Intelligent test method and system for vibration noise influence of rail transit

Abstract

The invention discloses a rail transit vibration noise influence intelligent test method and system, and mainly relates to the technical field of noise test. Comprising the steps of obtaining a target test track, performing sensor layout based on route characteristics of the target test track, and obtaining a noise sensor array after layout is completed; obtaining a noise monitoring data sequence array; traversing the noise monitoring data sequence array to perform anomaly monitoring identification on the noise sensor array to obtain an abnormal noise sensor sub-array and a conventional noise sensor sub-array; and performing online monitoring analysis on the abnormal noise sensor sub-array and the conventional noise sensor sub-array by using the abnormal test analysis channel set and the conventional test analysis network layer to obtain a target vibration noise test result. The method has the beneficial effects that the technical problem of low test result reliability caused by lack of continuous monitoring and analysis on the vibration noise influence of the rail transit in the prior art is solved, and the technical effect of improving the test accuracy is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of noise testing, and particularly to an intelligent testing method and system for the vibration and noise impact of rail transit. Background Art

[0002] At present, the construction scale of rail transit is continuously expanding, and the resulting vibration and noise pollution problems are becoming increasingly serious. Most of the existing rail transit noise monitoring methods adopt the methods of manual collection or regular spot checks, which cannot effectively achieve real-time online monitoring, and the timeliness and accuracy of data collection are insufficient. It is difficult to quickly discover and accurately locate abnormal noise and vibration areas, which brings difficulties to the safe operation and maintenance management of rail transit.

[0003] There is a technical problem in the prior art that the continuous monitoring and analysis of the vibration and noise impact of rail transit are lacking, resulting in low reliability of test results. Summary of the Invention

[0004] The present application provides an intelligent testing method and system for the vibration and noise impact of rail transit, which is used to solve the technical problem in the prior art that the continuous monitoring and analysis of the vibration and noise impact of rail transit are lacking, resulting in low reliability of test results.

[0005] In view of the above problems, the present application provides an intelligent testing method and system for the vibration and noise impact of rail transit.

[0006] In the first aspect of the present application, an intelligent testing method for the vibration and noise impact of rail transit is provided. The method includes: Obtain a target test track, and based on the route characteristics of the target test track, arrange sensors to obtain a completed noise sensor array; Conduct a train passing test on the target test track according to a preset test plan, and synchronously collect the data of the sensor array to obtain a noise monitoring data sequence array; Traverse the noise monitoring data sequence array to perform abnormal monitoring and identification on the noise sensor array, and obtain an abnormal noise sensor sub-array and a normal noise sensor sub-array. Among them, the abnormal noise sensor sub-array includes an abnormal monitoring scale sub-array; Based on the abnormal monitoring scale sub-array, construct a set of abnormal test analysis channels, and based on the normal monitoring frequency, construct a normal test analysis network layer. Use the set of abnormal test analysis channels and the normal test analysis network layer to perform online monitoring and analysis on the abnormal noise sensor sub-array and the normal noise sensor sub-array respectively, and obtain a target vibration and noise test result.

[0007] In the second aspect of the present application, an intelligent testing system for the vibration and noise impact of rail transit is provided. The system includes: A noise sensor array acquisition module, configured to obtain a target test track, arrange sensors based on the route characteristics of the target test track, and obtain a completed noise sensor array; A monitoring data sequence array acquisition module, configured to perform a train passing test on the target test track according to a preset test plan, synchronously collect data of the sensor array, and obtain a noise monitoring data sequence array; A sensor sub-array acquisition module, configured to traverse the noise monitoring data sequence array to perform abnormal monitoring and identification on the noise sensor array, and obtain an abnormal noise sensor sub-array and a normal noise sensor sub-array, where the abnormal noise sensor sub-array includes an abnormal monitoring scale sub-array; A test result acquisition module, configured to construct a set of abnormal test analysis channels based on the abnormal monitoring scale sub-array, construct a normal test analysis network layer based on the normal monitoring frequency, and use the set of abnormal test analysis channels and the normal test analysis network layer to perform online monitoring and analysis on the abnormal noise sensor sub-array and the normal noise sensor sub-array respectively, so as to obtain a target vibration noise test result.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: In this application, by obtaining a target test track, arranging sensors based on the route characteristics of the target test track to obtain a completed noise sensor array, then performing a train passing test on the target test track according to a preset test plan, synchronously collecting data of the sensor array to obtain a noise monitoring data sequence array, and then traversing the noise monitoring data sequence array to perform abnormal monitoring and identification on the noise sensor array to obtain an abnormal noise sensor sub-array and a normal noise sensor sub-array, where the abnormal noise sensor sub-array includes an abnormal monitoring scale sub-array, and then constructing a set of abnormal test analysis channels based on the abnormal monitoring scale sub-array and constructing a normal test analysis network layer based on the normal monitoring frequency, using the set of abnormal test analysis channels and the normal test analysis network layer to perform online monitoring and analysis on the abnormal noise sensor sub-array and the normal noise sensor sub-array respectively to obtain a target vibration noise test result. It achieves the technical effect of performing intelligent online monitoring on rail transit and improving the reliability of test results. Description of the Drawings

