A parameter analysis method and system for automobile suspension spring

Abstract

The present invention relates to the technical field of spring anomaly detection, and specifically to a parameter analysis method and system for an automobile suspension spring. The present invention first monitors suspension springs under multiple working conditions to obtain a first data set; constructs a spring multidimensional data fusion model to analyze the first data set, and generates spring multidimensional associated data by correlating and identifying the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring; constructs an anomaly identification model to analyze the spring multidimensional associated data to generate a first anomaly coefficient; adjusts a spring multidimensional data threshold by analyzing the road condition, constructs a parameter adjustment model to analyze the spring multidimensional associated data and the first data set to generate a second anomaly coefficient; and generates a spring comprehensive anomaly coefficient based on the first anomaly coefficient and the second anomaly coefficient.

Description

Technical Field

[0001] The invention relates to the technical field of spring anomaly detection, and in particular to a parameter analysis method and system for an automobile suspension spring. Background Art

[0002] In the modern automobile industry, the suspension system, as a key component, has a significant impact on the stability, comfort and safety of the vehicle. The suspension spring is one of the core components of the suspension system, which plays the role of buffering and maintaining the balance of the vehicle body. With the complexity of the driving environment and the improvement of vehicle performance requirements, the traditional spring detection method has gradually exposed some shortcomings.

[0003] At present, the existing suspension spring detection technology on the market mainly relies on single parameter monitoring, such as stress and displacement. However, in actual applications, spring performance is affected by many factors, such as load, vibration frequency, temperature and humidity. Single parameter detection is not adaptable enough in complex environments, resulting in deviations in spring detection, making it difficult to accurately judge the actual health status of the spring, and unable to effectively deal with the impact of different vehicle models, different road conditions and environmental changes on spring performance, resulting in limited detection results.

[0004] Therefore, a parameter analysis method and system for automobile suspension springs are proposed. Summary of the invention

[0005] The object of the present invention is to provide a parameter analysis method and system for automobile suspension springs. The present invention first monitors suspension springs under multiple working conditions to obtain a first data set; constructs a spring multidimensional data fusion model to analyze the first data set, and generates spring multidimensional associated data by correlating and identifying the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring; constructs an abnormality recognition model to analyze the spring multidimensional associated data to generate a first abnormality coefficient; constructs a parameter adjustment model to analyze the spring multidimensional associated data and the first data set, and generates a second abnormality coefficient by performing abnormality detection on the spring multidimensional data threshold adjusted under the road condition; and generates a spring comprehensive abnormality coefficient based on the first abnormality coefficient and the second abnormality coefficient.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A parameter analysis method for an automobile suspension spring, comprising:

[0008] Monitoring the suspension spring under multiple working conditions to obtain a first data set; the first data set includes dynamic stress data of the spring, spring compression data, vibration frequency data, spring temperature data, load data, road conditions, ambient temperature and ambient humidity;

[0009] Constructing a spring multi-dimensional data fusion model to analyze the first data set, and generating spring multi-dimensional correlation data by correlating and identifying the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring;

[0010] Constructing an abnormality recognition model to analyze the multi-dimensional correlation data of the spring and generate a first abnormality coefficient;

[0011] By analyzing the road condition, adjusting the spring multidimensional data threshold, constructing a parameter adjustment model to analyze the spring multidimensional correlation data and the first data set, and generating a second abnormal coefficient;

[0012] The spring comprehensive abnormal coefficient is generated based on the first abnormal coefficient and the second abnormal coefficient; the specific calculation formula of the spring comprehensive abnormal coefficient is:

[0013] ;

[0014] in, is the comprehensive abnormal coefficient of the spring, is the first abnormal coefficient weight, is the first abnormal coefficient, is the second abnormal coefficient weight, is the second abnormal coefficient;

[0015] According to the comprehensive abnormal coefficient of the spring, an early warning is given to the automobile suspension spring.

[0016] Preferably, the multidimensional data fusion model includes a first data processing layer, a feature extraction layer and a multidimensional correlation analysis layer;

[0017] The first data processing layer obtains first standard data by cleaning and normalizing the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring;

[0018] The feature extraction layer obtains a first feature set by extracting features from the first standard data; the first feature set includes stress fluctuation features, compression curve features, vibration period features and temperature change rate features;

[0019] The multidimensional correlation analysis layer obtains spring multidimensional correlation data by performing multidimensional data fusion on the stress fluctuation features, compression curve features, vibration period features and temperature change rate features in the first feature set.

[0020] Preferably, the specific process of multidimensional data fusion is:

[0021] The first feature set is represented as a feature matrix, and the feature matrix is ​​decomposed to obtain a potential feature corresponding to each feature; and a feature association between features is generated by analyzing a change trend of the features;

[0022] Comprehensively analyzing all the features in the first feature set, obtaining feature associations among all the features, and generating multi-dimensional association data of the spring;

[0023] The spring multi-dimensional association data includes feature association data and a first feature set.

