Test environment abnormity monitoring method and system

Abstract

The invention discloses a test environment abnormity monitoring method and system. The anomaly monitoring method adopted by the invention comprises the steps of collecting test environment data in real time, preprocessing the collected test environment data, and performing intelligent enhancement processing on the preprocessed test environment data by using a multi-scale fractional order adaptive feedback enhancement algorithm to obtain environment data after intelligent enhancement processing; performing feature extraction optimization on the test environment data after intelligent enhancement processing through a nonlinear adaptive perturbation mapping and recursive cross optimization algorithm; performing feature fusion processing on the optimized feature data to obtain fused feature data; and performing anomaly monitoring analysis based on the fused feature data to obtain an anomaly monitoring result. According to the invention, tiny fluctuations in environmental changes can be captured; the enhanced data are more sensitive and have higher resolution, and a high-precision basis is provided for subsequent feature extraction and anomaly detection.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular, to a method and system for abnormal monitoring of a test environment with multi-sensor fusion. Background Art

[0002] In the current field of environmental monitoring, multi-sensor fusion technology has been widely applied to achieve real-time monitoring and precise analysis of environmental conditions. With the development of the Internet of Things technology, various sensors can simultaneously obtain data such as temperature, humidity, air pressure, vibration, and light intensity in the environment. However, due to problems such as heterogeneity of sensor data, noise interference, and high complexity of multi-source data fusion, traditional monitoring methods often struggle to achieve high-precision and low false-alarm rate monitoring in the face of complex and dynamic environments. At the same time, traditional environmental monitoring methods mostly rely on single sensors or simple data fusion methods, lacking the ability to deeply analyze multi-dimensional and multi-source data. Especially when there are minor changes or sudden anomalies in the environment, it is often difficult to detect anomalies in a timely manner due to insufficient monitoring sensitivity.

[0003] In addition, sensor data often contains random noise and periodic interference signals. Without effective signal enhancement and denoising processing, the environmental monitoring system is easily affected by noise interference, resulting in false alarms or missed alarms. Summary of the Invention

[0004] The present invention provides a method and system for abnormal monitoring of a test environment with multi-sensor fusion to solve the defects of inaccurate environmental data processing and incomplete feature analysis in the prior art for multi-sensor environmental monitoring.

[0005] For this purpose, the present invention adopts the following technical solutions.

[0006] In a first aspect, the present invention provides a method for abnormal monitoring of a test environment, which includes the steps of: S1. Real-time collect test environment data, preprocess the collected test environment data, and perform intelligent enhancement processing on the preprocessed test environment data using a multi-scale fractional-order adaptive feedback enhancement algorithm to obtain the environment data after intelligent enhancement processing; The multi-scale fractional-order adaptive feedback enhancement algorithm realizes intelligent enhancement processing through multi-scale variational decomposition, fractional-order derivative feature enhancement, multi-variable adaptive feedback mapping, and non-linear incremental weight optimization; S2. Optimize feature extraction for the test environment data after intelligent enhancement processing through a non-linear adaptive perturbation mapping and recursive cross-optimization algorithm; then perform feature fusion processing on the optimized feature data to obtain the fused feature data; The described non - linear adaptive perturbation mapping and recursive cross - optimization algorithm realizes high - complexity data optimization and fusion through multi - step and multi - process non - linear adaptive perturbation mapping, recursive cross - weight optimization, adaptive feedback perturbation processing, and recursive fusion; S3. Conduct anomaly monitoring and analysis based on the fused feature data to obtain the anomaly monitoring result.

[0007] Furthermore, in S1, the original test environment data is collected in real - time through multiple sensors to obtain the original test environment data; pre - processing such as time synchronization, denoising, data missing value processing, and data normalization is performed on the original test environment data to obtain the pre - processed test environment data.

[0008] Furthermore, in S1, during the implementation of the multi - scale fractional - order adaptive feedback enhancement algorithm, when performing multi - scale variational decomposition on the pre - processed test environment data, through multi - scale signal decomposition, the input signal is decomposed into multiple scale components at different frequency scales to capture the subtle changes in the test environment and extract the key information in the environmental changes.

