Intelligent inspection method and system based on multi-dimensional data fusion analysis

Abstract

The invention discloses an intelligent inspection method and system based on multi-dimensional data fusion analysis, and relates to the technical field of intelligent monitoring and maintaining.The intelligent inspection method comprises the steps that telemetering data, telecommand data and remote vision data of equipment are collected from multiple management platforms; constructing a time sequence model based on the telemetry data to process the telemetry data to obtain processed telemetry data; performing primary fusion on the processed telemetering data and telesignalling data to obtain primary fusion data; performing visual feature vector extraction on the remote vision data; performing advanced fusion on the primary fusion data and the visual feature vector to obtain advanced fusion data; and evaluating the equipment state based on the advanced fusion data, and finally generating an inspection report. According to the method, abnormal points can be accurately identified and removed, and complex association among different data sources is fully mined. The precision and reliability of equipment state evaluation are improved, a detailed inspection report is provided for inspection personnel, potential problems can be found in time, and an effective maintenance plan can be made.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring and maintenance technology, and in particular to an intelligent inspection method and system based on multi-dimensional data fusion analysis. Background Art

[0002] With the development of industrial automation and intelligence, equipment inspection has gradually shifted from traditional manual inspection to intelligent inspection. Modern inspection systems collect data to achieve real-time monitoring and evaluation of equipment status. However, existing technologies have many shortcomings in processing these multi-source heterogeneous data. First, the time series modeling of telemetry data usually relies on simple statistical methods, resulting in large deviations in subsequent analysis results. Second, existing data fusion methods mostly use traditional weighted average or linear combination methods, which limits the fusion effect.

[0003] In addition, the processing of remote viewing data mostly stays at the stage of image enhancement and segmentation, lacking in-depth feature extraction and fusion mechanisms, and unable to fully utilize visual information for equipment status assessment. When processing multidimensional data fusion, existing technologies often ignore the inherent connection between different data sources, resulting in the fused data failing to fully reflect the true status of the equipment. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent inspection method based on multidimensional data fusion analysis to solve the problems of abnormal point detection and complex association mining in multi-source heterogeneous data fusion.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides an intelligent inspection method based on multi-dimensional data fusion analysis, which includes collecting status data of equipment from multiple management platforms, wherein the status data includes telemetry data, telesignaling data, and televiewing data; A time series model is built based on telemetry data, the residual between the predicted value and the observed value is calculated, and the Dixon criterion is used to judge and remove abnormal points to obtain processed telemetry data; Performing primary fusion on the processed telemetry data and telesignaling data to obtain primary fusion data; Use convolutional neural network model to extract visual feature vectors from remote viewing data; Performing advanced fusion on the primary fusion data and the visual feature vector to obtain advanced fusion data; Evaluate equipment status based on advanced fusion data and ultimately generate inspection reports.

[0007] As a preferred solution of the intelligent inspection method based on multi-dimensional data fusion analysis described in the present invention, the telemetry data includes temperature and humidity, the telesignaling data refers to the switch status, and the remote viewing data refers to the equipment operation status image taken by the camera.

[0008] As a preferred solution of the intelligent inspection method based on multidimensional data fusion analysis described in the present invention, the time series model is constructed based on the telemetry data, the residual between the predicted value and the observed value is calculated, and the Dixon criterion is used to judge and remove the abnormal points to obtain the processed telemetry data, which includes the following steps: Use the ARIMA model as the time series model and use the AIC criterion to determine the optimal parameter combination; Use the ARIMA model with the optimal parameter combination to predict future observations and generate forecast values; Calculate the residual between each pair of observed values ​​and predicted values ​​to obtain the residual sequence. Use the standard deviation of the residual sequence to define the outlier threshold. Use the outlier threshold to check each residual, mark the outliers and remove them from the telemetry data to obtain the processed telemetry data.