[0024] Preferably, the anomaly identification model includes a second data processing layer and a first anomaly analysis layer;

[0025] The second data processing layer obtains second standard data by cleaning and normalizing the spring multi-dimensional correlation data;

[0026] The first abnormality analysis layer identifies potential abnormalities by analyzing the second standard data to obtain a first abnormality coefficient;

[0027] The specific calculation formula of the first abnormal coefficient is:

[0028] ;

[0029] in, is the first abnormal coefficient, is the time window length for data collection, is the number of features in the first feature set, For the The first feature weight, For Moment feature data, For the The mean of feature data, For the The standard deviation of the characteristic data, is the number of feature-related data, For the The first feature associated data weight, For Moment The correlation abnormality of feature data.

[0030] Preferably, the parameter adjustment model includes a working condition identification layer, a parameter adjustment layer, a real-time data analysis layer and a second abnormality identification layer;

[0031] The working condition identification layer identifies the load data, road conditions, ambient temperature and ambient humidity, and obtains the current working condition type through a clustering algorithm;

[0032] The parameter adjustment layer adjusts the monitoring data threshold by analyzing the working condition type, load data, ambient temperature and ambient humidity;

[0033] The real-time data analysis layer obtains second abnormal data by comparing the first data set acquired in real time with the adjusted threshold value;

[0034] The second abnormality identification layer obtains a second abnormality coefficient by analyzing the second abnormality data.

[0035] Preferably, the specific calculation formula of the second abnormal coefficient is:

[0036] ;

[0037] in, is the second abnormal coefficient, is the global adjustment coefficient, is the number of features in the first feature set, For the The second feature weight, For the feature data, For the The threshold value after feature data adjustment, is the number of feature-related data, For the The second feature associated data weight, is the maximum value function, For the The associated outliers of feature data, After adjusting the threshold A feature is associated with a data value.

[0038] A parameter analysis system for automobile suspension springs, comprising:

[0039] A data acquisition unit is used to monitor the suspension spring under multiple working conditions to obtain a first data set; the first data set includes dynamic stress data of the spring, spring compression data, vibration frequency data, spring temperature data, load data, road conditions, ambient temperature and ambient humidity;

[0040] A model building unit, used for building a spring multi-dimensional data fusion model to analyze the first data set, and generating spring multi-dimensional correlation data by correlating and identifying the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring;

[0041] A first abnormal coefficient generating unit, used for constructing an abnormal recognition model to analyze the multi-dimensional correlation data of the spring to generate a first abnormal coefficient;

[0042] A second abnormal coefficient generating unit is used to analyze the road condition, adjust the spring multi-dimensional data threshold, build a parameter adjustment model to analyze the spring multi-dimensional correlation data and the first data set, and generate a second abnormal coefficient;

[0043] A comprehensive abnormal coefficient generating unit is used to generate a spring comprehensive abnormal coefficient based on the first abnormal coefficient and the second abnormal coefficient; the specific calculation formula of the spring comprehensive abnormal coefficient is:

[0044] ;

[0045] in, is the comprehensive abnormal coefficient of the spring, is the first abnormal coefficient weight, is the first abnormal coefficient, is the second abnormal coefficient weight, is the second abnormal coefficient;

[0046] According to the comprehensive abnormal coefficient of the spring, an early warning is given to the automobile suspension spring.

[0047] Preferably, the multidimensional data fusion model includes a first data processing layer, a feature extraction layer and a multidimensional correlation analysis layer;

[0048] The first data processing layer obtains first standard data by cleaning and normalizing the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring;

[0049] The feature extraction layer obtains a first feature set by extracting features from the first standard data; the first feature set includes stress fluctuation features, compression curve features, vibration period features and temperature change rate features;

[0050] The multidimensional correlation analysis layer obtains spring multidimensional correlation data by performing multidimensional data fusion on the stress fluctuation features, compression curve features, vibration period features and temperature change rate features in the first feature set.

[0051] Preferably, the anomaly identification model includes a second data processing layer and a first anomaly analysis layer;

[0052] The second data processing layer obtains second standard data by cleaning and normalizing the spring multi-dimensional correlation data;

[0053] The first abnormality analysis layer identifies potential abnormal situations by analyzing the second standard data to obtain a first abnormality coefficient.