[0009] Furthermore, in S1, during the implementation of the multi - scale fractional - order adaptive feedback enhancement algorithm, fractional - order derivative feature enhancement processing is performed on each scale component obtained through multi - scale variational decomposition to further amplify the subtle changes and enhance the feature expressiveness.

[0010] Furthermore, in S1, the specific content of the multi - variable adaptive feedback mapping and non - linear incremental weight optimization is as follows: an adaptive feedback mechanism between multiple sensor data is constructed, and incremental weight optimization is performed through the method of multi - variable non - linear mapping to achieve the output of dynamic feature enhancement.

[0011] Furthermore, in S2, the specific process of the non - linear adaptive perturbation mapping and recursive cross - optimization algorithm is as follows: feature extraction is performed on the test environment data after intelligent enhancement processing, and the extracted features form an initial feature matrix. Non - linear adaptive perturbation mapping is achieved for each feature in the initial feature matrix through the combined action of non - linear mapping and sine modulation to improve the robustness and discrimination ability of the features; after completing the non - linear adaptive perturbation mapping, recursive cross - weight optimization is performed on the perturbed feature data; the feature data obtained from the recursive cross - weight optimization is subjected to adaptive feedback perturbation processing.

[0012] Even further, in S2, all the feature data after adaptive feedback perturbation processing are fused to obtain the fused initial feature data, and further non - linear feature fusion mapping is performed on the fused initial feature data to capture higher - level feature associations.

[0013] Further, in step S2, after obtaining the fused feature data after non-linear feature fusion mapping, the fused feature data is gradually recursively optimized through cross-recursive iterative fusion to obtain the finally fused feature data, so as to achieve comprehensive expression among different sensor features.

[0014] Further, the finally fused feature data is subjected to standardization processing, and single-point anomaly, dynamic anomaly, and clustering anomaly detection are performed based on the standardized fused feature data to obtain an anomaly detection result.

[0015] In a second aspect, the present invention provides an experimental environment anomaly monitoring system for implementing the above experimental environment anomaly monitoring method, which includes: Intelligent enhancement processing unit: Real-time collect experimental environment data, preprocess the collected experimental environment data, and perform intelligent enhancement processing on the preprocessed experimental environment data by using a multi-scale fractional-order adaptive feedback enhancement algorithm to obtain the intelligent enhanced environment data; The multi-scale fractional-order adaptive feedback enhancement algorithm realizes intelligent enhancement processing through multi-scale variational decomposition, fractional-order derivative feature enhancement, multivariable adaptive feedback mapping, and non-linear incremental weight optimization; Feature optimization and fusion unit: Extract and optimize features of the experimentally enhanced environment data through a non-linear adaptive perturbation mapping and recursive cross-optimization algorithm; then perform feature fusion processing on the optimized feature data to obtain the fused feature data; The non-linear adaptive perturbation mapping and recursive cross-optimization algorithm realizes high-complexity data optimization and fusion through multi-step and multi-process non-linear adaptive perturbation mapping, recursive cross-weight optimization, adaptive feedback perturbation processing, and recursive fusion; Anomaly monitoring and analysis unit: Perform anomaly monitoring and analysis based on the fused feature data to obtain an anomaly detection result.

[0016] The beneficial effects of the present invention are: 1. Through the multi-scale fractional-order adaptive feedback enhancement algorithm, the experimental environment data is refined and analyzed at different frequency scales, and it is possible to capture the minute fluctuations in the environmental changes. Multi-scale variational decomposition and fractional-order derivative feature enhancement not only improve the feature expression of the data, but also significantly amplify the subtle dynamic changes in the experimental environment data at different scales, making the enhanced data more sensitive and having higher resolution, providing a high-precision basis for subsequent feature extraction and anomaly detection.

[0017] 2. During the optimization of feature extraction, the non-linear adaptive perturbation mapping and recursive cross-weight optimization algorithm are used to further process the experimental environment data after intelligent enhancement. Through non-linear perturbation mapping and recursive cross-weight optimization, this algorithm effectively enhances the robustness and discrimination ability of the feature data, enabling the features of each sensor to be highly sensitive to external changes and resistant to internal fluctuations, thus ensuring the stability and consistency of the feature data.