[0009] As a preferred solution of the intelligent inspection method based on multi-dimensional data fusion analysis described in the present invention, the primary fusion includes the following steps: Convert the processed telemetry data and telesignaling data into quantum state representation using quantum principal component analysis and reconstruct the data; The reconstructed data is used to perform adaptive weighted average fusion optimization weights, and the optimized weights are used for fusion to obtain primary fusion data.

[0010] As a preferred solution of the intelligent inspection method based on multi-dimensional data fusion analysis described in the present invention, the convolutional neural network model is used to extract visual feature vectors from remote viewing data, including the following steps: Enhance the equipment operation status image in the remote viewing data, and use semantic segmentation technology to segment the equipment area in the equipment operation status image; Based on the multi-scale convolutional neural network, each path processes image features of different scales, and feature fusion layers are added at different levels of the multi-scale convolutional neural network to fuse feature maps of different scales; A global average pooling operation is used before the fully connected layer of the multi-scale convolutional neural network to compress the spatial dimension of each feature map into a single value, thereby generating a high-dimensional feature vector of fixed length; Use principal component analysis to reduce the dimensionality of the extracted high-dimensional feature vector; A contrastive learning strategy is adopted to optimize the high-dimensional feature vector after dimensionality reduction by constructing positive and negative sample pairs, and finally a visual feature vector is obtained.

[0011] As a preferred solution of the intelligent inspection method based on multi-dimensional data fusion analysis described in the present invention, the advanced fusion includes the following steps: Combine the visual feature vectors of primary fusion data and remote viewing data, and use the covariance crossover algorithm to calculate the mean vector and covariance matrix after advanced fusion; In the process of advanced fusion, quantum heuristic algorithm is used to optimize advanced fusion parameters; The high-level fusion parameters optimized by the quantum heuristic algorithm are substituted into the covariance crossover algorithm, and the optimized mean vector and covariance matrix are calculated to obtain the high-level fusion data.

[0012] As a preferred solution of the intelligent inspection method based on multi-dimensional data fusion analysis described in the present invention, the following steps are included: evaluating the equipment status based on advanced fusion data and generating an inspection report: Set the normal range of equipment status based on the equipment's historical data and operating specifications; Extract state parameters from the mean vector, compare the extracted state parameters with the normal range, and determine whether the device state is normal; Evaluate the uncertainty of state parameters based on the covariance matrix; According to the status parameters and uncertainty assessment results, the equipment status is divided into normal status, warning status and abnormal status; Integrate the assessed equipment status with the equipment status data to generate an inspection report.

[0013] In a second aspect, the present invention provides an intelligent inspection system based on multi-dimensional data fusion analysis, including: a data acquisition module responsible for collecting status data of equipment from multiple management platforms, wherein the status data includes telemetry data, telesignaling data and televiewing data; The processing module is responsible for building a time series model based on telemetry data, calculating the residual between the predicted value and the observed value, and using the Dixon criterion to determine and remove outliers to obtain processed telemetry data; The primary fusion module is responsible for primary fusion of the processed telemetry data and telesignaling data to obtain primary fusion data; The feature extraction module is responsible for extracting visual feature vectors from remote viewing data using a convolutional neural network model; The advanced fusion module is responsible for performing advanced fusion of primary fusion data and visual feature vectors to obtain advanced fusion data; The report generation module is responsible for evaluating the equipment status based on advanced fusion data and finally generating an inspection report.

[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the intelligent inspection method based on multidimensional data fusion analysis as described in the first aspect of the present invention is implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent inspection method based on multidimensional data fusion analysis as described in the first aspect of the present invention.

