Bridge construction monitoring data abnormity identification method and system and related equipment

Abstract

The invention discloses a bridge construction monitoring data anomaly identification method and system and related equipment, and the method comprises the steps: firstly filtering an incoming data stream, then calculating and extracting a plurality of statistical feature values for the filtered data stream, forming a composition vector based on the plurality of statistical feature values, and forming a one-dimensional feature vector X; selecting a plurality of corresponding characteristic values from a plurality of statistical characteristic values of the determined characteristic vector X according to a preset standard to form a new characteristic vector Y; then, an SVDD model single-class support vector machine is adopted, a basic classification library is called, the feature vector Y after dimension reduction is recognized, and whether the feature vector Y belongs to abnormal data or not is judged; and finally, calling the basic classification library by adopting a binary-classification support vector machine based on an SMO algorithm, and re-judging the data which are judged to be abnormal. According to the scheme provided by the invention, high-precision construction monitoring abnormal data identification can be realized, the early warning efficiency is improved, and missing report and false report can be effectively avoided.

Description

Technical Field

[0001] The invention relates to engineering construction monitoring technology, and in particular to bridge construction monitoring technology. Background Art

[0002] At present, in the process of infrastructure construction, such as hydraulic climbing formwork for bridge piers, hanging basket casting of bridge beams and other construction environments, the current operating status inspection methods for most construction equipment are usually on-site feedback from construction personnel and regular inspections by management personnel, which have a certain lag. Once an emergency occurs, it is impossible to make timely feedback and implement corresponding emergency measures, which poses certain safety hazards.

[0003] The existing real-time safety monitoring system can effectively overcome the problems existing in manual monitoring. However, the existing real-time safety monitoring system also mostly issues warnings for certain numerical values ​​through manually set thresholds, which often leads to missed alarms or false alarms due to occasional abnormalities in construction equipment, environment or collection equipment. Summary of the invention

[0004] In view of the problems existing in the existing real-time safety monitoring scheme for bridge construction in terms of the accuracy and reliability of the monitoring structure, the purpose of the present invention is to provide a bridge construction monitoring data anomaly identification scheme, which is based on a joint machine learning model improved and optimized by a single-classification support vector machine and a binary-classification support vector machine to achieve high-precision construction monitoring anomaly data identification, effectively improve the early warning efficiency, and effectively avoid underreporting and false alarms.

[0005] In order to achieve the above object, the present invention provides a method for identifying abnormalities in bridge construction monitoring data, the identification method comprising:

[0006] (1) Data filtering: filtering the incoming data stream to remove the sharp data in the data stream;

[0007] (2) generating a feature vector, calculating and extracting a number of statistical feature values ​​from the data stream filtered in step (1), and forming a component vector based on the number of statistical feature values ​​to form a one-dimensional feature vector X;

[0008] (3) Data dimensionality reduction: a number of corresponding eigenvalues ​​of the eigenvector X determined in step (2) are selected according to a preset standard to form a new eigenvector Y;

[0009] (4) Initial analysis: using the SVDD model single-class support vector machine, calling the basic classification library, identifying the feature vector Y after dimensionality reduction, and determining whether it is abnormal data;

[0010] (5) Optimization analysis: using a binary support vector machine based on the SMO algorithm, calling the basic classification library, and re-judging the data judged as abnormal in step (4). If it is judged as abnormal data, it is finally judged as abnormal data; if it is judged as normal data, it is manually confirmed whether it is abnormal.

[0011] In some embodiments of the present invention, the wavelet denoising method is used to filter the data in step (1).

[0012] In some embodiments of the present invention, in step (3), the dimension reduction target features determined in the basic classification library are called to extract corresponding feature values.

[0013] In some embodiments of the present invention, in step (5), two point groups are formed in the vector space by calling all feature vectors Y1 formed by normal data and after dimensionality reduction, and all feature vectors Y2 formed by abnormal data and after dimensionality reduction in the basic classification library; on this basis, the feature vector Y to be classified is placed in the formed vector space, and its data type is determined according to which point group it is in.