[0018] 3. Through non-linear adaptive perturbation mapping, recursive cross-weight optimization, adaptive feedback perturbation processing, and recursive fusion steps, multi-sensor feature data are fused into a comprehensive expression. The multi-step fusion process ensures the strengthening of the correlation of features in the multi-dimensional space. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flowchart of a method for monitoring anomalies in an experimental environment with multi-sensor fusion according to the present invention; Figure 2 is a composition diagram of a system for monitoring anomalies in an experimental environment with multi-sensor fusion according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0022] The following specifically describes the specific solution of a method for monitoring anomalies in an experimental environment with multi-sensor fusion provided by the present invention in conjunction with the accompanying drawings.

[0023] Refer to the attached Figure 1 , which shows a flowchart of a method for monitoring anomalies in an experimental environment with multi-sensor fusion provided by an embodiment of the present invention. The method includes the following steps: S1. Real-time collect the experimental environment data, preprocess the collected experimental environment data, and use the multi-scale fractional-order adaptive feedback enhancement algorithm to perform intelligent enhancement processing on the preprocessed experimental environment data to obtain the intelligent enhanced environment data. Specifically: Collect the test environment data in real time through multiple sensors to obtain the original test environment data; the types of sensors include temperature sensors, humidity sensors, barometric pressure sensors, vibration sensors, light intensity sensors, etc. Perform preprocessing such as time synchronization, denoising, missing data processing, and data normalization on the original test environment data to obtain the preprocessed environment data. The technologies used in the preprocessing process are all well-known existing technologies in the art and will not be elaborated here.

[0024] Perform intelligent enhancement processing on the preprocessed test environment data using the multi-scale fractional-order adaptive feedback enhancement algorithm. The multi-scale fractional-order adaptive feedback enhancement algorithm realizes intelligent enhancement processing through multi-scale variational decomposition, fractional-order derivative feature enhancement, multivariable adaptive feedback mapping, and non-linear incremental weight optimization. The specific implementation process is as follows: First, perform multi-scale variational decomposition on the preprocessed test environment data to extract the key information in the environmental changes. The core of multi-scale variational decomposition is to decompose the input signal into multiple scale components at different frequency scales through multi-scale signal decomposition, so as to accurately capture the subtle changes in the test environment. The formula is as follows:

[0025] Among them, represents the signal collected from the th sensor, is the time variable; is the scale ordinal number of the decomposition, with the range , used to represent that the signal is decomposed into multi-scale components at different frequency scales; is the frequency th decomposition component at the th order, used to characterize the signal characteristics at this frequency; is the frequency variable, used to represent the signal characteristics of different frequency bands; is the total number of decomposition scales; in order to better simulate the environmental noise and periodic fluctuations in the sensor data, add a noise term to the decomposed data, is the amplitude of the random noise, simulating the influence of sensor noise; is the frequency of the noise, representing the frequency components of the sensor noise; is the phase of the noise, reflecting the position of the noise signal on the time axis.

[0026] Secondly, for each scale component Perform fractional derivative feature enhancement processing to further amplify subtle changes and enhance feature expressiveness. Fractional derivatives are used to enhance signal variations within a local range, enabling even the slightest environmental changes to be significantly amplified. The formula for fractional derivatives is as follows:

[0027] where is the result of the fractional derivative operation, and the th order is decomposed into components The enhancement value at frequency ; is the order of the fractional derivative, with a value range of , representing the degree of local enhancement and used to control the distribution characteristics of the derivative in the local area; is the normalized gamma function, used to normalize the results to ensure comparability of results for different orders; is the integration variable, used for the cumulative effect in the frequency domain; is at frequency The th order scale component, representing the frequency components after decomposition. Through this fractional operation, local variations can be enhanced on the basis of global features, extracting subtle dynamic information in the data, making the data information expression more delicate, and laying a foundation for the next feedback mapping.