[0054] Preferably, the parameter adjustment model includes a working condition identification layer, a parameter adjustment layer, a real-time data analysis layer and a second abnormality identification layer;

[0055] The working condition identification layer identifies the load data, road conditions, ambient temperature and ambient humidity, and obtains the current working condition type through a clustering algorithm;

[0056] The parameter adjustment layer adjusts the monitoring data threshold by analyzing the working condition type, load data, ambient temperature and ambient humidity;

[0057] The real-time data analysis layer obtains second abnormal data by comparing the first data set acquired in real time with the adjusted threshold value;

[0058] The second abnormality identification layer obtains a second abnormality coefficient by analyzing the second abnormality data.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] 1. The present invention constructs a multidimensional data fusion model to preprocess and extract features of the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring in the first data set to obtain a first feature set; the first feature set includes stress fluctuation features, compression curve features, vibration period features and temperature change rate features; by performing multidimensional data fusion on the stress fluctuation features, compression curve features, vibration period features and temperature change rate features in the first feature set, multidimensional correlation data of the spring is obtained, thereby reducing the error caused by single parameter fluctuations and timely discovering abnormalities of the automobile suspension spring during operation.

[0061] 2. The present invention constructs an abnormality recognition model to clean and normalize the multi-dimensional correlation data of the spring to obtain the second standard data; by analyzing the second standard data, potential abnormal conditions are identified to obtain the first abnormality coefficient; the first abnormality coefficient calculation formula reflects the correlation between features through weight distribution, so that the model can realize comprehensive analysis of multi-dimensional data while considering feature fluctuations, thereby improving the sensitivity and accuracy of abnormality detection of automobile suspension springs.

[0062] 3. The present invention constructs a parameter adjustment model to identify the load data, road conditions, ambient temperature and ambient humidity, and analyzes the data through a clustering algorithm to obtain the current working condition type; obtains a second abnormality coefficient by comparing the first data set acquired in real time with the adjusted threshold; and generates a spring comprehensive abnormality coefficient based on the first abnormality coefficient and the second abnormality coefficient; based on the spring comprehensive abnormality coefficient, an efficient early warning of suspension spring abnormalities is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A schematic flow chart of a parameter analysis method for an automobile suspension spring according to the present invention;

[0064] Figure 2 It is a structural schematic diagram of a parameter analysis system for an automobile suspension spring according to the present invention;

[0065] Figure 3 It is a structural schematic diagram of the multidimensional data fusion model of the present invention;

[0066] Figure 4 This is an interface diagram of a parameter analysis system for an automobile suspension spring provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0067] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0068] The present invention provides a parameter analysis method for an automobile suspension spring. The method is applied to a parameter analysis system for an automobile suspension spring. The flowchart of the specific method and system is shown in FIG. Figure 1 and Figure 2 .

[0069] Embodiment 1

[0070] As an embodiment of the present invention, refer to Figure 1 S10 in the figure is applied to a data acquisition unit of a parameter analysis system of an automobile suspension spring, and the data acquisition unit is used to monitor the suspension spring under multiple working conditions to obtain a first data set; the first data set includes dynamic stress data of the spring, spring compression data, vibration frequency data, spring temperature data, load data, road conditions, ambient temperature and ambient humidity.

[0071] The present invention installs force sensors, displacement sensors, acceleration sensors and temperature sensors at key parts of suspension springs to monitor the stress, compression, vibration frequency and working temperature of the springs respectively; by real-time monitoring of the dynamic stress, compression, vibration frequency, temperature, load, road conditions, ambient temperature and humidity of automobile suspension springs under multiple working conditions and forming a first data set, the key parameters of the suspension springs are fully covered. This data collection provides a basis for subsequent abnormality detection and adaptive analysis, enabling the system to identify performance changes of the springs under different working conditions and detect potential abnormalities in a timely manner.

[0072] Reference Figure 1 S20 in the figure is applied to a model building unit of a robot equipment health status assessment system, and the model building unit is used to build a spring multi-dimensional data fusion model to analyze the first data set, and generate spring multi-dimensional correlation data by associating and identifying the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring.

[0073] Further, the multidimensional data fusion model includes a first data processing layer, a feature extraction layer and a multidimensional correlation analysis layer;

[0074] The first data processing layer obtains first standard data by cleaning and normalizing the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring;

[0075] The feature extraction layer obtains a first feature set by extracting features from the first standard data; the first feature set includes stress fluctuation features, compression curve features, vibration period features and temperature change rate features;

[0076] The stress fluctuation feature identifies the stress fluctuation frequency and amplitude by analyzing the stress changes of the spring under different load and vibration conditions, thereby reflecting the dynamic stress bearing capacity of the spring under different working conditions; the compression curve feature tracks the curve of the spring compression changing with time and load, extracts the change mode and peak value of the compression, and analyzes the elastic deformation and recovery performance of the spring; the vibration period feature is used to obtain the vibration frequency and period of the spring under the action of external force, so as to evaluate the stability of the spring and its response at different frequencies; the temperature change rate feature is used to record the rate of change of the spring temperature and identify the potential impact of temperature on the performance of the spring, especially the temperature rise under high-frequency vibration or heavy load conditions;

[0077] The multidimensional correlation analysis layer obtains spring multidimensional correlation data by performing multidimensional data fusion on the stress fluctuation features, compression curve features, vibration period features and temperature change rate features in the first feature set.

