Bridge monitoring sensor fault identification method and system

Abstract

The invention relates to the technical field of bridge structure health monitoring, in particular to a bridge monitoring sensor fault identification method and system. The method comprises the following steps: acquiring monitoring data of a bridge monitoring sensor; constructing a sensor mutual prediction model, and obtaining a first prediction value of the bridge monitoring sensor by using the sensor mutual prediction model; analyzing historical monitoring data of the bridge monitoring sensor, and reconstructing a data reconstruction value of the bridge monitoring sensor according to an analysis result; and recognizing the working state of the bridge monitoring sensor in combination with the first predicted value and the data reconstruction value. According to the method, the predicted value and the data reconstruction value of the monitoring data are obtained based on the historical data of the sensor, then the credible space of the monitoring data is constructed to evaluate whether the monitoring data is abnormal or not, the problem of efficiently and accurately recognizing the fault sensor is solved, and judgment of the bridge health monitoring system on the bridge state is facilitated.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge structure health monitoring, and particularly relates to a method and system for identifying faults of bridge monitoring sensors. Background Art

[0002] As an important transportation facility, bridges play a decisive role in social development and economic growth. However, most bridges are located in harsh natural environments and are often threatened by various natural disasters such as floods, strong winds, earthquakes, and landslides. To ensure the long-term stable operation of bridges, a method of using a health monitoring system to monitor the structural parameters of key parts of bridges, detect potential problems early, and take corresponding maintenance and reinforcement measures is widely used. The key to mastering the real-time operation status of bridges through this method lies in ensuring that the monitoring system can stably collect effective data. The normal operation of sensors is directly related to the accuracy and reliability of bridge monitoring data, which in turn affects the assessment and decision-making of the bridge structure's health status. However, in actual applications, sensors are easily affected by environmental factors such as high temperature, humidity, and corrosion, resulting in failures or aging, and the monitoring system collects invalid data that does not match the true output of the bridge, thus making it impossible to accurately understand the actual situation of the bridge or making misjudgments about the bridge operation status. Therefore, it is crucial to detect sensor anomalies in a timely manner.

[0003] Current methods for detecting abnormal data of bridge monitoring sensors include supervised learning methods, which have the following limitations: (1) High demand for data labeling, requiring a large amount of manually labeled data for model training. However, abnormal data during bridge monitoring is usually much less than normal data, resulting in high data labeling costs; moreover, the scarcity of abnormal data compared to normal data causes an imbalanced classification problem, which affects the accuracy and robustness of the model. (2) Insufficient generalization ability, the distribution of monitoring data of different bridges varies significantly, and a trained model is difficult to generalize to other bridges, which means that each bridge needs to be separately labeled and trained, consuming time and effort. Facing a large number of bridges and massive data, manual labeling is almost impossible to complete, restricting the widespread application of existing methods in actual projects.

[0004] On the other hand, existing methods for identifying faults of bridge monitoring sensors mainly rely on single characteristic parameters or simple statistical analysis, such as setting thresholds to determine whether sensor data is abnormal. However, this method often ignores the spatial correlation and time series characteristics between sensor data, resulting in low accuracy and reliability of fault identification.

[0005] Therefore, finding a method and system for efficiently and accurately identifying faulty sensors is an urgent problem to be solved at present. Summary of the Invention

[0006] In view of the deficiencies of existing methods and the requirements of practical applications, in order to improve the ability to identify faulty sensors in a bridge monitoring system. On the one hand, the present invention provides a method for identifying faults in bridge monitoring sensors, comprising the following steps: obtaining monitoring data of the bridge monitoring sensors; constructing a mutual prediction model for the sensors, and obtaining a first predicted value of the bridge monitoring sensors by using the mutual prediction model for the sensors; analyzing the historical monitoring data of the bridge monitoring sensors, and reconstructing a data reconstruction value of the bridge monitoring sensors according to the analysis result; combining the first predicted value and the data reconstruction value to identify the working state of the bridge monitoring sensors.

[0007] Based on the historical data of the sensors, the present invention obtains the predicted value and the data reconstruction value of the monitoring data, and then constructs a credible space for the monitoring data to evaluate whether the monitoring data is abnormal, solving the problem of efficiently and accurately identifying faulty sensors, which is beneficial to the judgment of the bridge state by the bridge health monitoring system.