[0028] Finally, after obtaining the multi-scale data enhanced by fractional derivatives, perform multi-variable adaptive feedback mapping and non-linear incremental weight optimization processing. Construct an adaptive feedback mechanism between multiple sensor data, and perform non-linear incremental weight optimization through multi-variable mapping to achieve the output of dynamic feature enhancement. The specific formula is:

[0029] where is the output data of sensor after intelligent enhancement processing at time ; is the total number of frequencies after sensor data decomposition, used to represent the fusion of multi-frequency components; is the feedback weight between sensor and sensor at the th order scale component; is the result after fractional derivative enhancement of the th order decomposed component at frequency ; is the feedback adjustment factor, controlling the feedback intensity; is the bias parameter of the nonlinear mapping, used to adjust the response of the nonlinear feedback. The feedback mapping process can adaptively adjust the data augmentation effect according to the changes in the environmental signals dynamically.

[0030] After the above processing, the environmental data after intelligent enhancement processing is obtained.

[0031] S2. Feature extraction optimization is performed on the experimental environmental data after intelligent enhancement processing through the nonlinear adaptive perturbation mapping and recursive cross-optimization algorithm; then, feature fusion processing is performed on the optimized feature data to obtain the fused feature data. The nonlinear adaptive perturbation mapping and recursive cross-optimization algorithm realizes high-complexity data optimization through multi-step and multi-process nonlinear adaptive perturbation mapping, recursive cross-weight optimization, and adaptive feedback perturbation processing. The specific implementation process is as follows: First, use existing feature engineering techniques to extract features from the environmental data after intelligent enhancement processing, and form an initial feature matrix with the extracted features , , where any represents the feature data of the th sensor, and is the total number of sensors. For each feature in the initial feature matrix , an adaptive nonlinear perturbation mapping is realized through the combined action of nonlinear mapping and sine modulation to improve the robustness and discrimination ability of the features. The adaptive nonlinear perturbation mapping function is defined as:

[0032] where, is the feature data after perturbation mapping; is the adaptive perturbation factor for each sensor feature, with an initial value of 1; is the nonlinear perturbation mapping function, including a logarithmic amplification term and a sine modulation term; is the logarithmic nonlinear mapping, used to amplify the relative change of low-value features and enhance the influence of weak signals; A sine perturbation term is introduced to increase the periodic modulation of the features, making the features more sensitive to external changes.

[0033] Secondly, after completing the adaptive nonlinear perturbation mapping, recursive cross-weight optimization is performed on the feature data after perturbation mapping. The recursive cross-weight optimization is defined as follows:

[0034] where, is the feature data after recursive cross-weight optimization; is the cumulative index used to traverse all sensors, with a value range , is the total number of sensors; is the feature attenuation adjustment parameter; is the perturbation factor adjustment parameter, controlling the response speed to the features of other sensors; is the cross-perturbation amplitude, adjusting the interaction strength between features; is the non-linear interaction term based on feature differences, increasing the perturbation effect; is a small constant to prevent the denominator from being zero and ensure the stability of the calculation. The purpose of the above recursive cross-weight optimization is to maintain the independence of each feature while having a reasonable correlation between sensors, thereby enhancing the recognition of environmental changes.

[0035] Then, the features obtained from the recursive cross-weight optimization are subjected to adaptive feedback perturbation processing. The adaptive feedback perturbation processing process aims to strengthen the expression ability of the features, making them more sensitive to small changes in the environment. The formula for the adaptive feedback perturbation processing is:

[0036] where, represents the finally optimized feature data, including high sensitivity to environmental changes; is the feedback amplitude control factor, used to adjust the strength of the feedback; is the non-linear suppression term, suppressing excessive fluctuations of the features to ensure stability; is the non-linear feedback adjustment parameter; is the feedback perturbation strength control factor; is the feature after the recursive cross-weight optimization of the standard deviation, used for adaptive perturbation; is the inverse enhancement term of the standard deviation, enhancing the recognition of small features.

[0037] Finally, after the adaptive feedback perturbation processing is completed, all the optimized features are subjected to feature fusion to form a comprehensive feature expression to improve the overall expression ability of environmental information. The goal of feature fusion is to achieve mutual complementation and enhancement between the features of different sensors through reasonable weights, non-linear mappings, and recursive fusions.