[0078] In this embodiment, the dynamic stress, compression, vibration frequency, temperature and load data of the spring are processed by a multi-dimensional data fusion model to extract key features. By analyzing the key features, the performance of the automobile suspension spring under different working conditions is reflected, such as bearing capacity, elastic recovery, stability and the impact of temperature changes on performance, see Table 1.

[0079] Table 1 analyzes the changes in key spring characteristics and their impact on spring performance under different working conditions. It reflects the working performance of springs under various road conditions and environmental conditions. By analyzing these characteristic changes, the status of automobile suspension springs in actual use can be more accurately evaluated.

[0080] Furthermore, the specific process of the multi-dimensional data fusion is as follows:

[0081] The first feature set is represented as a feature matrix, and the feature matrix is ​​decomposed to obtain a potential feature corresponding to each feature; and a feature association between features is generated by analyzing a change trend of the features;

[0082] Comprehensively analyzing all the features in the first feature set, obtaining feature associations among all the features, and generating multi-dimensional association data of the spring;

[0083] The spring multi-dimensional association data includes feature association data and a first feature set.

[0084] In this embodiment, by comprehensively analyzing all the features in the first feature set, the correlation data between the features is obtained, and the multi-dimensional correlation data of the spring is generated. This process effectively reveals the internal connection between the features, improves the accuracy and reliability of spring performance evaluation, and provides theoretical support for the health status monitoring of automobile suspension springs under different working conditions.

[0085] Table 1 Multidimensional data association feature table

[0086]

[0087] Reference Figure 1 S30 in which S30 is applied to a first abnormal coefficient generating unit of a robot equipment health status assessment system, wherein the first abnormal coefficient generating unit is used to construct an abnormality recognition model to analyze the multi-dimensional correlation data of the spring to generate a first abnormal coefficient;

[0088] Further, the anomaly identification model includes a second data processing layer and a first anomaly analysis layer;

[0089] The second data processing layer obtains second standard data by cleaning and normalizing the spring multi-dimensional correlation data;

[0090] The first abnormality analysis layer identifies potential abnormalities by analyzing the second standard data to obtain a first abnormality coefficient;

[0091] The specific calculation formula of the first abnormal coefficient is:

[0092] ;

[0093] in, is the first abnormal coefficient, is the time window length of data collection, in seconds, is the number of features in the first feature set, For the The first feature weight, For Moment feature data, For the The mean of feature data, For the The standard deviation of the characteristic data, is the number of feature-related data, For the The first feature associated data weight, For Moment The correlation abnormality of feature data.

[0094] In this embodiment, the present invention identifies potential abnormalities of automobile suspension springs under different working conditions by establishing an abnormality recognition model, combining the second data processing layer and the first abnormality analysis layer. Standard data is obtained by cleaning and normalizing the multi-dimensional correlation data of the springs, and then abnormal analysis is performed to calculate the first abnormality coefficient. This method can accurately reflect the abnormal performance of the springs under different conditions, and improves the accuracy and real-time performance of fault diagnosis.

[0095] Reference Figure 1 S40 in which S40 is applied to a second abnormal coefficient generating unit of a robot equipment health status assessment system, wherein the second abnormal coefficient generating unit is used to analyze the road surface condition, adjust the spring multidimensional data threshold, construct a parameter adjustment model to analyze the spring multidimensional correlation data and the first data set, and generate a second abnormal coefficient;

[0096] Furthermore, the parameter adjustment model includes a working condition identification layer, a parameter adjustment layer, a real-time data analysis layer and a second abnormality identification layer;

[0097] The working condition identification layer identifies the load data, road conditions, ambient temperature and ambient humidity, and obtains the current working condition type through a clustering algorithm;

[0098] The parameter adjustment layer adjusts the monitoring data threshold by analyzing the working condition type, load data, ambient temperature and ambient humidity;

[0099] The real-time data analysis layer obtains second abnormal data by comparing the first data set acquired in real time with the adjusted threshold value;

[0100] The second abnormality identification layer obtains a second abnormality coefficient by analyzing the second abnormality data.

[0101] In this embodiment, the parameter adjustment model includes working condition identification, parameter adjustment, real-time data analysis and abnormality identification layers. By comprehensively analyzing the multi-dimensional correlation data of the spring and the first data set, and adjusting the data threshold according to the load, road conditions and environmental changes under different working conditions, it can accurately identify abnormal conditions and generate a second abnormality coefficient. The flexibility and accuracy of abnormality detection are enhanced, providing reliable support for real-time monitoring and fault warning of equipment.