[0014] In order to achieve the above object, the present invention provides a bridge construction monitoring data anomaly identification system, the identification system comprising:

[0015] A data filtering module, wherein the data filtering module is configured to filter the incoming data stream and remove sharp data in the data stream;

[0016] A feature vector generation module, wherein the feature vector generation module is configured to interact with the data filtering module and is capable of calculating and extracting a number of statistical feature values ​​for the filtered data stream, and forming a component vector based on the number of statistical feature values ​​to form a one-dimensional feature vector X;

[0017] A data dimension reduction module, wherein the data dimension reduction module is configured to be able to perform data interaction with the feature vector generation module, and to select a number of corresponding feature values ​​from a number of statistical feature values ​​in the feature vector X generated by the feature vector generation module according to a preset standard to form a new feature vector Y;

[0018] An initial analysis module, wherein the initial analysis module is configured to perform data interaction with the data dimension reduction module and the basic classification library, and can use the SVDD model single-class support vector machine to call the basic classification library to perform anomaly recognition on the feature vector Y after the dimension reduction processing by the data dimension reduction module;

[0019] The optimization analysis module is configured to interact with the initial analysis module and the basic classification library, and can use the binary support vector machine of the SMO algorithm to call the basic classification library to re-identify and determine the data determined to be abnormal by the initial analysis module: if it is determined to be abnormal data, it is finally determined to be abnormal data; if it is determined to be normal data, it is manually confirmed whether it is abnormal;

[0020] A basic classification library, wherein the basic classification library contains a number of feature vectors Y1 formed by normal data and after dimensionality reduction, and a number of feature vectors Y2 formed by abnormal data and after dimensionality reduction.

[0021] In some embodiments of the present invention, the data filtering module uses a wavelet denoising method to filter the data.

[0022] In some embodiments of the present invention, the data dimension reduction module calls the dimension reduction target features determined in the basic classification library to extract corresponding feature values.

[0023] In some embodiments of the present invention, the optimization analysis module forms two point groups in the vector space by calling all feature vectors Y1 formed by normal data and after dimensionality reduction in the basic classification library and all feature vectors Y2 formed by abnormal data and after dimensionality reduction; on this basis, the feature vector Y to be classified is placed in the formed vector space, and its data type is determined according to which point group it is in.

[0024] In order to achieve the above-mentioned object, the present invention also provides a computer-readable storage medium on which a program is stored, and when the program is executed by a processor, the steps of the above-mentioned bridge construction monitoring data anomaly identification method are implemented.

[0025] In order to achieve the above-mentioned object, the present invention also provides a processor, which is used to run a program, and when the program is running, the steps of the above-mentioned bridge construction monitoring data anomaly identification method are executed.

[0026] In order to achieve the above-mentioned purpose, the present invention also provides a terminal device, which includes a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program code is loaded and executed by the processor to implement the steps of the above-mentioned bridge construction monitoring data anomaly identification method.

[0027] In order to achieve the above-mentioned object, the present invention also provides a computer program product, which, when executed on a data processing device, is suitable for executing the steps of the above-mentioned bridge construction monitoring data anomaly identification method.

[0028] Compared with existing methods, the bridge construction monitoring data anomaly identification solution provided by the present invention can greatly improve the identification accuracy, reduce the misclassification rate, and effectively overcome the problems existing in the prior art.

[0029] The bridge construction monitoring data anomaly identification scheme provided by the present invention has a wide range of applications and can be promoted and applied in most bridge construction monitoring; specifically, it can adapt to various construction environments such as bridge hydraulic climbing formwork, hanging basket construction, beam jacking, and support casting; at the same time, it can generate considerable social and economic benefits in project construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0031] Figure 1 It is a basic flow chart of the bridge construction monitoring data anomaly identification method in the present invention;

[0032] Figure 2 This is a system principle diagram of the bridge construction monitoring data anomaly identification system in the present invention;

[0033] Figure 3 This is a schematic diagram of the initial analysis performed in the present invention;

[0034] Figure 4 A schematic diagram of the optimization analysis performed in the present invention;

[0035] Figure 5 The present invention is a flowchart for specifically implementing the method for identifying abnormalities in bridge construction monitoring data. DETAILED DESCRIPTION

[0036] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further described below with reference to specific diagrams.

[0037] In view of the characteristics of bridge construction safety monitoring data, the scheme of the present invention forms an improved and optimized joint machine learning algorithm based on single-classification support vector machine and binary-classification support vector machine, thereby realizing high-precision construction monitoring abnormal data identification, thereby improving early warning efficiency and effectively avoiding underreporting and false alarms.

[0038] On this basis, the solution of the present invention also innovatively incorporates a dimensionality reduction method and an outlier dynamic warehousing method into the joint machine learning algorithm, thereby further improving the detection accuracy.