[0009] Att Figure 1 is a schematic flow chart of an intelligent test method for vibration and noise impact of rail transit provided by an embodiment of the present invention.

[0010] Att Figure 2 is a schematic structural diagram of an intelligent test system for vibration and noise impact of rail transit provided by an embodiment of the present invention.

[0011] Reference numerals shown in the drawings: Noise sensor array acquisition module 11, monitoring data sequence array acquisition module 12, sensor sub-array acquisition module 13, test result acquisition module 14. Specific implementation manners

[0012] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of this application. It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product or server including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0013] Embodiment 1, as shown in the appendix Figure 1 The present application provides an intelligent test method for the vibration and noise impact of rail transit. Among them, the method includes: S1: Obtain a target test track, perform sensor layout based on the route characteristics of the target test track, and obtain a completed noise sensor array. In a possible embodiment, the target test track refers to a specific rail transit line that needs to be tested for vibration and noise impact, which can be a complete line or a specific area to be monitored on the line. The route characteristics include track structure characteristics (such as rail type, subgrade condition, turnout position, etc.) and surrounding environment characteristics (such as tunnels, elevated sections, residential areas, etc.). The noise sensor array refers to a monitoring network formed by combining multiple noise sensors according to a preset spatial layout strategy, which is used to comprehensively and accurately obtain the noise and vibration data along the track.

[0014] In the specific implementation process, first conduct on-site investigation or historical data analysis on the target track, identify and determine different route characteristics along the line, such as different track condition areas such as straight sections, curve sections, turnout areas, tunnels and elevated sections. Then, select appropriate noise sensors and plan the layout positions according to these route characteristics. For example, increase the sensor density appropriately in the turnout area and curve area to improve the detection accuracy of these key areas. Finally, form a noise sensor array with reasonable spatial coverage and moderate density through installation and commissioning. This step can effectively improve the accuracy and representativeness of the subsequent test data, provide an accurate data basis for subsequent online real-time monitoring and analysis, and ensure that the test results can objectively and accurately reflect the actual vibration and noise impact of rail transit.

[0015] S2: Conduct train passing tests on the target test track according to a preset test plan, and synchronously collect the data of the sensor array to obtain a noise monitoring data sequence array. In a possible embodiment, the preset test plan is the specific content of the train passing test planned and formulated in advance by those skilled in the art, including the running speed of the train, the driving route, the test time and frequency, etc. By obtaining the preset test plan, it lays the foundation for the subsequent real train operation on the target test track to trigger track vibration and noise in the actual operation state. Each noise monitoring data in the noise monitoring data sequence array refers to the data set collected by the noise sensor arranged in chronological order, and the monitoring data generated by each sensor constitutes a separate time series.

[0016] First, according to the pre-formulated test plan, clarify the operating conditions of the train on the target test track, including specific running speeds (such as different speed levels of 60 km / h, 80 km / h, 120 km / h, etc.), the types of test trains (such as bullet trains, subway vehicles, urban rail vehicles, etc.), and the time periods and frequencies of the tests. Then, start the noise sensor array when the train passes through the target track. Each sensor is precisely synchronized relying on GPS timing technology or network timing protocols (such as NTP) to realize the real-time and continuous collection of vibration and noise data during the train passing process, forming a noise monitoring data sequence array with a clear time dimension. This step realizes the accurate monitoring of the real working conditions of rail transit, and the obtained data array can accurately reflect the dynamic change characteristics of track vibration and noise in different regions and working conditions, providing high-quality basic data for the next step of anomaly identification and online analysis.

