state parameter, and extracting an instantaneous change rate of the key oil state parameter; based on the determined oil state evolution mode, applying a change point detection algorithm to identify a statistically significant inflection point of the key oil state parameter from the time series data of the key oil state parameter; if the instantaneous change rate of the key oil state parameter continuously exceeds a historical reference rate of the corresponding oil state evolution mode, a fault acceleration warning is triggered; if a new statistically significant inflection point appears in the time series sequence of the key oil state parameter, a fault mechanism mutation alarm is triggered; 10. A system for diagnosing faults in a special vehicle based on wide near infrared spectroscopy analysis, for implementing the method for diagnosing faults in a special vehicle based on wide near infrared spectroscopy analysis according to any one of claims 1 to 9, characterized in that, comprehensive oil state evolution mode, instantaneous change rate state and inflection point detection result, output the final fault diagnosis method of the key component of the special vehicle. The method comprises the following steps of: a wide near-infrared spectrum acquisition module is used to acquire oil sample ultraviolet to near-infrared wide spectrum data, complete spectrum data fusion and pretreatment to obtain final spectrum data; a characteristic wavelength screening module is used to screen a characteristic wavelength set representing oil state from the final spectrum data based on a continuous projection algorithm; a dynamic correlation analysis module is used to identify an initial characteristic wavelength through time series data, construct a strengthening and weakening characteristic wavelength set and calculate a mixed period; a fault diagnosis decision module is used to construct a state evolution mode library by using the mixed period, and generate a fault diagnosis conclusion in combination with a change rate and an inflection point detection.

Images