[0039] The solution of the present invention can effectively overcome the shortcomings of the single classification method, which lacks data dimension reduction optimization means, has a high false warning rate, and the binary classification method has too high requirements for the basic training database, thereby achieving high-precision recognition and warning.

[0040] It should be noted that the bridge construction monitoring data targeted by the scheme of the present invention involves various data during the bridge construction process. Taking the high-altitude hanging basket casting of the bridge as an example, it mainly includes: suspension rod tension, main rod stress, on-site temperature and humidity, wind speed, acceleration, inclination angle, etc.

[0041] In this regard, the present invention provides a method for identifying abnormalities in bridge construction monitoring data, such as Figure 1 As shown, the abnormality identification method is specifically composed of the following steps:

[0042] S1: Data filtering.

[0043] This step filters the incoming bridge construction monitoring data stream and removes extremely rare sharp data in the data stream, thereby reducing the noise in the data and improving the accuracy of subsequent abnormal data identification.

[0044] S2: Generate feature vector.

[0045] This step calculates and extracts a number of statistical eigenvalues ​​for a segment of data stream that has been filtered in step S1, and forms a component vector based on the statistical eigenvalues ​​to form a one-dimensional feature vector X; each statistical eigenvalue in the one-dimensional feature vector X formed here reflects a certain characteristic corresponding to the data in a specific dimension. Based on this, the segment of data stream that has been filtered is converted into a one-dimensional feature vector X.

[0046] Since a data stream is composed of dozens, hundreds or even thousands of values, in order to improve the accuracy and efficiency of subsequent data processing, this step converts it into a small number of feature vector values ​​through eigenvalue extraction and statistics, thereby representing this data stream from different angles.

[0047] Each value of the one-dimensional feature vector X generated here is a statistical feature value of this segment of data flow, including the average value x1, maximum value x2, minimum value x3, peak-to-peak value x4, rectified mean x5, standard deviation x6, skewness x7, kurtosis x8, root mean square x9, waveform index x 10 , peak index x 11 , Pulse index x 12 , frequency center x 13 , RMS frequency x 14 , frequency standard deviation x 15 .

[0048] The one-dimensional feature vector X thus formed represents this data stream from different angles through these 15 values.

[0049] S3: Data dimensionality reduction.

[0050] In this step, several corresponding eigenvalues ​​of the eigenvalues ​​of the eigenvector X determined in step (2) are selected according to a preset standard to form a new eigenvector Y.

[0051] Since the feature vector X generated in step (2) includes multiple feature vector values, in order to be more suitable for the current project, the most suitable feature values ​​are selected through dimensionality reduction according to the current project, and the feature values ​​with poor discrimination are eliminated, thereby forming a new vector Y.

[0052] S4: Initial analysis.

[0053] In this step, the SVDD model single-class support vector machine is used to call the basic classification library to identify the feature vector Y after the dimension reduction in step S3 to determine whether it is abnormal data: if it is determined to be abnormal data, it goes to the next section.

[0054] S5: Optimization analysis.

[0055] In this step, a binary classification support vector machine of the SMO algorithm is used to call the basic classification library to re-judge the data determined to be abnormal in step S4: if it is determined to be abnormal data, it is finally determined to be abnormal data; if it is determined to be normal data, it is manually confirmed whether it is abnormal.

[0056] This step further optimizes the analysis based on the basic classification library on the basis of the initial analysis in step (4), effectively overcoming the shortcomings of the single classification method, which lacks data dimension reduction optimization means, has a high false alarm rate, and the binary classification method has high requirements for the basic training database, thereby achieving high-precision recognition and warning.

[0057] At the same time, the data finally judged to be abnormal will be entered into the basic classification library for subsequent abnormality identification.

[0058] As a further explanation, the present abnormality identification method preferably uses a wavelet denoising method to filter data in step S1. Here, the wavelet denoising method is used to efficiently filter data such as wind speed, acceleration, and dynamic inclination angle to ensure filtering effect.

[0059] As a further explanation, in step S3 of the anomaly identification method, the corresponding statistical features selected as the target features of dimensionality reduction in the feature vector X are preferably determined when the basic classification library is established to ensure the efficiency of the subsequent dimensionality reduction processing.

[0060] As a further explanation, when the anomaly recognition method performs initial analysis based on the SVDD model single-class support vector machine in step S4, the corresponding feature vector is specifically mapped to a high-dimensional feature space, and a hypersphere surrounding the target vector is found in the high-dimensional space.