[0016] The beneficial effects of the present invention are as follows: by constructing a time series model and using the Dixon criterion to remove anomalies in telemetry data, the accuracy and reliability of the data are ensured. On this basis, quantum principal component analysis and adaptive weighted average fusion are used to process telemetry data and telesignaling data, which significantly improves the quality of primary fusion data. At the same time, a convolutional neural network model is used to extract visual feature vectors from remote viewing data, and advanced fusion is performed in combination with the covariance crossover algorithm and the quantum heuristic algorithm, thereby achieving efficient fusion and deep mining of multi-source heterogeneous data. The present invention can not only accurately identify and remove anomalies, but also fully mine the complex associations between different data sources, thereby generating more accurate equipment status assessment results. This not only improves the accuracy and reliability of equipment status assessment, but also provides detailed inspection reports for inspection personnel, helping to promptly identify potential problems and formulate effective maintenance plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0018] Figure 1 This is a flow chart of the intelligent inspection method based on multi-dimensional data fusion analysis in Example 1.

[0019] Figure 2 Schematic diagram of the intelligent inspection system based on multi-dimensional data fusion analysis in Example 1. DETAILED DESCRIPTION

[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0021] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0023] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides an intelligent inspection method based on multi-dimensional data fusion analysis, comprising the following steps: S1. Collect equipment status data from multiple management platforms. The status data includes telemetry data, remote signaling data and remote viewing data.

[0024] Telemetry data includes temperature and humidity, telesignaling data refers to switch status, and remote viewing data refers to images of equipment operation status taken by cameras.

[0025] It should be noted that the status data of the equipment is collected from various management platforms (such as SCADA systems, IoT platforms, etc.), including telemetry data (temperature, humidity), telesignaling data (switch status), and remote viewing data (images of equipment operation taken by cameras). By collecting multi-source heterogeneous data from various management platforms, the acquisition of all-round status information of the equipment is achieved, ensuring the comprehensiveness and diversity of the data, and providing a solid foundation for subsequent data analysis and equipment status evaluation. This step ensures that the data processed and fused subsequently is sufficiently rich and accurate.

[0026] S2. Build a time series model based on telemetry data, calculate the residual between the predicted value and the observed value, and use the Dixon criterion to judge and remove outliers to obtain the processed telemetry data.

[0027] The ARIMA (AutoRegressive Integrated Moving Average) model is used as the time series model, and the AIC (Akaike Information Criterion) criterion is used to determine the optimal parameter combination, which includes the number of autoregressive terms, the number of differences, and the number of moving average terms; Specifically, the historical telemetry data were differentially processed to eliminate the trend and seasonal components, the AIC values ​​under different parameter combinations were calculated, and the combination with the smallest AIC value was selected as the optimal parameter combination; Use the ARIMA model with the optimal parameter combination to predict future observations and generate forecast values; It should be noted that the original data is first processed for stationarity (such as difference), all possible parameter combinations are traversed through grid search, and the Akaike Information Criterion is used to screen out the optimal parameters that balance the model complexity and fitting accuracy. After the ARIMA model is trained based on the optimal parameters, the prediction results of future observations are generated point by point using a recursive method.

[0028] Calculate the residual between each pair of observed values ​​and predicted values ​​to obtain a residual sequence, use the standard deviation of the residual sequence to define an outlier threshold, use the outlier threshold to check each residual, mark the outlier points and remove them from the telemetry data to obtain the processed telemetry data; Specifically, the calculation of the outlier threshold is expressed as, ; in, represents the outlier threshold, represents the residual mean, represents the confidence interval (95% or 99%), which is a constant, usually taking the value of 2 or 3. represents the standard deviation of the residual series; Check each residual ,when , it is judged as an outlier, marked and removed.

[0029] It should be noted that a time series model (ARIMA) is constructed based on telemetry data (such as temperature and humidity), the residual between the predicted value and the observed value is calculated, and the Dixon criterion is used to judge and remove the outliers to obtain the processed telemetry data. By using the ARIMA model for time series modeling and combining the AIC criterion to optimize the parameter combination, the trend and seasonal components of historical telemetry data are effectively processed. The Dixon criterion is used to mark and remove outliers, which improves the quality and reliability of the data, reduces errors in subsequent analysis, and improves the prediction accuracy and stability.