[0102] The specific calculation formula of the second abnormal coefficient is:

[0103] ;

[0104] in, is the second abnormal coefficient, is the global adjustment coefficient, which is used to adjust the overall sensitivity of the abnormal coefficient. is the number of features in the first feature set, For the The second feature weight, For the feature data, For the The threshold value after feature data adjustment, is the number of feature-related data, For the The second feature associated data weight, is the maximum value function, For the The associated outliers of feature data, After adjusting the threshold A feature is associated with a data value.

[0105] In this embodiment, the present invention comprehensively considers the number of features, weights and abnormality of associated data in the first feature set through the calculation formula of the second abnormal coefficient, thereby generating an accurate abnormality identification result. By comparing the feature data with the adjusted threshold value and calculating the abnormal value of the associated feature, the abnormal operation state of the spring under multiple working conditions can be accurately reflected.

[0106] Reference Figure 1 S50 in the embodiment is applied to a comprehensive abnormality coefficient generating unit of a robot equipment health status assessment system, wherein the comprehensive abnormality coefficient generating unit is used to generate a spring comprehensive abnormality coefficient based on the first abnormality coefficient and the second abnormality coefficient; the specific calculation formula of the spring comprehensive abnormality coefficient is:

[0107] ;

[0108] in, is the comprehensive abnormal coefficient of the spring, is the first abnormal coefficient weight, is the first abnormal coefficient, is the second abnormal coefficient weight, is the second abnormal coefficient;

[0109] Based on different working conditions, the dynamic stress, compression, vibration frequency, temperature and other parameters of the automobile suspension spring are studied to obtain the first abnormal coefficient and the second abnormal coefficient. Through the training of historical data, the weights of the first abnormal coefficient and the second abnormal coefficient are obtained, and then the comprehensive abnormal coefficient of the spring is generated through comprehensive calculation. Specifically referring to the data shown in Table 2, by comparing the spring performance under different working conditions, potential abnormalities can be identified more accurately;

[0110] Table 2 Spring abnormal coefficient table

[0111]

[0112] Based on the abnormal coefficient data shown in Table 2, the spring comprehensive abnormal coefficient under different working conditions is calculated by weighted summation. Specifically, the first abnormal coefficient and the second abnormal coefficient are used to generate the final spring comprehensive abnormal coefficient after weight adjustment, thereby providing a quantitative basis for the health status assessment of the suspension system.

[0113] Through the analysis of historical data of automobile suspension springs, it is determined that the normal range of the spring comprehensive abnormal coefficient is between 0.1 and 0.3. When the spring comprehensive abnormal coefficient exceeds 0.5, it indicates that the spring may have a significant abnormality. This range is set based on a large amount of experimental data and the performance of the spring under normal working conditions.

[0114] The present invention realizes accurate evaluation of the health status of automobile suspension springs under various working conditions through the comprehensive application of multi-dimensional data fusion model, abnormal identification model and parameter adjustment model. In the present invention, the weights are obtained by training the historical automobile suspension spring data through the machine learning model. First, by arranging force sensors, displacement sensors, acceleration sensors and temperature sensors at the key parts of the spring, key data such as dynamic stress, compression, vibration frequency and temperature under multiple working conditions are collected in real time to form a first data set, providing a complete data basis for subsequent analysis. Through data cleaning, normalization processing and feature extraction, multi-dimensional spring correlation data containing features such as stress fluctuation, compression curve, vibration cycle and temperature change rate are generated. In the abnormal identification stage, the abnormal identification model is constructed through the second data processing layer and the first abnormal analysis layer, the first abnormal coefficient is calculated, the potential abnormal situation is identified, and the accurate identification of the abnormal situation is ensured. In addition, the parameter adjustment model adaptively adjusts the threshold according to changes in road surface, load, environment, etc., and generates the second abnormal coefficient through real-time data analysis and the second abnormal identification layer, thereby improving the flexibility and accuracy of abnormal detection. Finally, the comprehensive abnormal coefficient generation unit calculates the comprehensive abnormal coefficient of the spring based on the weights of the first abnormal coefficient and the second abnormal coefficient, providing a comprehensive and real-time early warning capability for the suspension spring.

[0115] Embodiment 2

[0116] As one of the key components of the vehicle suspension system, automobile suspension springs directly affect the stability and driving safety of the vehicle under different road conditions. Suspension springs not only bear the task of supporting the weight of the vehicle body, but also play an important role in absorbing road vibrations and ensuring passenger comfort. However, the performance of suspension springs will gradually decay over time, and may even fail due to wear or fatigue, which will affect the stability and safety of vehicle driving. Therefore, how to effectively monitor and analyze the status of suspension springs has become an important research direction in automotive safety technology.