[0061] At the same time, the hypersphere formed in this step meets the following two requirements: (1) the feature vector is enclosed as much as possible; (2) the hypersphere encloses a range as small as possible.

[0062] As a further explanation, this anomaly identification method also has a basic classification library establishment step. The established basic classification library specifically includes a number of feature vectors Y1 formed by normal data and after dimensionality reduction, and also includes a number of feature vectors Y2 formed by abnormal data and after dimensionality reduction, so as to provide data category judgment for steps S4 and S5.

[0063] The abnormality identification method for bridge construction monitoring data provided by the present invention can be used as a corresponding software program to form a corresponding abnormality identification system for bridge construction monitoring data. When the software program is running, it will execute the above-mentioned abnormality identification method for bridge construction monitoring data and store it in a corresponding storage medium for the processor to retrieve and execute.

[0064] See also Figure 2 The bridge construction monitoring data anomaly identification system 100 formed in this way mainly includes six functional modules: data filtering module 110, feature vector generation module 120, data dimension reduction module 130, initial analysis module 140, optimization analysis module 150 and basic classification library 160.

[0065] The data filtering module 110 in the system is configured to filter the incoming bridge construction monitoring data stream and remove extremely individual sharp data in the data stream.

[0066] Furthermore, the data filtering module 110 is specifically configured to perform data filtering using a wavelet noise reduction method.

[0067] The feature vector generation module 120 in the present system is configured to interact with the data filtering module 110, and can calculate and extract a number of statistical feature values ​​for the data stream filtered by the data filtering module 110, and form a component vector based on the extracted statistical feature values ​​to form a one-dimensional feature vector X. Thus, the feature vector generation module 120 can convert a filtered data stream into a one-dimensional feature vector X, and each statistical feature value in the vector reflects a certain characteristic of the data in a specific dimension.

[0068] The data dimension reduction module 130 in the system is configured to be able to interact with the feature vector generation module 120 for data, and to select several corresponding feature values ​​from several statistical feature values ​​in the feature vector X generated by the feature vector generation module 120 according to a preset standard to form a new feature vector Y.

[0069] Furthermore, the statistical features selected by the data dimensionality reduction module 130 from the feature vector X as the dimensionality reduction target features are determined when the basic classification library is established.

[0070] The initial analysis module 140 in this system is configured to interact with the data dimension reduction module 130 and the basic classification library 160, and can use the SVDD model single-class support vector machine to call the basic classification library to perform anomaly identification on the feature vector Y after dimension reduction processing by the data dimension reduction module to determine whether it is abnormal data: if it is determined to be abnormal data, it will enter the next section.

[0071] Furthermore, when the initial analysis module 140 performs initial analysis based on the SVDD model single-class support vector machine, it specifically maps the corresponding feature vector to a high-dimensional feature space, and finds a hypersphere that surrounds the target vector in the high-dimensional space. At the same time, the hypersphere formed meets the following two requirements: 1. The feature vector is surrounded as much as possible; 2. The hypersphere surrounds the smallest possible range.

[0072] The optimization analysis module 150 in this system is configured to interact with the initial analysis module 140 and the basic classification library 160 for data, and can utilize the binary support vector machine of the SMO algorithm to call the basic classification library to re-identify and determine the data determined to be abnormal by the initial analysis module: if it is determined to be abnormal data, it will be finally determined to be abnormal data; if it is determined to be normal data, it will be manually confirmed whether it is abnormal; the data finally determined to be abnormal will be entered into the basic classification library for subsequent abnormality identification.

[0073] The basic classification library 160 in this system is specifically configured to include a number of feature vectors Y1 formed by normal data and after dimensionality reduction, and also includes a number of feature vectors Y2 formed by abnormal data and after dimensionality reduction; at the same time, the basic classification library 160 is also configured to be able to interact with the initial analysis module 140 and the optimization analysis module 150 for the two analysis modules to perform data category determination.

[0074] With respect to the bridge construction monitoring data anomaly identification solution provided by the present invention, the implementation process and corresponding technical features of the solution of the present invention are explained below through specific application examples.

[0075] When the bridge construction monitoring data anomaly identification scheme provided by the present invention is implemented, a corresponding bridge construction monitoring data anomaly identification system 100 is constructed based on the above scheme, such as Figure 2 shown.