[0030] S3. Perform primary fusion on the processed telemetry data and telesignaling data to obtain primary fusion data.

[0031] Primary fusion refers to using quantum principal component analysis to convert the processed telemetry data and telesignaling data into quantum state representation, reconstruct the data, use the reconstructed data to perform adaptive weighted average fusion optimization weights, and use the optimized weights for fusion to obtain primary fusion data; Specifically, the processed telemetry data and telesignaling data are mapped to the quantum bit space through quantum state encoding, and the main components of the data are extracted using QPCA (quantum principal component analysis). Based on the main components, the processed telemetry data and telesignaling data are projected into a low-dimensional space, and the data is reconstructed. The reconstructed data is used for adaptive weighted average fusion, and the initial weight of the reconstructed data is set to a uniform distribution. The fusion value is calculated using the adaptive weighted average method. Based on the fusion value, the minimum mean square error is used as the objective function to dynamically adjust the weight, and the gradient descent method is used to minimize the mean square error until convergence to obtain the optimized weight. The processed telemetry data and telesignaling data are fused using the optimized weight to obtain primary fused data. It should be noted that before using quantum principal component analysis, the telemetry data and telesignaling data must be processed using timestamp alignment.

[0032] It should be noted that the processed telemetry data and telesignaling data are converted into quantum state representation using quantum principal component analysis (QPCA), and the data is reconstructed, and the weights are optimized using adaptive weighted average fusion to obtain primary fusion data. Through quantum principal component analysis and adaptive weighted average fusion, efficient fusion of telemetry data and telesignaling data is achieved. This method not only extracts the main components of the data, but also optimizes the fusion weights, so that the primary fusion data can more accurately reflect the true state of the equipment and enhance the integrity and consistency of the data.

[0033] S4. Use the convolutional neural network model to extract visual feature vectors from remote viewing data.

[0034] Enhance the equipment operation status images in the remote viewing data, including but not limited to contrast enhancement, histogram equalization, noise removal, etc., to improve the quality of the equipment operation status images, and use semantic segmentation technology (such as U-Net, DeepLab, etc.) to segment the equipment area in the equipment operation status images, remove background interference, and focus on the key areas of equipment operation; It should be noted that the enhanced image is classified at the pixel level through a semantic segmentation model (such as U-Net) to separate the equipment body and core components (such as circuit boards, bearings, valves, etc.) from the background. The key area is defined as a continuous pixel area where the semantic segmentation model predicts a probability higher than a set pixel threshold (such as 0.9), which usually includes functional components of the equipment and potential fault indications.

[0035] Based on the multi-scale convolutional neural network, each path processes the image features of the device operation status at different scales. For example, one path processes the original resolution device operation status image, and the other path processes the downsampled device operation status image. Feature fusion layers are added at different levels of the multi-scale convolutional neural network to fuse feature maps of different scales to capture richer visual information. A global average pooling operation is used before the fully connected layer of the multi-scale convolutional neural network to compress the spatial dimension of each feature map into a single value, thereby generating a high-dimensional feature vector of fixed length; The extracted high-dimensional feature vectors are reduced in dimension using principal component analysis to reduce computational complexity and retain the most important visual information. A contrastive learning strategy is adopted to optimize the high-dimensional feature vector after dimensionality reduction by constructing positive and negative sample pairs, and finally a visual feature vector is obtained.

[0036] It should be noted that the convolutional neural network (CNN) model is used to enhance the remote viewing data (equipment operation status images), perform semantic segmentation and multi-scale feature extraction, generate a high-dimensional feature vector of fixed length, and further optimize the feature vector through principal component analysis and contrastive learning strategies. Through multi-scale convolutional neural network and semantic segmentation technology, deep feature extraction of equipment operation status images is achieved, background interference is removed, and key areas are focused. Combined with principal component analysis and contrastive learning strategies, the feature vector is further optimized, the most important visual information is retained, and the expressiveness and recognition accuracy of equipment operation status image data are significantly improved.