[0008] Optionally, the method for identifying faults in bridge monitoring sensors further comprises the following steps: Constructing a data denoising model; using the data denoising model to perform denoising processing on the monitoring data. The present invention performs denoising processing on the monitoring data through the data denoising model, eliminating and suppressing the influence of noise, which is beneficial to improving the accuracy and availability of the data.

[0009] Optionally, the constructing of the mutual prediction model for the sensors comprises the following steps: Using an improved dung beetle algorithm to obtain the optimal parameters of the machine learning network; constructing a mutual prediction model for the sensors based on the optimal parameters. The present invention optimizes the optimal parameters of the machine learning network through the improved dung beetle algorithm, which is beneficial to quickly and accurately obtaining the prediction result and further improving the identification accuracy of the present invention.

[0010] Optionally, the improvement of the dung beetle algorithm comprises the following steps: Decomposing the optimal foraging area of the small dung beetles, and adjusting the foraging direction of the small dung beetles according to the decomposition result; adjusting the number of stealing dung beetles according to the optimization result based on the number of iterations. The present invention improves the local optimization efficiency and global optimization ability of the dung beetle algorithm by adjusting the foraging direction of the small dung beetles and adjusting the number of stealing dung beetles, which is further beneficial to improving the reliability of the first predicted value.

[0011] Optionally, the analyzing of the historical monitoring data of the bridge monitoring sensors and reconstructing the data reconstruction value of the bridge monitoring sensors according to the analysis result comprises the following steps: Set the sequence window length of the historical monitoring data; based on the sequence window length, fit the first linear function of the historical monitoring data under the corresponding window length; combine the first linear function and the historical monitoring data to obtain the calculated value of the first linear function of the corresponding window monitoring data; evaluate the calculated value of the first linear function, and reconstruct the data reconstruction value of the bridge monitoring sensor according to the evaluation result. The present invention reconstructs the monitoring data by using the calculated values of the first linear functions under multiple windows, which is beneficial to timely detect abnormal data and improves the recognition efficiency of the present invention.

[0012] Optionally, the evaluating the calculated value of the first linear function and reconstructing the data reconstruction value of the bridge monitoring sensor according to the evaluation result includes the following steps: Set a difference threshold; compare the difference threshold with the difference value between the calculated value of the first linear function and the historical monitoring data to obtain an evaluation result; fit a second linear function according to the evaluation result, and reconstruct the data reconstruction value of the bridge monitoring sensor through the second linear function. The present invention screens out abnormal data through the difference threshold, which is further beneficial to constructing an accurate and effective credible space in the subsequent steps.

[0013] Optionally, the setting the difference threshold includes the following steps: Calculate the actual difference value between the calculated value of the first linear function and the historical monitoring data; based on the percentile method, set the difference threshold according to the actual difference value. The present invention dynamically sets the difference threshold, reduces the influence of human factors, and improves the adaptability of the present invention.

[0014] Optionally, the combining the first predicted value and the data reconstruction value to identify the working state of the bridge monitoring sensor includes the following steps: Obtain a first residual sequence according to the first predicted value and the monitoring data; obtain a second residual sequence according to the first predicted value and the data reconstruction value; combine the first residual sequence and the second residual sequence to establish a confidence interval; identify the working state of the bridge monitoring sensor according to the confidence interval. The present invention constructs a confidence interval through the residual sequences between the monitoring data, the predicted data, and the reconstructed data, eliminating various error relationships or unnecessary correlation relationships.

[0015] Optionally, the combining the first residual sequence and the second residual sequence to establish a confidence interval includes the following steps: Calculate the mean and standard deviation of the first residual sequence and the mean absolute deviation of the second residual sequence; obtain a confidence space according to the mean, the standard deviation, and the mean absolute deviation; and establish a confidence interval in combination with the confidence space, the first predicted value, and the data reconstruction value. The confidence space obtained by the present invention is objective and accurate, effectively restricting the credible range of the monitoring data, and further facilitating the improvement of the recognition accuracy of the present invention.