[0076] The data filtering module 110 in the system performs filtering processing on the bridge construction monitoring data stream.

[0077] The monitoring data here is a one-dimensional signal, so the data filtering module 110 uses a wavelet noise reduction method to perform data filtering.

[0078] Specifically, the process of wavelet denoising by the data filtering module 110 is as follows:

[0079] (1) Select a suitable wavelet basis and determine the decomposition level, and then perform wavelet decomposition on the signal.

[0080] (2) For the high-frequency coefficients of each layer, select an appropriate threshold to perform hard threshold or soft threshold filtering.

[0081] (3) Reconstruct the one-dimensional signal based on the deepest coefficients of wavelet decomposition and the high-frequency coefficients of each layer after processing.

[0082] Furthermore, in order to ensure the data processing efficiency of real-time noise reduction, the wavelet threshold noise reduction method is adopted in this example, which can further improve the operation efficiency of the algorithm on the existing equipment while ensuring the noise reduction effect.

[0083] Threshold denoising sets the high-frequency coefficients of wavelet decomposition that are less than the threshold to 0. Fixed threshold:

[0084]

[0085] Among them, N represents the length of the wavelet coefficient of this layer, σ represents the wavelet noise intensity of this layer, and the general calculation formula is:

[0086]

[0087] d i (t) represents the wavelet decomposition coefficient of the i-th layer.

[0088] The feature vector generation module 120 in the system specifically adopts the following 15 statistical feature values: average value x1, maximum value x2, minimum value x3, peak-to-peak value x4, rectified mean x5, standard deviation x6, skewness x7, kurtosis x8, root mean square x9, waveform index x1, peak value x2, peak value x3, peak value x4, rectified mean x5, standard deviation x6, skewness x7, kurtosis x8, root mean square x9, waveform index x1, peak value x1, peak value x1, peak value x2, peak value x3, peak value x4, peak value ... peak value x6, peak value x7, peak value x8, peak value x9, peak value x10, peak value 10 , peak index x 11 , Pulse index x 12 , frequency center x 13 , RMS frequency x 14 , frequency standard deviation x 15 .

[0089] After calculating each statistical eigenvalue, the vector X=[x1,x2,x3,x4,x5,x6,x7,x8,x9,x 10 ,x 11 ,x 12 ,x 13 ,x 14 ,x15 ], which is the feature vector used for subsequent identification. This enables each statistical feature value in the vector to reflect a certain characteristic of the data in a specific dimension.

[0090] The data dimension reduction module 130 in the system specifically selects corresponding statistical features as dimension reduction target features according to the feature vectors determined when the basic classification library is established. The subsequent data dimension reduction to be determined only needs to be extracted according to the determined dimension reduction target features.

[0091] Furthermore, in this example, when establishing the basic classification library, the dimension reduction target features are determined, and the determination method is shown as follows:

[0092] Convert 10 to 50 data streams into 10 to 50 feature vectors, and use these feature vectors as column vectors to form a feature matrix:

[0093]

[0094] Where: x ji represents the i-th eigenvalue of the j-th random variable.

[0095] The basic calculation process of the univariate feature selection algorithm is as follows:

[0096] (1) Data standardization

[0097]

[0098] in: The sample mean of each row of the matrix.

[0099] S i : The sample standard deviation of each row of the matrix.

[0100] (2) Set the dimension reduction target and select the evaluation function

[0101] An F test or a chi-square test is performed on each eigenvalue, and its separability score is obtained according to the two evaluation functions.

[0102] Then, according to the set number of dimension reduction target features, several statistical eigenvalues ​​with high scores are selected to form the eigenvector Y.

[0103] (3) Composition of feature vectors

[0104] According to the selected eigenvalue, the new eigenvector Y is used to replace the original eigenvector X for output, and the names of the selected eigenvalues ​​are output in sequence.

[0105] When the initial analysis module 140 in the system performs initial analysis, Figure 3As shown, all feature vectors Y1 formed by normal data and after dimension reduction in the basic classification library are specifically called, which form a minimum hypersphere in the vector space; the feature vector Y to be classified is placed in this vector space, and its data type is determined according to whether it is inside the hypersphere:

[0106] (1) Y is inside the hypersphere: normal data;

[0107] (2) Y is outside the hypersphere: abnormal data.

[0108] See also Figure 3 In the figure, the black dot represents Y1, the circle represents the smallest hypersphere composed of all Y1, and the white dot represents the feature vector Y to be classified.