[0038] Introduce a diagonal weighted matrix for initial weighting. The formula for the fused feature matrix is:

[0039] where, is the initial feature data after fusion; Represents the finally optimized feature matrix; is a diagonal weighting matrix used to assign different weights to the features of different sensors. Any element represents the adaptive fusion weight of the

[0040] After weighting, further perform a non - linear fusion mapping on to capture higher - level feature associations. The non - linear fusion mapping formula is as follows:

[0041] where, represents the fused feature data after non - linear mapping; the parameters , and control the influence of first - order, second - order, and third - order non - linear mappings. This mapping formula performs non - linear transformation on the fused features through a combination of polynomials and trigonometric functions, where expresses the non - linear combination of first - order and second - order features, captures the non - linear association of third - order features. The non - linear mapping process is to enhance the high - order interaction characteristics between features and improve the detection sensitivity to environmental anomalies.

[0042] After obtaining , perform step - by - step recursive optimization on the features through cross - recursive iterative fusion to achieve comprehensive expression among different sensor features. The recursive iterative fusion formula is:

[0043] where, is the fused feature data of the th iteration; is the fused feature data of the th iteration; is the fused feature of the feature vector at index ; is the adjustment parameter for non - linear interaction, used to control the non - linear degree of cross - terms; is the perturbation adjustment factor, used to enhance the perturbation effect of the fused features during the recursive process, prevent premature convergence of feature values, and increase the dynamic variability of the fused features.

[0044] After iteration, when the fused feature data reaches a stable value during the iteration process, that is, when the difference between the results of two adjacent iterations is less than the convergence threshold preset according to the empirical method or reaches the preset number of iterations, stop the iteration to obtain the final fused feature data , the fused feature contains the comprehensive feature expression of multi-sensors in complex environments.

[0045] Furthermore, the fused feature data is normalized to eliminate the scale differences between different features, ensuring the effectiveness and stability of the detection process, and the fused feature data after normalization is obtained; anomaly detection of single-point anomaly, dynamic anomaly, and clustering anomaly is performed based on the fused feature data after normalization, and the anomaly detection results are obtained, specifically including: Based on the fused feature data after normalization, an anomaly detection method based on statistical characteristics is adopted to quickly identify the feature changes that deviate significantly from the normal state. By calculating the single-point anomaly score of the normalized features, it is judged whether there is a single-point anomaly in the data. The calculation of the single-point anomaly score is the mean value of the fused feature data after normalization, and when the single-point anomaly score exceeds the single-point anomaly threshold preset according to the empirical method, it is marked as a single-point anomaly, which is caused by accidental factors.

[0046] To capture the mutation trend in the environment, a dynamic detection method based on temporal variation is introduced. By analyzing the change trend of features in the time series, progressive or sudden anomalies, that is, dynamic anomalies, are identified. The change rate of the features at time t of the fused feature data after normalization is calculated as the incremental change rate and used as the dynamic anomaly score; when the dynamic anomaly score exceeds the dynamic anomaly threshold preset according to the empirical method, it is marked as a dynamic anomaly, indicating the existence of a sudden or gradual change trend.

[0047] To improve the accuracy of anomaly detection, an existing density-based clustering method is used to identify the anomaly points in the data clustering, and the clustering anomaly score is calculated. When the clustering anomaly score exceeds the dynamic anomaly threshold preset according to the empirical method, it is marked as a clustering anomaly, indicating the existence of discrete changes in the environment.

[0048] The beneficial effects of the above technical solutions are: 1. Through the multi-scale fractional-order adaptive feedback enhancement algorithm, the environmental data is refined and analyzed at different frequency scales, and the tiny fluctuations in the environmental changes can be captured. This multi-scale variational decomposition and fractional-order derivative feature enhancement not only improve the feature expression of the data, but also significantly amplify the subtle dynamic changes in the environmental data at different scales, making the enhanced data more sensitive and having higher resolution, providing a high-precision basis for subsequent feature extraction and anomaly detection.