[0016] In a second aspect, in order to efficiently execute a method for identifying faults in bridge monitoring sensors provided by the present invention, the present invention further provides a system for identifying faults in bridge monitoring sensors, including a processor, an input device, an output device, and a memory. The processor, the input device, the output device, and the memory are interconnected. Among them, the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute a method for identifying faults in bridge monitoring sensors as described in the first aspect of the present invention. The system for identifying faults in bridge monitoring sensors of the present invention has a compact structure and stable performance, and can stably execute a method for identifying faults in bridge monitoring sensors provided by the present invention, further improving the overall applicability and practical application ability of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of a method for identifying faults in bridge monitoring sensors provided by an embodiment of the present invention; Figure 2 It is a framework diagram of a system for identifying faults in bridge monitoring sensors provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described here are only for illustrative purposes and are not used to limit the present invention. In the following description, in order to provide a thorough understanding of the present invention, a large number of specific details are set forth. However, it is obvious to those of ordinary skill in the art that the present invention does not have to employ these specific details. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the present invention.

[0019] Throughout the specification, references to "an embodiment", "embodiments", "an example", or "examples" mean that the specific features, structures, or characteristics described in connection with the embodiment or example are included in at least one embodiment of the present invention. Thus, the phrases "in an embodiment", "in embodiments", "an example", or "examples" that appear throughout the specification do not necessarily all refer to the same embodiment or example. In addition, the specific features, structures, or characteristics may be combined in any suitable combination and / or sub-combination in one or more embodiments or examples. Furthermore, those of ordinary skill in the art should understand that the diagrams provided herein are for illustrative purposes only and are not necessarily drawn to scale.

[0020] Please refer to Figure 1 , in order to improve the ability to identify faulty sensors in a bridge monitoring system. The present invention provides a method for identifying faults in bridge monitoring sensors. As Figure 1 shown, in one embodiment, the method includes the following steps: S1. Obtain the monitoring data of the bridge monitoring sensors.

[0021] In an embodiment, by arranging various types of sensors such as strain sensors, acceleration sensors, and displacement sensors at key parts of the bridge, such as the main girder, piers, and bearings, each sensor collects the response data of the bridge structure in real time according to the set sampling frequency, such as strain values, acceleration values, displacement values, etc. At the same time, the working state parameters of the sensors are collected, such as the supply voltage and the temperature of the sensor itself.

[0022] Further, after obtaining the monitoring data of the bridge monitoring sensors, the monitoring data is further subjected to noise reduction processing.

[0023] Specifically, the noise reduction processing of the monitoring data includes the following steps: S11. Construct a data noise reduction model.

[0024] In an embodiment, the data noise reduction model is constructed based on the wavelet denoising algorithm. The core idea of the wavelet denoising algorithm is to decompose the noisy signal into the wavelet domain of different scales (frequencies) by using wavelet transform. Since the wavelet coefficients of noise and effective signals have different characteristics at different scales, by processing the wavelet coefficients (such as threshold processing), the wavelet coefficients corresponding to noise can be suppressed, while the wavelet coefficients corresponding to effective signals are retained or enhanced. Finally, the signal is reconstructed by inverse wavelet transform to achieve the purpose of denoising.

[0025] Specifically, according to the characteristics of the data, a suitable wavelet basis and decomposition level are selected, the data is subjected to multi-layer wavelet decomposition, the high-frequency components where the noise is located are removed, and then wavelet reconstruction is performed to obtain the denoised data.

[0026] In some other embodiments, other noise reduction algorithms such as the Kalman filter algorithm or the Fourier transform denoising algorithm can also be used to construct a data noise reduction model to achieve the purpose of reducing noise in the data.

[0027] Furthermore, it is also necessary to perform normalization processing on the denoised data to map the data to the range of [0, 1]. The normalization processing satisfies the following formula:

[0028] where represents the value after normalization processing, represents the value before normalization processing, represents the maximum value before normalization processing, represents the minimum value before normalization processing.

[0029] S12. Use the data noise reduction model to perform noise reduction processing on the monitoring data.

[0030] Specifically, according to the constructed data noise reduction model, appropriate parameters are set. For example, for the wavelet denoising algorithm, a wavelet basis function and a threshold need to be selected; for the Kalman filter algorithm, a state transition matrix and a noise covariance matrix need to be defined.

[0031] Furthermore, apply the data noise reduction model to the monitoring data, such as performing wavelet transform on the data, performing threshold processing on the wavelet coefficients, and then reconstructing the signal.