[0109] The optimization analysis module 150 in the system is used to make further determinations at this stage on the data determined to be abnormal by the initial analysis (ie, the abnormal feature vector Y).

[0110] Combination Figure 4 , call all the feature vectors Y1 formed by normal data and after dimension reduction in the basic classification library, and all the feature vectors Y2 formed by abnormal data and after dimension reduction, to form two point groups in the vector space. Place the feature vector Y to be classified in this vector space, and determine its data type according to which point group it is in:

[0111] (1) Y is in the Y1 point group: normal data;

[0112] (2) Y is in the Y2 point group: it is abnormal data.

[0113] See also Figure 4 In the figure, the square dot represents Y1, the circle represents Y2, and the solid line represents the optimal hyperplane. If the feature vector Y to be classified is on the side of the solid line close to Y1, it belongs to the Y1 point group, otherwise it belongs to the Y2 point group.

[0114] Here, the wind speed monitoring data during bridge construction is taken as an example to specifically explain the process of abnormal data identification of the wind speed monitoring data during bridge construction formed by the example.

[0115] like Figure 5 As shown, the bridge construction monitoring data anomaly identification system 100 processes a section of wind speed monitoring data stream input as follows:

[0116] (1) First, determine whether it is used for abnormality identification and warning or for establishing a basic classification library.

[0117] (2) If used to establish a basic classification library, 15 eigenvalues ​​are calculated to form a eigenvector X. All eigenvectors in the library are combined, and several eigenvalues ​​with the best separability are selected to form a new eigenvector Y. Then the incoming data is calibrated to indicate whether it is normal data or abnormal data, and stored in the basic classification library.

[0118] (3) If used for abnormality identification and early warning, the data is restricted and filtered, and then the feature vector X is generated in the same way. According to several feature values ​​determined in the basic classification library, the dimension is reduced to generate a new feature vector Y.

[0119] (4) Generate a new feature vector Y for dimensionality reduction and perform an initial analysis to see if the data is abnormal. If not, the process ends directly. If the preliminary analysis shows that the data is abnormal, proceed to step (5).

[0120] (5) Based on the optimization analysis of the data determined to be abnormal:

[0121] (5.1) If it is determined to be abnormal again, it will enter the subsequent warning process, be entered into the basic classification library and marked as abnormal data.

[0122] (5.2) If it is judged to be normal again, it will enter the manual identification process to determine the final result.

[0123] From the above example, it can be seen that the bridge construction monitoring data anomaly identification scheme given in this paper is based on the joint machine learning model improved and optimized by single-classification support vector machine and binary-classification support vector machine, which can realize high-precision construction monitoring abnormal data identification, improve early warning efficiency, and effectively avoid underreporting and false alarms.

[0124] Based on the above-mentioned bridge construction monitoring data anomaly identification scheme, an embodiment of the present invention further provides a computer-readable storage medium on which a program is stored. When the program is executed by a processor, the steps of the above-mentioned bridge construction monitoring data anomaly identification method are implemented.

[0125] An embodiment of the present invention further provides a processor, which is used to run a program, wherein the program executes the steps of the above-mentioned bridge construction monitoring data anomaly identification method when running.

[0126] An embodiment of the present invention also provides a terminal device, which includes a processor, a memory, and a program stored in the memory and executable on the processor, wherein the program code is loaded and executed by the processor to implement the steps of the above-mentioned bridge construction monitoring data anomaly identification method.

[0127] The present invention also provides a computer program product, which, when executed on a data processing device, is suitable for executing the steps of the above-mentioned bridge construction monitoring data anomaly identification method.

[0128] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0129] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0130] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0131] The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products of the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0132] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0133] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0134] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0135] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0136] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0137] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0138] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0139] The above-mentioned method of the present invention, or a specific system unit, or a part of the same, is a pure software architecture, which can be arranged on a physical medium, such as a hard disk, an optical disk, or any electronic device (such as a smart phone, a computer-readable storage medium) through program code. When the machine loads the program code and executes it (such as a smart phone loading and executing it), the machine becomes a device for implementing the present invention. The above-mentioned method and device of the present invention can also be transmitted in the form of program code through some transmission media, such as cables, optical fibers, or any transmission mode. When the program code is received, loaded and executed by a machine (such as a smart phone), the machine becomes a device for implementing the present invention.