[0037] S5. Perform advanced fusion on the primary fusion data and the visual feature vector to obtain advanced fusion data.

[0038] The primary fusion data (i.e., the fusion result of telemetry and telesignaling data) and the visual feature vector of remote viewing data are combined, and the mean vector and covariance matrix after advanced fusion are calculated using the covariance crossover algorithm, which is expressed as: ; ; in, represents the covariance matrix after high-level fusion, represents the inverse of the covariance matrix after high-level fusion, represents the mean vector after high-level fusion, and denote the inverse of the covariance matrix of the primary fusion data and the visual feature vector, respectively, and Represent the covariance matrix of primary fusion data and visual feature vector respectively, and Represent the mean vector of primary fusion data and visual feature vector respectively, Represents high-level fusion weights, usually satisfying , used to balance the importance of primary fusion data and visual feature vectors; In the process of advanced fusion, quantum heuristic algorithms (such as quantum genetic algorithm, quantum annealing, etc.) are used to optimize advanced fusion parameters; Specifically, the fusion weights and other parameters are encoded into quantum bits, a set of quantum states are randomly generated as the initial population, the covariance matrix determinant of the fusion result corresponding to each quantum state is calculated as the fitness value, the phase of the quantum bit is adjusted according to the fitness value to guide the search direction, the information of different quantum states is exchanged through quantum entanglement operations to increase the population diversity, and the fitness evaluation and quantum operations are repeated until the convergence conditions are met (such as the fitness value change is less than the set fitness value threshold or the maximum number of iterations is reached), and the fusion weights and other parameters are decoded from the optimal quantum state; The high-level fusion parameters optimized by the quantum heuristic algorithm are substituted into the covariance crossover algorithm, and the optimized mean vector and covariance matrix are calculated to obtain the high-level fusion data.

[0039] It should be noted that the primary fusion data (the fusion result of telemetry and telesignaling data) and the visual feature vector are combined, and the covariance crossover algorithm and quantum heuristic algorithm are used for advanced fusion to generate advanced fusion data. The covariance crossover algorithm and quantum heuristic algorithm are used to achieve efficient fusion of primary fusion data and visual feature vectors, optimize fusion weights and other parameters, and enable advanced fusion data to more comprehensively reflect the status of the equipment. This method not only improves the quality of fusion data, but also enhances the robustness and interpretability of the data.

[0040] S6. Evaluate equipment status based on advanced fusion data and finally generate an inspection report.

[0041] Set the normal range of equipment status based on the equipment's historical data and operating specifications; Specifically, perform statistical analysis on historical data and calculate the mean , Standard Deviation For example, for temperature data, set the normal range. ,in is a constant (such as 2 or 3) that represents the confidence interval, which is used to obtain the operating specifications and recommended operating parameter ranges provided by the equipment manufacturer and further adjust and optimize the normal range; Extract state parameters (such as temperature, humidity, switch status, etc.) from the mean vector, compare the extracted state parameters with the normal range, and determine whether the device status is normal; Specifically, assume that the mean vector is , from which the temperature is extracted ,humidity and switch status ; Based on the advanced fusion data, the covariance matrix of the state parameters is calculated to evaluate the uncertainty of the state parameters. If the diagonal elements (variance) of the covariance matrix are large, it means that the uncertainty of the parameter estimation is high and further verification is required. Combined with the visual feature vectors in the remote viewing data, cross-validation is performed. For example, check whether there are overheated areas in the equipment operation image and verify whether the temperature parameters are accurate. If the remote viewing data verification results show that the state parameters are biased, the state parameters are adjusted according to the information provided by the remote viewing data. For example, if the equipment operation image shows that the equipment is overheated but the temperature parameters are within the normal range, the estimated value of the temperature parameters is appropriately adjusted upward. Based on the results of the remote viewing data verification, the covariance matrix of the state parameters is updated to reflect the new level of uncertainty. According to the status parameters and uncertainty assessment results, the equipment status is divided into normal status, warning status and abnormal status; It should be noted that the normal state is when all state parameters are within the normal range and the uncertainty is low; the warning state is when some state parameters are close to the normal range or the uncertainty is high and needs attention; the abnormal state is when the state parameters are beyond the normal range or the uncertainty is extremely high and needs to be handled immediately; Specifically, for the temperature parameter , check whether it is in within the range.