[0117] Based on the importance of suspension springs, the present invention proposes a parameter analysis system for automobile suspension springs. Through real-time monitoring of multiple parameters and generation of comprehensive abnormal coefficients, the health status of suspension springs can be evaluated and fault warnings can be given, providing a scientific basis for vehicle driving safety and suspension maintenance. Figure 4 .

[0118] A parameter analysis system for automobile suspension springs, comprising:

[0119] The data acquisition unit is used to monitor the suspension spring under multiple working conditions to obtain a first data set; the first data set includes dynamic stress data of the spring, spring compression data, vibration frequency data, spring temperature data, load data, road conditions, ambient temperature and ambient humidity; Figure 4Click the data acquisition button corresponding to the data acquisition unit in the data acquisition unit to obtain the data in the first data set. The data acquisition progress will be displayed in the progress bar below the output box. When the acquisition is completed, the data box will output "Data acquisition completed!";

[0120] A model building unit is used to build a spring multi-dimensional data fusion model to analyze the first data set, and generate spring multi-dimensional correlation data by correlating and identifying the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring; Figure 4 Click the model building button corresponding to the model building unit in the output box to build the spring multidimensional data fusion model and normalize the data in the first data set. The model building progress will be displayed in the progress bar below the output box. When the building is completed, the data box will output "Model building completed!";

[0121] The first abnormal coefficient generating unit is used to construct an abnormal recognition model to analyze the multi-dimensional correlation data of the spring to generate a first abnormal coefficient; Figure 4 The button for generating the first abnormal coefficient corresponding to the first abnormal coefficient generating unit outputs the value of the first abnormal coefficient;

[0122] The second abnormal coefficient generating unit is used to analyze the road condition, adjust the spring multidimensional data threshold, build a parameter adjustment model to analyze the spring multidimensional correlation data and the first data set, and generate a second abnormal coefficient; Figure 4 The second abnormal coefficient generating unit in the outputs the second abnormal coefficient value by a button corresponding to the second abnormal coefficient generating unit;

[0123] A comprehensive abnormal coefficient generating unit is used to generate a spring comprehensive abnormal coefficient based on the first abnormal coefficient and the second abnormal coefficient; Figure 4 The button corresponding to the spring comprehensive abnormal coefficient generation unit outputs the spring comprehensive abnormal coefficient value;

[0124] The specific calculation formula of the spring comprehensive abnormal coefficient is:

[0125] ;

[0126] in, is the comprehensive abnormal coefficient of the spring, is the first abnormal coefficient weight, is the first abnormal coefficient, is the second abnormal coefficient weight, is the second abnormal coefficient;

[0127] According to the comprehensive abnormal coefficient of the spring, an early warning is given to the automobile suspension spring.

[0128] Furthermore, the multidimensional data fusion model includes a first data processing layer, a feature extraction layer and a multidimensional correlation analysis layer. Figure 3 ;

[0129] The first data processing layer obtains first standard data by cleaning and normalizing the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring;

[0130] The feature extraction layer obtains a first feature set by extracting features from the first standard data; the first feature set includes stress fluctuation features, compression curve features, vibration period features and temperature change rate features;

[0131] The multidimensional correlation analysis layer obtains spring multidimensional correlation data by performing multidimensional data fusion on the stress fluctuation features, compression curve features, vibration period features and temperature change rate features in the first feature set.

[0132] Furthermore, the specific process of the multi-dimensional data fusion is as follows:

[0133] The first feature set is represented as a feature matrix, and the feature matrix is ​​decomposed to obtain a potential feature corresponding to each feature; and a feature association between features is generated by analyzing a change trend of the features;

[0134] All features in the first feature set are comprehensively analyzed to obtain feature associations between all features, and to generate spring multidimensional association data; the spring multidimensional association data includes feature association data and the first feature set.

[0135] Table 3 Automobile suspension spring data analysis comparison table

[0136]

[0137] In this embodiment, multidimensional data fusion obtains potential features through the construction and decomposition of feature matrices, generates associations between features, and realizes multidimensional correlation data between various parameters of automobile suspension springs. Compared with the analysis of single data, it can more comprehensively and dynamically reflect the change trend and mutual influence of springs under different working conditions, and provide higher accuracy and reliability for abnormal detection and early warning. For specific differences, refer to Table 3.