[0049] 2. During the feature extraction optimization process, the non-linear adaptive perturbation mapping and recursive cross-weight optimization algorithm are used to further process the environment data after intelligent enhancement processing. Through non-linear perturbation mapping and recursive cross-weight optimization, this algorithm effectively enhances the robustness and discrimination ability of the feature data, enabling the features of each sensor to be both highly sensitive to external changes and resistant to internal fluctuations, thus ensuring the stability and consistency of the feature data.

[0050] 3. Through the steps of adaptive weighting, non-linear mapping, and recursive fusion, the multi-sensor feature data are fused into a comprehensive expression. The multi-step fusion process ensures the strengthening of the correlation of features in the multi-dimensional space. At the same time, it adaptively adjusts the sensor weights, magnifying the contribution of important sensors in the comprehensive feature expression, and ensuring the best expression effect of multi-source information in the fused data.

[0051] Refer to the appendix Figure 2 , which shows the composition diagram of an experimental environment anomaly monitoring system for multi-sensor fusion of the present invention, used to implement the above-mentioned experimental environment anomaly monitoring method. It consists of an intelligent enhancement processing unit, a feature optimization and fusion unit, and an anomaly monitoring and analysis unit.

[0052] Intelligent enhancement processing unit: It collects the experimental environment data in real time, preprocesses the collected experimental environment data, and uses the multi-scale fractional-order adaptive feedback enhancement algorithm to perform intelligent enhancement processing on the preprocessed experimental environment data to obtain the environment data after intelligent enhancement processing.

[0053] The multi-scale fractional-order adaptive feedback enhancement algorithm realizes intelligent enhancement processing through multi-scale variational decomposition, fractional-order derivative feature enhancement, multi-variable adaptive feedback mapping, and non-linear incremental weight optimization.

[0054] Feature optimization and fusion unit: It optimizes the feature extraction of the experimental environment data after intelligent enhancement processing through the non-linear adaptive perturbation mapping and recursive cross-optimization algorithm; then performs feature fusion processing on the optimized feature data to obtain the fused feature data.

[0055] The non-linear adaptive perturbation mapping and recursive cross-optimization algorithm realizes high-complexity data optimization through multi-step and multi-process non-linear adaptive perturbation mapping, recursive cross-weight optimization, and adaptive feedback perturbation processing.

[0056] Anomaly monitoring and analysis unit: It performs anomaly monitoring and analysis based on the fused feature data to obtain the anomaly monitoring result.

[0057] It should be noted that each unit in the above-mentioned test environment anomaly monitoring system for multi-sensor fusion can be implemented in whole or in part by software, hardware, or a combination thereof. The above-mentioned units can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules. For the specific limitations of a test environment anomaly monitoring system for multi-sensor fusion, refer to the limitations of a test environment anomaly monitoring method for multi-sensor fusion in the above text. The two have the same functions and effects, and will not be elaborated here.

[0058] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for monitoring abnormalities in a test environment, characterized in that: Includes steps: S1. Collect test environment data in real time, pre-process the collected test environment data, and use a multi-scale fractional-order adaptive feedback enhancement algorithm to perform intelligent enhancement processing on the pre-processed test environment data to obtain the environment data after intelligent enhancement processing; The multi-scale fractional-order adaptive feedback enhancement algorithm realizes intelligent enhancement processing through multi-scale variational decomposition, fractional-order derivative feature enhancement, multi-variable adaptive feedback mapping and nonlinear incremental weight optimization; S2. Perform feature extraction optimization on the test environment data after intelligent enhancement processing through nonlinear adaptive perturbation mapping and recursive cross optimization algorithm; then perform feature fusion processing on the optimized feature data to obtain fused feature data; The nonlinear adaptive perturbation mapping and recursive cross optimization algorithm achieves high-complexity data optimization through multi-step and multi-process nonlinear adaptive perturbation mapping, recursive cross weight optimization and adaptive feedback perturbation processing; S3. Perform anomaly monitoring analysis based on the fused feature data to obtain anomaly monitoring results.