[0032] Furthermore, draw a comparison chart of the original data and the denoised data to visually evaluate the denoising effect; the signal-to-noise ratio (SNR), mean square error (MSE) and other indicators can also be used to quantitatively evaluate the denoising effect. According to the evaluation results, adjust the algorithm parameters to improve the denoising effect. If the effect is not ideal, other algorithms can be tried or multiple algorithms can be combined.

[0033] S2. Construct a mutual prediction model for sensors, and use the mutual prediction model for sensors to obtain the first predicted value of the bridge monitoring sensors.

[0034] In the embodiment, the construction of the mutual prediction model for sensors includes the following steps: S21. Use the improved dung beetle algorithm to obtain the optimal parameters of the machine learning network.

[0035] Specifically, the machine learning network refers to one or an ensemble learning model of machine learning algorithms such as support vector machines, random forests, etc. and deep learning networks such as convolutional neural networks. Determine the types of hyperparameters according to the constructed machine learning network. The hyperparameters include: a. Hyperparameters related to the network structure, including hyperparameters of the convolutional layer such as the convolutional kernel size, the number of convolutional kernels, the convolutional stride, and the padding method, hyperparameters of the pooling layer such as the pooling window size, the pooling stride, and the pooling method, and hyperparameters of the fully connected layer such as the number of layers and the number of neurons in each layer.

[0036] b. Hyperparameters related to training, including the learning rate, batch size, number of iterations, and optimizer selection.

[0037] The learning rate controls the step size of the weight update during the training of the model and is a key parameter affecting the convergence speed and stability of the model; the batch size determines the number of samples used when updating the model parameters each time, affecting the training speed and the stability of the model; the number of iterations refers to the number of times the entire training set data is input into the neural network model to complete a full training process, and the number of iterations affects the training degree and performance of the model; optimizer selection such as SGD (Stochastic Gradient Descent), Adam, etc., different optimizers have an impact on the training effect and convergence speed of the model.

[0038] c. Hyperparameters related to regularization, including the Dropout rate and the L2 regularization coefficient. The Dropout rate is the proportion of randomly discarded neurons during the training process, used to prevent overfitting. The L2 regularization coefficient (weight decay) is the weight decay term added to the optimizer, used to control the model complexity and prevent overfitting.

[0039] Furthermore, the dung beetle algorithm simulates the foraging behavior and information exchange mechanism of dung beetles in the natural environment, and through iterative search and self - adjustment, automatically finds a set of hyperparameter combinations that can optimize the model performance.

[0040] The dung beetle algorithm contains various dung beetles. The rolling dung beetle: imitates the behavior of dung beetles rolling dung balls to find suitable locations, and its position update strategy focuses on global search to explore new solution spaces; the breeding dung beetle: simulates the process of dung beetles reproducing offspring, and its position update strategy tends to improve the solution through information exchange within the population; the small dung beetle: reflects the local search behavior of dung beetles when looking for food, and its position update method focuses on local fine - tuning to fine - tune and improve the current better solution; the stealing dung beetle: imitates the behavior of dung beetles competing for resources, obtaining resources from other dung beetles to improve its own position, which helps the spread of information within the population and the dissemination of high - quality solutions.

[0041] It should be understood that in the traditional dung beetle algorithm, the small dung beetle forages in the optimal foraging area in a random direction, and as the number of iterations increases, the position of the thief dung beetle will be updated towards the global optimal direction, which may lead to the problem of falling into a local optimal solution. In order to accelerate the optimization speed of the small dung beetle in the foraging area, improve the global optimization ability and local search efficiency, the present invention improves the traditional dung beetle algorithm, including the following steps: First, decompose the optimal foraging area of the small dung beetles.

[0042] Specifically, divide the upper and lower boundaries of the optimal foraging area, that is, the boundary values of the optimal problem solution, proportionally according to the number of small dung beetles.

[0043] Secondly, adjust the foraging directions of the small dung beetles according to the decomposition results.