[0042] For humidity parameters , check whether it is in within the range.

[0043] For switch status parameters , check whether it matches the set on / off status (such as 0 for off and 1 for on).

[0044] The comparison results of each status parameter are recorded as the basis for subsequent analysis and decision-making to obtain the evaluation results of the equipment status; Integrate the assessed equipment status with the equipment status data to generate an inspection report.

[0045] It should be noted that by setting the normal range and combining it with statistical analysis, an accurate assessment of the equipment status is achieved. Based on the status parameters and uncertainty assessment results, the equipment status is classified as normal, warning or abnormal, and a detailed inspection report is generated. This step not only helps inspectors to identify potential problems in a timely manner, but also provides a basis for formulating effective maintenance plans.

[0046] This embodiment also provides an intelligent inspection system based on multi-dimensional data fusion analysis, including: a data acquisition module, responsible for collecting status data of equipment from multiple management platforms, the status data including telemetry data, telesignaling data and televiewing data; The processing module is responsible for building a time series model based on telemetry data, calculating the residual between the predicted value and the observed value, and using the Dixon criterion to determine and remove outliers to obtain processed telemetry data; The primary fusion module is responsible for primary fusion of the processed telemetry data and telesignaling data to obtain primary fusion data; The feature extraction module is responsible for extracting visual feature vectors from remote viewing data using a convolutional neural network model; The advanced fusion module is responsible for performing advanced fusion of primary fusion data and visual feature vectors to obtain advanced fusion data; The report generation module is responsible for evaluating the equipment status based on advanced fusion data and finally generating an inspection report.

[0047] This embodiment also provides a computer device, which is suitable for the case of an intelligent inspection method based on multidimensional data fusion analysis, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the intelligent inspection method based on multidimensional data fusion analysis proposed in the above embodiment.

[0048] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0049] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the intelligent inspection method based on multi-dimensional data fusion analysis proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0050] In summary, the present invention ensures the accuracy and reliability of the data by constructing a time series model and using the Dixon criterion to remove anomalies in the telemetry data. On this basis, quantum principal component analysis and adaptive weighted average fusion are used to process telemetry data and telesignaling data, which significantly improves the quality of primary fusion data. At the same time, a convolutional neural network model is used to extract visual feature vectors from remote viewing data, and advanced fusion is performed in combination with the covariance crossover algorithm and the quantum heuristic algorithm, thereby achieving efficient fusion and deep mining of multi-source heterogeneous data. The present invention can not only accurately identify and remove anomalies, but also fully mine the complex associations between different data sources, thereby generating more accurate equipment status assessment results. This not only improves the accuracy and reliability of equipment status assessment, but also provides detailed inspection reports for inspection personnel, helping to promptly identify potential problems and formulate effective maintenance plans.

[0051] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent inspection method based on multi-dimensional data fusion analysis, characterized in that: include, Collecting equipment status data from multiple management platforms, the status data including telemetry data, telesignaling data and televiewing data; A time series model is built based on telemetry data, the residual between the predicted value and the observed value is calculated, and the Dixon criterion is used to judge and remove outliers to obtain processed telemetry data; Performing primary fusion on the processed telemetry data and telesignaling data to obtain primary fusion data; Use convolutional neural network model to extract visual feature vectors from remote viewing data; Performing advanced fusion on the primary fusion data and the visual feature vector to obtain advanced fusion data; Evaluate equipment status based on advanced fusion data and ultimately generate inspection reports.