[0017] S3: Traverse the noise monitoring data sequence array to perform anomaly monitoring and identification on the noise sensor array, and obtain an abnormal noise sensor sub-array and a normal noise sensor sub-array, where the abnormal noise sensor sub-array includes an abnormal monitoring scale sub-array. Further, traverse the noise monitoring data sequence array to perform anomaly monitoring and identification on the noise sensor array, and obtain an abnormal noise sensor sub-array and a normal noise sensor sub-array, where the abnormal noise sensor sub-array includes an abnormal monitoring scale sub-array. Step S3 of the embodiment of the present application further includes: Arbitrarily extract a first noise monitoring data sequence from the noise monitoring data sequence array. Extract the noise monitoring characteristics of the first noise monitoring data sequence to obtain a first noise monitoring data feature set. In the pre-built anomaly monitoring and recognition prototype library, match the first noise monitoring data feature set. If the match is successful, mark the noise sensor corresponding to the first noise monitoring data sequence as abnormal, add it to the abnormal noise sensor sub-array, and add the first abnormal monitoring scale obtained from the match to the abnormal monitoring scale sub-array.

[0018] In a possible embodiment, the abnormal noise sensor sub-array refers to the subset composed of sensors with identified abnormal data. The conventional noise sensor sub-array is the subset composed of sensors without identified abnormalities. The abnormal monitoring scale sub-array represents the interval scale used for monitoring and analyzing the abnormal noise sensor sub-array.

[0019] Preferably, randomly extract the first noise monitoring data sequence from the noise monitoring data sequence array. Use signal processing and analysis algorithms (such as Fourier transform FFT, short-time Fourier transform STFT, wavelet transform, or statistical feature extraction) to extract features from this data sequence to generate the first noise monitoring data feature set, including indicators such as spectral features, energy distribution, peak positions, and vibration amplitude change trends. Optionally, obtain multiple sample noise monitoring data sequences and multiple sample noise monitoring data feature sets as training data, and perform supervised training on the framework constructed based on the feedforward neural network until the training converges to obtain a trained feature recognizer. Use the feature recognizer to perform feature recognition on the first noise monitoring data sequence to obtain the first noise monitoring data feature set.

[0020] Furthermore, match and compare the extracted first noise monitoring data feature set with the pre-built anomaly monitoring and recognition prototype library. That is to say, calculate the cosine similarity between the first noise monitoring data feature set and the recognition internal nodes in the anomaly monitoring and recognition tree in the anomaly monitoring and recognition prototype library. When the calculation result is greater than or equal to the preset similarity (the similarity preset by those skilled in the art), the match is successful, the corresponding noise sensor is determined to be abnormal and classified into the abnormal noise sensor sub-array, and the corresponding abnormal scale is recorded and added to the abnormal monitoring scale sub-array. If the match is not successful, the noise sensor is classified into the conventional noise sensor sub-array.

[0021] By using the anomaly monitoring and recognition prototype library to quickly identify abnormal noise data in the actual operation of rail transit, so as to timely discover abnormal working conditions and potential risks, provide effective data partitioning and feature support for subsequent online intelligent analysis of abnormal noise, and achieve precise and efficient management of rail noise anomalies.

[0022] Further, step S3 of this application embodiment further includes: Obtain a set of historical noise monitoring data sequences, a corresponding set of historical noise types, and a set of historical noise response durations; Taking the noise type as an index, perform homogeneous aggregation on the set of historical noise types, and perform mapping clustering on the set of historical noise monitoring data sequences and the historical noise response durations according to the homogeneous aggregation result to obtain K sets of clustered historical noise monitoring data sequences and K sets of clustered historical noise response durations, where K is a positive integer; Traverse the K sets of clustered historical noise monitoring data sequences to extract noise monitoring features, and obtain K sets of clustered historical noise monitoring feature sets; Perform secondary aggregation on the K sets of clustered historical noise monitoring feature sets, combine the K sets of clustered historical noise response durations, construct K abnormal monitoring recognition trees, and summarize the K abnormal monitoring recognition trees to generate the abnormal monitoring recognition prototype library.

[0023] Further, perform secondary aggregation on the K sets of clustered historical noise monitoring feature sets, combine the K sets of clustered historical noise response durations, and construct K abnormal monitoring recognition trees. Step S3 of the embodiment of the present application further includes: Extract a first set of clustered historical noise monitoring features from the K sets of clustered historical noise monitoring feature sets; Randomly extract M first clustered historical noise monitoring features from the first set of clustered historical noise monitoring features as M initial recognition internal nodes; Use the M initial recognition internal nodes to aggregate the first set of clustered historical noise monitoring features to obtain M sets of aggregated historical noise monitoring features; Use an aggregation verification function to perform aggregation authentication on the M sets of aggregated historical noise monitoring features and the M first recognition internal nodes. If the authentication passes, then take the M initial clustered historical noise monitoring features as the M recognition internal nodes, and perform monitoring scale identification on the M recognition internal nodes based on the corresponding first set of clustered historical noise response durations to obtain the constructed first abnormal monitoring recognition tree; Analyze the K sets of clustered historical noise monitoring feature sets and the K sets of clustered historical noise response durations to obtain the K abnormal monitoring recognition trees.