[0044] Specifically, adjusting the foraging directions of the small dung beetles according to the decomposition results satisfies the following formula:

[0045] Wherein, represents the position of the th small dung beetle at the th iteration, represents the position of the th small dung beetle at the th iteration, represents the current iteration number, represents the maximum iteration number, represents the th small dung beetle at the th iteration, and is the central value of the decomposition result that is closest. In the embodiment, the degree of closeness is judged according to the Euclidean distance between the position of the small dung beetle and the central value of the decomposition result.

[0046] Finally, based on the iteration number, adjust the number of stealing dung beetles according to the optimization result.

[0047] Specifically, adjusting the number of stealing dung beetles based on the iteration number according to the optimization result satisfies the following formula:

[0048] Wherein, represents the judgment value of whether the th dung beetle mutates into a stealing dung beetle at the th iteration, represents mutating into a stealing dung beetle, represents not mutating into a stealing dung beetle, represents the fitness of the th dung beetle, represents the global worst fitness, represents a random number belonging to (0, 1), represents the current iteration number, represents the maximum iteration number.

[0049] It should be understood that the fitness is used to evaluate the quality of the position of the dung beetle, that is, the problem solution. The smaller the fitness value, the worse the problem solution.

[0050] S22. Construct a mutual prediction model for sensors based on the optimal parameters.

[0051] Specifically, apply the obtained optimal hyperparameters to the constructed mutual prediction model for sensors.

[0052] Furthermore, obtaining the first predicted value of the bridge monitoring sensors using the mutual prediction model for sensors includes the following steps: First, based on the monitoring data of the sensors respectively, with the current sensor monitoring value as the output and the values of other sensors as the input, establish a training data set. Second, use the training data set to train and validate the mutual prediction model for sensors, and then obtain the first predicted value of the corresponding sensors according to the real-time monitoring data.

[0053] S3. Analyze the historical monitoring data of the bridge monitoring sensors, and reconstruct the data reconstruction value of the bridge monitoring sensors according to the analysis result.

[0054] In the embodiment, analyzing the historical monitoring data of the bridge monitoring sensors and reconstructing the data reconstruction value of the bridge monitoring sensors according to the analysis result in step S3 includes the following steps: S31. Set the sequence window length of the historical monitoring data.

[0055] According to the monitoring frequency, data storage conditions, computing resources and other actual situations, set the sequence window length of the historical monitoring data. In the embodiment, the sequence window length is set to 10, that is, 10 sequence data.

[0056] S32. Based on the sequence window length, fit the first linear function of the historical monitoring data under the corresponding window length.

[0057] Specifically, through 10 sequence data, fit the first linear function of the corresponding sequence data. The least squares method can be used to find the best fit line. The least squares method determines the slope and intercept of the line by minimizing the sum of the squares of the perpendicular distances from the data points to the line.

[0058] Exemplarily, the sequence data is (t1, y1), (t2, y2), … (t10, y10), where t represents time and y represents the monitoring value. The first linear function obtained after fitting is .

[0059] S33. Combine the first linear function and the historical monitoring data to obtain the calculated value of the first linear function of the corresponding window monitoring data.

[0060] Specifically, substitute t10 into the first linear function obtained after fitting to obtain the calculated value of the first linear function of the corresponding window monitoring data.

[0061] S34. Evaluate the calculated value of the first linear function, and reconstruct the data reconstruction value of the bridge monitoring sensor according to the evaluation result.

[0062] In the embodiment, evaluating the calculated value of the first linear function and reconstructing the data reconstruction value of the bridge monitoring sensor according to the evaluation result includes the following steps: First, set a difference threshold.

[0063] Specifically, calculate the actual difference value between the calculated value of the first linear function and the historical monitoring data to obtain the actual difference values of consecutive multiple sequence windows.

[0064] Based on the percentile method, set the difference threshold according to the actual difference value.

[0065] Specifically, arrange the actual difference values in ascending order. Based on the percentile method, select an appropriate percentile according to requirements, and find the corresponding value from the sorted difference value sequence as the difference threshold. For example, the difference value corresponding to the 75th percentile is the difference threshold.

[0066] Second, compare the difference threshold with the difference value between the calculated value of the first linear function and the historical monitoring data to obtain an evaluation result.

[0067] Specifically, exclude the corresponding sequence windows where the difference threshold and the difference value between the calculated value of the first linear function and the historical monitoring data are greater than the difference threshold, and do not perform the next data reconstruction.