[0138] The data correlation of automobile suspension springs is obtained by collecting dynamic data of automobile suspension springs in actual operation, such as stress, compression, vibration frequency, temperature, etc., and then calculating the correlation coefficient between data through data mining and feature correlation analysis. The correlation coefficient is obtained based on the correlation, covariance and dependence between features; while the single data analysis is calculated by the independent analysis of the feature itself. The accuracy of anomaly detection is determined by trend analysis and correlation detection of multiple features, identifying anomalies through anomaly recognition models, and calculating the accuracy of detection by training models with real data sets and marked anomalies; while single data is detected by the degree of deviation of the data curve using a single feature. Data redundancy processing can obtain the reduction ratio of data redundancy errors by cleaning historical automobile suspension spring data, calculating the similarity and redundant error correction ratio between features, multi-dimensional data fusion technology noise processing and data redundancy reduction; single data analysis generally relies on filters and single feature statistical verification through noise processing of a single feature. Fault location is determined by obtaining historical automobile suspension spring data multiple times and determining it by the correctness ratio. Multi-dimensional data fusion is based on abnormal pattern analysis of multiple features, which can locate potential sources of faults. Through correlation analysis of features such as dynamic stress and compression, the cause and location of the abnormality can be more accurately identified and located, thereby achieving a high positioning accuracy rate. A single feature is difficult to reflect complex fault patterns, so the accuracy of fault location is low.

[0139] Multidimensional data fusion has significant advantages over single data analysis in many aspects. First, in terms of data correlation, multidimensional data fusion can discover potential correlations in automobile suspension spring data by analyzing the correlations between multiple features, while single data analysis only focuses on a single feature and it is difficult to effectively obtain the correlations between data features. The accuracy of anomaly detection is also significantly improved. Multidimensional data fusion combines the changing trends of multi-feature data and can detect anomalies more accurately, while single data analysis relies on a single feature, resulting in lower accuracy of anomaly detection.

[0140] In terms of data redundancy processing, multidimensional data fusion can reduce the error caused by noise between single features in automobile suspension spring data, thereby improving the robustness of the data, while the data analysis of a single feature is susceptible to noise interference and has low reliability. In terms of feature interpretability, multidimensional data fusion reveals the explanatory information of potential features through feature decomposition, which helps to understand the system status. In addition, in fault location, multidimensional data fusion can more accurately locate the source of potential faults by analyzing multi-feature related data, while single feature analysis is difficult to determine the specific location of the fault.

[0141] From the comparison in Table 3, it can be seen that multidimensional data fusion has obvious advantages over single data analysis in terms of correlation, anomaly detection accuracy, redundant processing and fault location, and provides a more reliable basis for system status assessment and early warning.

[0142] Further, the anomaly identification model includes a second data processing layer and a first anomaly analysis layer;

[0143] The second data processing layer obtains second standard data by cleaning and normalizing the spring multi-dimensional correlation data;

[0144] The first abnormality analysis layer identifies potential abnormalities by analyzing the second standard data to obtain a first abnormality coefficient;

[0145] The specific calculation formula of the first abnormal coefficient is:

[0146] ;

[0147] in, is the first abnormal coefficient, is the time window length for data collection, is the number of features in the first feature set, For the The first feature weight, For Moment feature data, For the The mean of feature data, For the The standard deviation of the characteristic data, is the number of feature-related data, For the The first feature associated data weight, For Moment The correlation abnormality of feature data.

[0148] Furthermore, the parameter adjustment model includes a working condition identification layer, a parameter adjustment layer, a real-time data analysis layer and a second abnormality identification layer;

[0149] The working condition identification layer identifies the load data, road conditions, ambient temperature and ambient humidity, and obtains the current working condition type through a clustering algorithm;

[0150] The parameter adjustment layer adjusts the monitoring data threshold by analyzing the working condition type, load data, ambient temperature and ambient humidity;

[0151] The real-time data analysis layer obtains second abnormal data by comparing the first data set acquired in real time with the adjusted threshold value;

[0152] The threshold value is obtained through historical data of automobile suspension springs under different operating conditions;

[0153] The second abnormality identification layer obtains a second abnormality coefficient by analyzing the second abnormality data.

[0154] Furthermore, the specific calculation formula of the second abnormal coefficient is:

[0155] ;

[0156] in, is the second abnormal coefficient, is the global adjustment coefficient, which is used to adjust the overall sensitivity of the abnormal coefficient. is the number of features in the first feature set, For the The second feature weight, For the feature data, For the The threshold value after feature data adjustment, is the number of feature-related data, For the The second feature associated data weight, is the maximum value function, For the The associated outliers of feature data, After adjusting the threshold A feature is associated with a data value.

[0157] The present invention uses a multi-dimensional data fusion model, anomaly identification model and parameter adjustment model to accurately evaluate the health status of automobile suspension springs under multiple working conditions. The system improves the health monitoring accuracy and fault diagnosis real-time performance of automobile suspension springs under multiple working conditions, which helps to extend the service life of the equipment and enhance safety.