[0024] Further, the aggregation verification function is: ; Wherein, is the aggregation degree output by the aggregation verification function, is the j-th aggregated historical noise monitoring feature in the i-th set of aggregated historical noise monitoring features in the M sets of aggregated historical noise monitoring features, It is the first clustering historical noise monitoring feature corresponding to the i-th initial recognition internal node in the M clustering historical noise monitoring feature sets. It is the total number of clustering historical noise monitoring features in the i-th clustering historical noise monitoring feature set; Input the M clustering historical noise monitoring feature sets and M first recognition internal nodes into the clustering verification function to obtain the clustering degree; Judge whether the clustering degree meets the preset clustering degree threshold. If so, the authentication passes.

[0025] In a possible embodiment, the historical noise monitoring data sequence set refers to the data set obtained from monitoring the rail transit noise and vibration in a certain period in the past. Each sequence represents the noise data recorded under a certain specific condition or event. The historical noise type set refers to the categories of different abnormal events corresponding to the historical monitoring data, such as track faults, vehicle faults or other specific abnormalities. The historical noise response duration set refers to the time duration from the start of the indication of an abnormal event to the occurrence of the abnormality. The abnormal monitoring recognition tree is a classification tree constructed based on historical data features and response duration, which is used for rapid feature comparison and classification during real-time abnormal recognition. The abnormal monitoring recognition prototype library is a knowledge base formed by summarizing and organizing all abnormal recognition trees, which can be used for matching and recognition of real-time abnormal noise data features.

[0026] First, obtain the historical noise monitoring data sequence set and the corresponding historical noise types and abnormal response durations, and construct an initial historical data set; then, index by noise type, classify and aggregate the data (such as aggregating switch abnormalities, vehicle faults, etc. separately), and perform mapping clustering processing on the noise monitoring data sequence and the corresponding response duration according to the classification results to generate K clustering historical data sets (for example, K = 5 means aggregating into 5 different abnormal categories); subsequently, use the above-obtained feature recognizer to traverse each clustering data set for feature extraction, such as extracting indicators such as spectral energy, peak frequency, and time-domain energy change trend, to form K clustering historical noise monitoring feature sets.

[0027] Furthermore, perform secondary aggregation on these feature sets, combine the response duration characteristics corresponding to the abnormalities, and construct K abnormal monitoring recognition trees through a tree structure (such as a decision tree or a random forest). Each tree can efficiently identify the data features of a specific abnormal type and the corresponding response duration features. Finally, summarize and integrate these K abnormal monitoring recognition trees to form a unified abnormal monitoring recognition prototype library.

[0028] By establishing an accurate and efficient historical data feature recognition model, a stable and fast feature matching standard is provided for the anomaly recognition of real-time noise monitoring data, achieving the technical effects of improving the anomaly recognition accuracy and efficiency of real-time monitoring and effectively supporting the intelligent management of vibration and noise in rail transit.

[0029] Preferably, the first clustered historical noise monitoring feature set is the first feature set selected from the previously obtained K clustered historical noise monitoring feature sets, representing the feature set of historical monitoring data for a specific type of anomaly. The initial recognition internal node is the feature node initially randomly selected as the classification basis or reference when constructing the anomaly monitoring recognition tree. The aggregated historical noise monitoring feature set is the feature set re-divided by using the initial recognition internal node as the clustering basis, and each set represents data with similar features. The aggregation verification function is a quantitative evaluation function used to judge the rationality and compactness (aggregation degree) of the clustering result to ensure the effectiveness of the clustering division.

[0030] Calculate the similarity between the first clustered historical noise monitoring feature set and the M initial recognition internal nodes respectively by using the cosine similarity formula, and classify the first clustered historical noise monitoring features into the set corresponding to the initial recognition internal node with the highest similarity, so as to obtain the M aggregated historical noise monitoring feature sets. Furthermore, use the aggregation verification function to analyze and quantify the aggregation degree of the M aggregated historical noise monitoring feature sets, and judge whether the aggregation degree meets the preset aggregation degree threshold (the minimum aggregation degree that can be certified and passed pre-set by those skilled in the art). If so, the certification passes. If not, re-select the M initial recognition internal nodes.

[0031] Furthermore, step S3 of the embodiment of the present application further includes: Based on the one-to-one mapping relationship between the historical noise response duration and the historical noise monitoring features, map and aggregate the first clustered historical noise response duration set based on the M aggregated historical noise monitoring feature sets to obtain M aggregated historical noise response duration sets; Traverse and calculate the mean values of the M aggregated historical noise response duration sets to obtain M monitoring scales, and label the M recognition internal nodes based on the M monitoring scales to obtain the first anomaly monitoring recognition tree.