[0068] Finally, fit a second linear function according to the evaluation result, and reconstruct the data reconstruction value of the bridge monitoring sensor through the second linear function.

[0069] Specifically, perform least squares fitting on the calculated values of the first linear function that meet the conditions in the evaluation result to obtain a second linear function. Based on the second linear function, reconstruct the data reconstruction value of the bridge monitoring sensor with the calculated values of the first linear function of multiple sequence windows.

[0070] S4. Combine the first predicted value and the data reconstruction value to identify the working state of the bridge monitoring sensor.

[0071] In the embodiment, combining the first predicted value and the data reconstruction value to identify the working state of the bridge monitoring sensor includes the following steps: First, obtain a first residual sequence according to the first predicted value and the monitoring data.

[0072] Specifically, obtain a first residual sequence according to the difference between the first predicted value and the monitoring data and the corresponding time series order.

[0073] Further, calculate the mean and standard deviation of the corresponding sequence window according to the first residual sequence.

[0074] Next, obtain a second residual sequence according to the first predicted value and the data reconstruction value.

[0075] Specifically, obtain the second residual sequence according to the difference between the first predicted value and the data reconstruction value and the corresponding time series order.

[0076] Further, calculate the mean absolute deviation of the corresponding sequence window according to the second residual sequence.

[0077] Then, combine the first residual sequence and the second residual sequence to establish a confidence interval.

[0078] Specifically, obtain the confidence space according to the mean, the standard deviation, and the mean absolute deviation.

[0079] Further, obtaining the confidence space according to the mean, the standard deviation, and the mean absolute deviation satisfies the following formula:

[0080] where represents the confidence amplitude, represents the mean, represents the standard deviation, represents the mean absolute deviation.

[0081] Finally, identify the working state of the bridge monitoring sensor according to the confidence interval.

[0082] Specifically, the confidence interval satisfies the following formula:

[0083] where represents the first predicted value, represents the data reconstruction value, represents the confidence interval.

[0084] Further, compare the monitoring data with the confidence interval. If the monitoring data is not within the confidence interval, it is considered that the working state of the bridge monitoring sensor is abnormal and further manual detection is required.

[0085] It should be understood that when identifying the working state of the bridge monitoring sensor, the following situations should also be filtered: 1. The sensor has no data output, resulting in the monitoring data being 0; 2. The data output by the sensor is a fixed value for a certain period of time, resulting in the monitoring data being constant.

[0086] Please refer to Figure 2 In an embodiment, to efficiently execute a method for identifying faults in bridge monitoring sensors provided by the present invention, the present invention further provides a system for identifying faults in bridge monitoring sensors, including: an input device, an output device, a processor, and a memory. The input device, the output device, the processor, and the memory are interconnected. The memory contains program instructions for the steps of the method for identifying faults in bridge monitoring sensors. The system for identifying faults in bridge monitoring sensors of the present invention has a compact structure and stable performance, and can stably execute the method for identifying faults in bridge monitoring sensors of the present invention, further improving the overall applicability and practical application ability of the present invention.

[0087] In an embodiment, the so-called processor may be a central processing unit (CPU), and this processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc. The input device may be used to obtain data information. The output device may be used to output the result obtained from the program instructions included in the computer program stored in the memory provided by the present invention. The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include a non-volatile random access memory.

[0088] In a possible implementation, the memory may include a program storage area and a data storage area. Among them, the program storage area may store an operating system and application programs required for at least one function, etc.; the data storage area may store data created during use. In addition, the memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include NVRAM. The memory stores an operating system and operation instructions, executable modules, or data structures, or subsets thereof, or extended sets thereof. Among them, the operation instructions may include various operation instructions for implementing various operations. The operating system may include various system programs for implementing various basic tasks and processing hardware-based tasks.

[0089] In an embodiment, a storage medium is further provided. A computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the above-mentioned method for identifying faults in bridge monitoring sensors are implemented.

[0090] The storage medium may include: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs.

[0091] In summary, based on the historical data of sensors, the present invention obtains the predicted values and data reconstruction values of monitoring data, and then constructs a credible space for the monitoring data to evaluate whether the monitoring data is abnormal, solving the problem of efficiently and accurately identifying faulty sensors, which is beneficial for the bridge health monitoring system to judge the state of the bridge. Therefore, the present invention effectively overcomes various drawbacks in the prior art and has high industrial utilization value.