2. The test environment abnormality monitoring method according to claim 1, characterized in that: In S1, the test environment data is collected in real time by a variety of sensors to obtain the original test environment data; the original test environment data is preprocessed by time synchronization, denoising, data missing processing and data normalization to obtain the preprocessed test environment data.

3. The test environment abnormality monitoring method according to claim 1, characterized in that: In S1, during the implementation of the multi-scale fractional-order adaptive feedback enhancement algorithm, when performing multi-scale variational decomposition on the pre-processed test environment data, the input signal is decomposed into multiple scale components at different frequency scales through multi-scale signal decomposition, so as to capture the slight changes in the test environment and extract the key information in the environmental changes.

4. The test environment abnormality monitoring method according to claim 1, characterized in that: In S1, during the implementation of the multi-scale fractional-order adaptive feedback enhancement algorithm, fractional-order derivative feature enhancement processing is performed on each scale component obtained through multi-scale variational decomposition to further amplify subtle changes and enhance feature expression.

5. The test environment abnormality monitoring method according to claim 1, characterized in that: In S1, the specific contents of multivariable adaptive feedback mapping and nonlinear incremental weight optimization are: constructing an adaptive feedback mechanism between multiple sensor data, optimizing incremental weights by multivariable nonlinear mapping, and realizing dynamic feature enhanced output.

6. The test environment abnormality monitoring method according to claim 1, characterized in that: In S2, the specific process of implementing the nonlinear adaptive perturbation mapping and the recursive cross optimization algorithm is as follows: feature extraction is performed on the test environment data after intelligent enhancement processing, and the extracted features constitute an initial feature matrix, and nonlinear adaptive perturbation mapping is implemented for each feature in the initial feature matrix through the joint action of nonlinear mapping and sinusoidal modulation; after completing the nonlinear adaptive perturbation mapping, recursive cross weight optimization is performed on the feature data after the perturbation mapping; and the feature data obtained by the recursive cross weight optimization is subjected to adaptive feedback perturbation processing.

7. The test environment abnormality monitoring method according to claim 6, characterized in that: In S2, all feature data after the adaptive feedback disturbance processing are subjected to feature fusion to obtain fused initial feature data, and nonlinear feature fusion mapping is further performed on the fused initial feature data to capture higher-level feature associations.

8. The test environment abnormality monitoring method according to claim 7, characterized in that: In S2, after obtaining the fused feature data after the nonlinear feature fusion mapping, the fused feature data is gradually recursively optimized through cross-recursive iterative fusion to obtain the final fused feature data, so that it can achieve comprehensive expression between different sensor features.

9. The test environment abnormality monitoring method according to claim 8, characterized in that: The final fused feature data is standardized, and anomaly detection of single point anomaly, dynamic anomaly and cluster anomaly is performed based on the standardized fused feature data to obtain the anomaly detection result.

10. A test environment abnormality monitoring system, used to implement the test environment abnormality monitoring method according to any one of claims 1 to 9, characterized in that: include: Intelligent enhancement processing unit: collects test environment data in real time, preprocesses the collected test environment data, and uses a multi-scale fractional-order adaptive feedback enhancement algorithm to perform intelligent enhancement processing on the preprocessed test environment data to obtain the environment data after intelligent enhancement processing; The multi-scale fractional-order adaptive feedback enhancement algorithm realizes intelligent enhancement processing through multi-scale variational decomposition, fractional-order derivative feature enhancement, multi-variable adaptive feedback mapping and nonlinear incremental weight optimization; Feature optimization and fusion unit: The test environment data after intelligent enhancement processing is subjected to feature extraction optimization through nonlinear adaptive perturbation mapping and recursive cross optimization algorithm; then the optimized feature data is subjected to feature fusion processing to obtain fused feature data; The nonlinear adaptive perturbation mapping and recursive cross optimization algorithm achieves high-complexity data optimization through multi-step and multi-process nonlinear adaptive perturbation mapping, recursive cross weight optimization and adaptive feedback perturbation processing; Abnormal monitoring and analysis unit: performs abnormal monitoring and analysis based on the fused feature data to obtain abnormal monitoring results.

Images