[0032] Preferably, first, according to the established corresponding relationship between the historical noise monitoring features and the response duration, based on the previously obtained M aggregated historical noise monitoring feature sets, the corresponding first clustering historical noise response duration sets are associated and clustered, so that each feature set corresponds to a unique response duration set, forming M new aggregated historical noise response duration sets; Subsequently, each response duration set is traversed and calculated, for example, the mean value of all response durations in each set is taken to obtain M monitoring scales representing this type of abnormal feature, and these scales can be used to represent the average response duration of different abnormal situations. Next, using the obtained M monitoring scales as identifiers, they are corresponded to the M recognition internal nodes in the abnormal monitoring recognition tree to form scale markings, so that each node of the abnormal recognition tree can reflect the quantization degree of the corresponding abnormal feature. Finally, through the above steps, the construction of the complete first abnormal monitoring recognition tree is completed. The important role of this process is to clarify the corresponding relationship between the abnormal features and the response duration, and intuitively reflect the degree of abnormality in a quantitative manner, providing an accurate and efficient judgment standard for subsequent real-time abnormal monitoring analysis.

[0033] S4: Construct an abnormal test analysis channel set based on the abnormal monitoring scale sub-array, and construct a conventional test analysis network layer based on the conventional monitoring frequency. Use the abnormal test analysis channel set and the conventional test analysis network layer to perform online monitoring and analysis on the abnormal noise sensor sub-array and the conventional noise sensor sub-array respectively to obtain the target vibration noise test results.

[0034] Furthermore, for constructing the abnormal test analysis channel set based on the abnormal monitoring scale sub-array, step S4 of the embodiment of the present application further includes: Obtain multiple sample noise monitoring data sequences, and perform abnormal recognition on the multiple sample noise monitoring data sequences based on the abnormal monitoring scale sub-array to obtain multiple sample abnormal recognition result arrays; Combine the multiple sample noise monitoring data sequences with the multiple sample abnormal recognition result arrays respectively for abnormal test analysis channel construction to obtain the abnormal test analysis channel set.

[0035] In a possible embodiment, the set of abnormal test analysis channels refers to multiple independent data analysis channels specifically for analyzing abnormal noise data constructed based on the sub-array of abnormal monitoring scales determined in the foregoing steps, and each channel uses a specific abnormal monitoring scale as the analysis benchmark. The conventional test analysis network layer is constructed according to the typical noise monitoring frequencies under the normal operating conditions of rail transit, and is used to analyze the conventional noise data that has not been determined to be abnormal in real time. The sample noise monitoring data sequence is a plurality of typical data sequences extracted from historical or real-time monitoring data for constructing and verifying the abnormal analysis channels. The sample abnormal recognition result array represents the abnormal determination result array formed after the abnormal recognition of the sample data sequence, and each result corresponds to the abnormal state (such as abnormal position, duration, etc.) of a specific data sequence.

[0036] Preferably, first, obtain multiple noise sample data sequences collected from the actual operation monitoring history or recent real-time of rail transit, such as typical data sequences collected under different track conditions (switches, curve sections, elevated sections). Then, based on the sub-array of abnormal monitoring scales established in step S3, perform abnormal recognition analysis on these sample data sequences one by one, determine whether there are abnormal features in each sample sequence, and output their respective abnormal recognition result arrays, such as abnormal information such as the occurrence position, frequency characteristics, vibration amplitude exceeding limit conditions, or duration.

[0037] Furthermore, combine multiple sample noise monitoring data sequences with multiple sample abnormal recognition result arrays respectively to obtain multiple training sample data combinations. Use the multiple training sample data combinations to perform supervised training on the framework constructed based on the feedforward neural network until the training converges to obtain the trained abnormal test analysis channels. By obtaining multiple independent channel combinations constructed for different abnormal types and scale characteristics to form the set of abnormal test analysis channels, the accurate, efficient, and intelligent analysis of abnormal situations during the real-time monitoring of rail transit vibration and noise is realized, achieving the technical effect of improving the reliability of test results.

[0038] Based on the same principle as constructing the set of abnormal test analysis channels, construct the conventional test analysis network layer based on the conventional monitoring frequency. Among them, the conventional test analysis network layer is used to simultaneously monitor and analyze the sub-array of conventional noise sensors.