[0092] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the various embodiments of the present invention, and they should all be covered by the scope recorded in the present invention.

Claims

1. A method for identifying faults in bridge monitoring sensors, characterized in that, It includes the following steps: Obtain the monitoring data of the bridge monitoring sensors; Construct a mutual prediction model for the sensors, and use the mutual prediction model for the sensors to obtain the first predicted value of the bridge monitoring sensors; Analyze the historical monitoring data of the bridge monitoring sensors, and reconstruct the data reconstruction value of the bridge monitoring sensors according to the analysis results; Combine the first predicted value and the data reconstruction value to identify the working state of the bridge monitoring sensors.

2. The method for identifying faults of bridge monitoring sensors according to claim 1, wherein It further includes the following steps: Construct a data denoising model; Use the data denoising model to perform denoising processing on the monitoring data.

3. The method for identifying faults of bridge monitoring sensors according to claim 1, characterized in that, The construction of the mutual prediction model for the sensors includes the following steps: Use an improved dung beetle algorithm to obtain the optimal parameters of the machine learning network; Based on the optimal parameters, construct a mutual prediction model for the sensors.

4. The method for identifying faults of bridge monitoring sensors according to claim 3, wherein, The improvement of the dung beetle algorithm includes the following steps: Decompose the optimal foraging area of the small dung beetles, and adjust the foraging direction of the small dung beetles according to the decomposition results; Based on the number of iterations, adjust the number of stealing dung beetles according to the optimization results.

5. The method for identifying faults of a bridge monitoring sensor according to claim 1, wherein The analysis of the historical monitoring data of the bridge monitoring sensors and the reconstruction of the data reconstruction value of the bridge monitoring sensors according to the analysis results include the following steps: Set the sequence window length of the historical monitoring data; Based on the sequence window length, fit the first linear function of the historical monitoring data under the corresponding window length; Combine the first linear function and the historical monitoring data to obtain the calculated value of the first linear function of the monitoring data corresponding to the window; Evaluate the calculated value of the first linear function, and reconstruct the data reconstruction value of the bridge monitoring sensors according to the evaluation results.

6. The method for identifying faults of a bridge monitoring sensor according to claim 5, characterized in that, The evaluation of the calculated value of the first linear function and the reconstruction of the data reconstruction value of the bridge monitoring sensors according to the evaluation results include the following steps: Set a difference threshold; Compare the difference threshold with the difference value between the calculated value of the first linear function and the historical monitoring data to obtain an evaluation result; Fit a second linear function according to the evaluation result, and reconstruct the data reconstruction value of the bridge monitoring sensors through the second linear function.

7. The method for identifying faults of a bridge monitoring sensor according to claim 6, wherein, The setting of the difference threshold includes the following steps: Calculate the actual difference value between the calculated value of the first linear function and the historical monitoring data; Based on the percentile method, set the difference threshold according to the actual difference value.

8. The method for identifying faults of bridge monitoring sensors according to claim 1, wherein The combination of the first predicted value and the data reconstruction value to identify the working state of the bridge monitoring sensors includes the following steps: According to the first predicted value and the monitoring data, obtain a first residual sequence; According to the first predicted value and the data reconstruction value, obtain a second residual sequence; Combine the first residual sequence and the second residual sequence to establish a confidence interval; According to the confidence interval, identify the working state of the bridge monitoring sensors.

9. The method for identifying the fault of the bridge monitoring sensor according to claim 8, wherein The combination of the first residual sequence and the second residual sequence to establish a confidence interval includes the following steps: Calculate the mean and standard deviation of the first residual sequence and the mean of the absolute deviation of the second residual sequence; According to the mean, the standard deviation and the mean of the absolute deviation, obtain a confidence space; Combine the confidence space, the first predicted value and the data reconstruction value to establish a confidence interval.

10. A bridge monitoring sensor fault identification system, characterized in that, The bridge monitoring sensor fault identification system includes: an input device, an output device, a processor, and a memory. The input device, the output device, the processor, and the memory are interconnected. The memory includes program instructions for executing the bridge monitoring sensor fault identification method according to any one of claims 1-9.

Images