[0158] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A parameter analysis method for automobile suspension springs, characterized in that: include: Monitoring the suspension spring under multiple working conditions to obtain a first data set; the first data set includes dynamic stress data of the spring, spring compression data, vibration frequency data, spring temperature data, load data, road conditions, ambient temperature and ambient humidity; Constructing a spring multi-dimensional data fusion model to analyze the first data set, and generating spring multi-dimensional correlation data by correlating and identifying the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring; The multidimensional data fusion model includes a first data processing layer, a feature extraction layer and a multidimensional correlation analysis layer; The first data processing layer obtains first standard data by cleaning and normalizing the dynamic stress data, spring compression data, vibration frequency data, spring temperature data and load data of the spring; The feature extraction layer obtains a first feature set by extracting features from the first standard data; the first feature set includes stress fluctuation features, compression curve features, vibration period features and temperature change rate features; The multidimensional correlation analysis layer obtains spring multidimensional correlation data by performing multidimensional data fusion on the stress fluctuation feature, compression curve feature, vibration period feature and temperature change rate feature in the first feature set; The specific process of multidimensional data fusion is as follows: The first feature set is represented as a feature matrix, and the feature matrix is ​​decomposed to obtain a potential feature corresponding to each feature; and a feature association between features is generated by analyzing a change trend of the features; Comprehensively analyzing all the features in the first feature set, obtaining feature associations among all the features, and generating multi-dimensional association data of the spring; The spring multi-dimensional association data includes feature association data and a first feature set; Constructing an abnormality recognition model to analyze the multi-dimensional correlation data of the spring and generate a first abnormality coefficient; The anomaly identification model includes a second data processing layer and a first anomaly analysis layer; The second data processing layer obtains second standard data by cleaning and normalizing the spring multi-dimensional correlation data; The first abnormality analysis layer identifies potential abnormalities by analyzing the second standard data to obtain a first abnormality coefficient; The specific calculation formula of the first abnormal coefficient is: in, is the first abnormal coefficient, is the time window length for data collection, is the number of features in the first feature set, For the The first feature weight, For Moment feature data, For the The mean of the feature data, For the The standard deviation of the characteristic data, is the number of feature-related data, For the The first feature associated data weight, For Moment The degree of abnormality associated with each feature data; By analyzing the road condition, adjusting the spring multidimensional data threshold, constructing a parameter adjustment model to analyze the spring multidimensional correlation data and the first data set, and generating a second abnormal coefficient; The parameter adjustment model includes a working condition identification layer, a parameter adjustment layer, a real-time data analysis layer and a second abnormality identification layer; The working condition identification layer identifies the load data, road conditions, ambient temperature and ambient humidity, and obtains the current working condition type through a clustering algorithm; The parameter adjustment layer adjusts the monitoring data threshold by analyzing the working condition type, load data, ambient temperature and ambient humidity; The real-time data analysis layer obtains second abnormal data by comparing the first data set acquired in real time with the adjusted threshold value; The second abnormality identification layer obtains a second abnormality coefficient by analyzing the second abnormality data; The specific calculation formula of the second abnormal coefficient is: in, is the second abnormal coefficient, is the global adjustment coefficient, is the number of features in the first feature set, For the The second feature weight, For the feature data, For the The threshold value after feature data adjustment, is the number of feature-related data, For the The second feature associated data weight, is the maximum value function, For the The associated outliers of feature data, After adjusting the threshold feature associated data values; The spring comprehensive abnormal coefficient is generated based on the first abnormal coefficient and the second abnormal coefficient; the specific calculation formula of the spring comprehensive abnormal coefficient is: in, is the spring comprehensive abnormal coefficient, is the first abnormal coefficient weight, is the first abnormal coefficient, is the second abnormal coefficient weight, is the second abnormal coefficient; According to the comprehensive abnormal coefficient of the spring, an early warning is given to the automobile suspension spring.

2. A parameter analysis system for automobile suspension springs, characterized in that: The method for analyzing parameters of a vehicle suspension spring according to claim 1 comprises: A data acquisition unit, used for monitoring the suspension spring under multiple working conditions to obtain a first data set; A model building unit, used for building a spring multi-dimensional data fusion model to analyze the first data set and generate spring multi-dimensional correlation data; A first abnormal coefficient generating unit, used for constructing an abnormal recognition model to analyze the multi-dimensional correlation data of the spring to generate a first abnormal coefficient; A second abnormal coefficient generating unit is used to analyze the road condition, adjust the spring multi-dimensional data threshold, build a parameter adjustment model to analyze the spring multi-dimensional correlation data and the first data set, and generate a second abnormal coefficient; A comprehensive abnormal coefficient generating unit, used for generating a spring comprehensive abnormal coefficient based on the first abnormal coefficient and the second abnormal coefficient; According to the comprehensive abnormal coefficient of the spring, an early warning is given to the automobile suspension spring.

Images