[0039] In summary, the embodiments of the present application have at least the following technical effects: This application obtains a target test track, arranges sensors based on the route characteristics of the target test track to obtain a completed noise sensor array, then conducts a train passing test on the target test track according to a preset test plan, synchronously collects the data of the sensor array to obtain a noise monitoring data sequence array, and then traverses the noise monitoring data sequence array to perform abnormal monitoring and identification on the noise sensor array to obtain an abnormal noise sensor sub-array and a normal noise sensor sub-array. Among them, the abnormal noise sensor sub-array includes an abnormal monitoring scale sub-array. Then, an abnormal test analysis channel set is constructed based on the abnormal monitoring scale sub-array, and a normal test analysis network layer is constructed based on the normal monitoring frequency. The abnormal test analysis channel set and the normal test analysis network layer are used to perform online monitoring and analysis on the abnormal noise sensor sub-array and the normal noise sensor sub-array respectively to obtain the target vibration noise test result. It achieves the technical effect of intelligently online monitoring rail transit and improving the reliability of test results.

[0040] Embodiment 2, based on the same inventive concept as the intelligent test method for vibration and noise impact of a rail transit in the foregoing embodiment, as shown in the appendix Figure 2 This application provides an intelligent test system for vibration and noise impact of a rail transit. The system in the embodiment of this application and the method embodiment are based on the same inventive concept. Among them, the system includes: A noise sensor array obtaining module 11, configured to obtain a target test track, arrange sensors based on the route characteristics of the target test track to obtain a completed noise sensor array; A monitoring data sequence array obtaining module 12, configured to conduct a train passing test on the target test track according to a preset test plan, synchronously collect the data of the sensor array to obtain a noise monitoring data sequence array; A sensor sub-array obtaining module 13, configured to traverse the noise monitoring data sequence array to perform abnormal monitoring and identification on the noise sensor array to obtain an abnormal noise sensor sub-array and a normal noise sensor sub-array. Among them, the abnormal noise sensor sub-array includes an abnormal monitoring scale sub-array; A test result obtaining module 14, configured to construct an abnormal test analysis channel set based on the abnormal monitoring scale sub-array, and construct a normal test analysis network layer based on the normal monitoring frequency, and use the abnormal test analysis channel set and the normal test analysis network layer to perform online monitoring and analysis on the abnormal noise sensor sub-array and the normal noise sensor sub-array respectively to obtain the target vibration noise test result.

[0041] Further, the sensor sub-array obtaining module 13 is configured to perform the following steps: Arbitrarily extract a first noise monitoring data sequence from the noise monitoring data sequence array; Extract noise monitoring features from the first noise monitoring data sequence to obtain a first set of noise monitoring data features; In the pre-constructed anomaly monitoring and recognition prototype library, match the first set of noise monitoring data features. If the match is successful, mark the noise sensor corresponding to the first noise monitoring data sequence as an anomaly, add it to the sub-array of abnormal noise sensors, and add the first anomaly monitoring scale obtained from the match to the sub-array of anomaly monitoring scales.

[0042] Furthermore, the sensor sub-array obtaining module 13 is used to perform the following steps: Obtain a set of historical noise monitoring data sequences and corresponding sets of historical noise types and historical noise response durations; Using the noise type as an index, perform homogeneous aggregation on the set of historical noise types, and perform mapping clustering on the set of historical noise monitoring data sequences and historical noise response durations according to the homogeneous aggregation result to obtain K sets of clustered historical noise monitoring data sequences and K sets of clustered historical noise response durations, where K is a positive integer; Traverse the K sets of clustered historical noise monitoring data sequences to extract noise monitoring features and obtain K sets of clustered historical noise monitoring features; Perform secondary aggregation on the K sets of clustered historical noise monitoring features, combine the K sets of clustered historical noise response durations, construct K anomaly monitoring and recognition trees, and summarize the K anomaly monitoring and recognition trees to generate the anomaly monitoring and recognition prototype library.

[0043] Furthermore, the sensor sub-array obtaining module 13 is used to perform the following steps: Extract a first set of clustered historical noise monitoring features from the K sets of clustered historical noise monitoring features; Randomly extract M first clustered historical noise monitoring features from the first set of clustered historical noise monitoring features as M initial recognition internal nodes; Use the M initial recognition internal nodes to aggregate the first set of clustered historical noise monitoring features to obtain M sets of aggregated historical noise monitoring features; Use the aggregation verification function to perform aggregation authentication on the M sets of aggregated historical noise monitoring features and the M first recognition internal nodes. If the authentication passes, use the M initial clustered historical noise monitoring features as the M recognition internal nodes, and perform monitoring scale marking on the M recognition internal nodes based on the corresponding first set of clustered historical noise response durations to obtain the first constructed anomaly monitoring and recognition tree; Analyze the K sets of clustered historical noise monitoring features and the K sets of clustered historical noise response durations to obtain the K anomaly monitoring and recognition trees.