[0140] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for identifying abnormalities in bridge construction monitoring data, characterized in that: The identification method comprises: (1) Data filtering: filtering the incoming data stream to remove the sharp data in the data stream; (2) generating a feature vector, calculating and extracting a number of statistical feature values ​​from the data stream filtered in step (1), and forming a component vector based on the number of statistical feature values ​​to form a one-dimensional feature vector X; (3) Data dimensionality reduction: a number of corresponding eigenvalues ​​of the eigenvector X determined in step (2) are selected according to a preset standard to form a new eigenvector Y; (4) Initial analysis: using the SVDD model single-class support vector machine, calling the basic classification library, identifying the feature vector Y after dimensionality reduction, and determining whether it is abnormal data; (5) Optimization analysis: using a binary support vector machine based on the SMO algorithm, calling the basic classification library, and re-judging the data judged as abnormal in step (4). If it is judged as abnormal data, it is finally judged as abnormal data; if it is judged as normal data, it is manually confirmed whether it is abnormal.

2. The bridge construction monitoring data anomaly identification method according to claim 1 is characterized in that: In the step (1), the wavelet noise reduction method is used to filter the data.

3. The bridge construction monitoring data anomaly identification method according to claim 1 is characterized in that: In the step (3), the dimension reduction target features determined in the basic classification library are called to extract the corresponding feature values.

4. The bridge construction monitoring data anomaly identification method according to claim 1 is characterized in that: In the step (5), two point groups are formed in the vector space by calling all the feature vectors Y1 formed by normal data and after dimensionality reduction, and all the feature vectors Y2 formed by abnormal data and after dimensionality reduction in the basic classification library; on this basis, the feature vector Y to be classified is placed in the formed vector space, and its data type is determined according to which point group it is in.

5. A bridge construction monitoring data anomaly identification system, characterized in that: The identification system comprises: A data filtering module, wherein the data filtering module is configured to filter the incoming data stream and remove sharp data in the data stream; A feature vector generation module, wherein the feature vector generation module is configured to interact with the data filtering module and is capable of calculating and extracting a number of statistical feature values ​​for the filtered data stream, and forming a component vector based on the number of statistical feature values ​​to form a one-dimensional feature vector X; A data dimension reduction module, wherein the data dimension reduction module is configured to be able to perform data interaction with the feature vector generation module, and to select a number of corresponding feature values ​​from a number of statistical feature values ​​in the feature vector X generated by the feature vector generation module according to a preset standard to form a new feature vector Y; An initial analysis module, wherein the initial analysis module is configured to perform data interaction with the data dimension reduction module and the basic classification library, and can use the SVDD model single-class support vector machine to call the basic classification library to perform anomaly recognition on the feature vector Y after the dimension reduction processing by the data dimension reduction module; The optimization analysis module is configured to interact with the initial analysis module and the basic classification library, and can use the binary support vector machine of the SMO algorithm to call the basic classification library to re-identify and determine the data determined to be abnormal by the initial analysis module: if it is determined to be abnormal data, it is finally determined to be abnormal data; if it is determined to be normal data, it is manually confirmed whether it is abnormal; A basic classification library, wherein the basic classification library contains a number of feature vectors Y1 formed by normal data and after dimensionality reduction, and a number of feature vectors Y2 formed by abnormal data and after dimensionality reduction.

6. The bridge construction monitoring data anomaly identification system according to claim 5 is characterized in that: The data filtering module uses a wavelet noise reduction method to filter the data.

7. The bridge construction monitoring data anomaly identification system according to claim 5 is characterized in that: The data dimension reduction module calls the dimension reduction target features determined in the basic classification library to extract corresponding feature values.

8. The bridge construction monitoring data anomaly identification system according to claim 5 is characterized in that: In the optimization analysis module, all feature vectors Y1 formed by normal data and after dimensionality reduction and all feature vectors Y2 formed by abnormal data and after dimensionality reduction in the basic classification library are called to form two point groups in the vector space; on this basis, the feature vector Y to be classified is placed in the formed vector space, and its data type is determined according to which point group it is in.

9. A terminal device, comprising a processor, a memory, and a program stored in the memory and executable on the processor, characterized in that: The program code is loaded and executed by the processor to implement the steps of the bridge construction monitoring data anomaly identification method according to any one of claims 1 to 4.

10. A computer program product, characterized in that When executed on a data processing device, it is suitable for executing the steps of the bridge construction monitoring data anomaly identification method as described in any one of claims 1-4.

Images