[0044] Further, the aggregation verification function is as follows: ; wherein, is the aggregation degree output by the aggregation verification function, is the j-th aggregated historical noise monitoring feature in the i-th aggregated historical noise monitoring feature set among the M aggregated historical noise monitoring feature sets, is the first clustered historical noise monitoring feature corresponding to the i-th initial recognition internal node among the M aggregated historical noise monitoring feature sets, is the total number of aggregated historical noise monitoring features in the i-th aggregated historical noise monitoring feature set; Input the M aggregated historical noise monitoring feature sets and the M first recognition internal nodes into the aggregation verification function to obtain the aggregation degree; Judge whether the aggregation degree meets a preset aggregation degree threshold. If so, the authentication passes.

[0045] Further, the sensor sub-array obtaining module 13 is used to perform the following steps: Based on the one-to-one mapping relationship between the historical noise response duration and the historical noise monitoring feature, map and aggregate the first clustered historical noise response duration set based on the M aggregated historical noise monitoring feature sets to obtain M aggregated historical noise response duration sets; Traverse and calculate the mean values of the M aggregated historical noise response duration sets to obtain M monitoring scales, and identify the M recognition internal nodes based on the M monitoring scales to obtain the first abnormal monitoring recognition tree.

[0046] Further, the test result obtaining module 14 is used to perform the following steps: Obtain a plurality of sample noise monitoring data sequences, perform abnormal identification on the plurality of sample noise monitoring data sequences based on the abnormal monitoring scale sub-array to obtain a plurality of sample abnormal identification result arrays; Combine the plurality of sample noise monitoring data sequences with the plurality of sample abnormal identification result arrays respectively to construct an abnormal test analysis channel, and obtain the abnormal test analysis channel set.

[0047] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above describes specific embodiments of this specification. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0048] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

[0049] This specification and the drawings are merely exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. An intelligent test method for the vibration and noise impact of rail transit, characterized in that The method includes: Obtain a target test track, deploy sensors based on the route characteristics of the target test track, and obtain a deployed noise sensor array; Conduct a train passing test on the target test track according to a preset test plan, synchronously collect the data of the sensor array, and obtain a noise monitoring data sequence array; Traverse the noise monitoring data sequence array to perform anomaly monitoring and identification on the noise sensor array, and obtain an abnormal noise sensor sub-array and a normal noise sensor sub-array, where the abnormal noise sensor sub-array includes an anomaly monitoring scale sub-array; Construct an abnormal test analysis channel set based on the anomaly monitoring scale sub-array, and construct a normal test analysis network layer based on the normal monitoring frequency. Use the abnormal test analysis channel set and the normal test analysis network layer to perform online monitoring and analysis on the abnormal noise sensor sub-array and the normal noise sensor sub-array respectively, and obtain a target vibration noise test result.

2. The intelligent test method for the vibration and noise impact of rail transit according to claim 1, wherein Traverse the noise monitoring data sequence array to perform anomaly monitoring and identification on the noise sensor array, and obtain an abnormal noise sensor sub-array and a normal noise sensor sub-array, where the abnormal noise sensor sub-array includes an anomaly monitoring scale sub-array, including: Arbitrarily extract a first noise monitoring data sequence from the noise monitoring data sequence array; Extract noise monitoring feature of the first noise monitoring data sequence to obtain a first noise monitoring data feature set; In a pre-constructed anomaly monitoring and identification prototype library, match the first noise monitoring data feature set. If the match is successful, mark the noise sensor corresponding to the first noise monitoring data sequence as abnormal, add it to the abnormal noise sensor sub-array, and add the first anomaly monitoring scale obtained by the match to the anomaly monitoring scale sub-array.

3. The intelligent test method for the vibration and noise impact of rail transit according to claim 2, characterized in that Including: Obtain a set of historical noise monitoring data sequences, a corresponding set of historical noise types, and a set of historical noise response durations; Taking the noise type as an index, perform homogeneous aggregation on the set of historical noise types, and perform mapping clustering on the set of historical noise monitoring data sequences and the historical noise response durations according to the homogeneous aggregation result, to obtain K sets of clustered historical noise monitoring data sequences and K sets of clustered historical noise response durations, where K is a positive integer; Traverse the K sets of clustered historical noise monitoring data sequences to extract noise monitoring features, and obtain K sets of clustered historical noise monitoring feature sets; Perform secondary aggregation on the K sets of clustered historical noise monitoring feature sets, combine the K sets of clustered historical noise response durations, construct K anomaly monitoring and identification trees, and summarize the K anomaly monitoring and identification trees to generate the anomaly monitoring and identification prototype library.

4. The intelligent test method for the vibration and noise impact of rail transit according to claim 3, wherein Perform secondary aggregation on the K sets of clustered historical noise monitoring feature sets, combine the K sets of clustered historical noise response durations, and construct K anomaly monitoring and identification trees, including: Extract a first set of clustered historical noise monitoring features from the K sets of clustered historical noise monitoring feature sets; Randomly extract M first clustered historical noise monitoring features from the first set of clustered historical noise monitoring features as M initial recognition internal nodes; Aggregate the first clustering historical noise monitoring feature set by using the M initial recognition internal nodes to obtain M aggregated historical noise monitoring feature sets; Use an aggregation verification function to perform aggregation authentication on the M aggregated historical noise monitoring feature sets and the M first recognition internal nodes. If the authentication passes, use the M initial clustering historical noise monitoring features as the M recognition internal nodes, and perform monitoring scale identification on the M recognition internal nodes based on the corresponding first clustering historical noise response duration set to obtain the constructed first anomaly monitoring recognition tree; Analyze the K clustering historical noise monitoring feature sets and the K clustering historical noise response duration sets to obtain the K anomaly monitoring recognition trees.

5. The intelligent test method for vibration and noise impact of rail transit according to claim 4, wherein The aggregation verification function is: ； Among them, is the aggregation degree output by the aggregation verification function, is the j-th aggregated historical noise monitoring feature in the i-th aggregated historical noise monitoring feature set among the M aggregated historical noise monitoring feature sets, is the first clustering historical noise monitoring feature corresponding to the i-th initial recognition internal node among the M aggregated historical noise monitoring feature sets, is the total number of aggregated historical noise monitoring features in the i-th aggregated historical noise monitoring feature set; Input the M aggregated historical noise monitoring feature sets and the M first recognition internal nodes into the aggregation verification function to obtain an aggregation degree; Judge whether the aggregation degree meets a preset aggregation degree threshold. If so, the authentication passes.

6. The intelligent test method for the vibration and noise impact of rail transit according to claim 5, characterized in that It includes: Based on the one-to-one mapping relationship between the historical noise response duration and the historical noise monitoring feature, perform mapping aggregation on the first clustering historical noise response duration set based on the M aggregated historical noise monitoring feature sets to obtain M aggregated historical noise response duration sets; Traverse and calculate the mean values of the M aggregated historical noise response duration sets to obtain M monitoring scales, and perform identification on the M recognition internal nodes based on the M monitoring scales to obtain the first anomaly monitoring recognition tree.

7. The intelligent test method for the vibration and noise impact of rail transit according to claim 1, wherein Construct an anomaly test analysis channel set based on the anomaly monitoring scale sub-array, including: Obtain multiple sample noise monitoring data sequences, and perform anomaly recognition on the multiple sample noise monitoring data sequences based on the anomaly monitoring scale sub-array to obtain multiple sample anomaly recognition result arrays; Combine the multiple sample noise monitoring data sequences with the multiple sample anomaly recognition result arrays respectively to construct an anomaly test analysis channel to obtain the anomaly test analysis channel set.

8. An intelligent test system for the vibration and noise impact of rail transit, characterized in that, The system is used to implement the intelligent test method for the vibration and noise impact of rail transit according to any one of claims 1-7. The system includes: A noise sensor array acquisition module, configured to acquire a target test track, perform sensor layout based on the route characteristics of the target test track to obtain a completed noise sensor array; A monitoring data sequence array acquisition module, configured to perform train passing tests on the target test track according to a preset test plan, and synchronously collect data of the sensor array to obtain a noise monitoring data sequence array; A sensor sub-array acquisition module, configured to traverse the noise monitoring data sequence array to perform anomaly monitoring and recognition on the noise sensor array to obtain an abnormal noise sensor sub-array and a normal noise sensor sub-array, wherein the abnormal noise sensor sub-array includes an anomaly monitoring scale sub-array; A test result acquisition module is used to construct a set of abnormal test analysis channels based on an abnormal monitoring scale sub-array, and construct a conventional test analysis network layer based on a conventional monitoring frequency. The abnormal test analysis channel set and the conventional test analysis network layer are used to perform online monitoring and analysis on the abnormal noise sensor sub-array and the conventional noise sensor sub-array respectively, so as to obtain the target vibration noise test result.

Images