2. The intelligent inspection method based on multi-dimensional data fusion analysis according to claim 1, characterized in that: The telemetry data includes temperature and humidity, the telesignaling data refers to the switch status, and the remote viewing data refers to the image of the equipment operation status taken by the camera.

3. The intelligent inspection method based on multi-dimensional data fusion analysis as claimed in claim 2, characterized in that: Based on the telemetry data, a time series model is constructed to calculate the residual between the predicted value and the observed value, and the Dixon criterion is used to judge and remove abnormal points to obtain the processed telemetry data, including the following steps: Use the ARIMA model as the time series model and use the AIC criterion to determine the optimal parameter combination; Use the ARIMA model with the optimal parameter combination to predict future observations and generate forecast values; Calculate the residual between each pair of observed values ​​and predicted values ​​to obtain the residual sequence. Use the standard deviation of the residual sequence to define the outlier threshold. Use the outlier threshold to check each residual, mark the outliers and remove them from the telemetry data to obtain the processed telemetry data.

4. The intelligent inspection method based on multi-dimensional data fusion analysis according to claim 3 is characterized in that: The primary fusion comprises the following steps: Convert the processed telemetry data and telesignaling data into quantum state representation using quantum principal component analysis and reconstruct the data; The reconstructed data is used to perform adaptive weighted average fusion optimization weights, and the optimized weights are used for fusion to obtain primary fusion data.

5. The intelligent inspection method based on multi-dimensional data fusion analysis as claimed in claim 4, characterized in that: The convolutional neural network model is used to extract visual feature vectors from remote viewing data, including the following steps: Enhance the equipment operation status image in the remote viewing data, and segment the equipment area in the equipment operation status image using semantic segmentation technology; Based on the multi-scale convolutional neural network, each path processes the image features of the equipment operation status at different scales, and adds feature fusion layers at different levels of the multi-scale convolutional neural network to fuse feature maps of different scales. A global average pooling operation is used before the fully connected layer of the network to compress the spatial dimension of each feature map into a single value, thereby generating a high-dimensional feature vector of fixed length; Use principal component analysis to reduce the dimensionality of the extracted high-dimensional feature vectors; A contrastive learning strategy is adopted to optimize the high-dimensional feature vector after dimensionality reduction by constructing positive and negative sample pairs, and finally a visual feature vector is obtained.

6. The intelligent inspection method based on multi-dimensional data fusion analysis according to claim 5, characterized in that: The advanced fusion comprises the following steps: Combine the visual feature vectors of primary fusion data and remote viewing data, and use the covariance crossover algorithm to calculate the mean vector and covariance matrix after advanced fusion; In the process of advanced fusion, quantum heuristic algorithm is used to optimize advanced fusion parameters; The high-level fusion parameters optimized by the quantum heuristic algorithm are substituted into the covariance crossover algorithm, and the optimized mean vector and covariance matrix are calculated to obtain the high-level fusion data.

7. The intelligent inspection method based on multi-dimensional data fusion analysis according to claim 6, characterized in that: Evaluate equipment status based on advanced fusion data and generate inspection reports, including the following steps: Set the normal range of equipment status based on the equipment's historical data and operating specifications; Extract state parameters from the mean vector, compare the extracted state parameters with the normal range, and determine whether the device state is normal; Evaluate the uncertainty of state parameters based on the covariance matrix; According to the status parameters and uncertainty assessment results, the equipment status is divided into normal status, warning status and abnormal status; Integrate the assessed equipment status with the equipment status data to generate an inspection report.

8. An intelligent inspection system based on multidimensional data fusion analysis, based on the intelligent inspection method based on multidimensional data fusion analysis according to any one of claims 1 to 7, characterized in that: include, The data acquisition module is responsible for collecting the status data of the equipment from various management platforms, and the status data includes telemetry data, telesignaling data and televiewing data; The processing module is responsible for building a time series model based on telemetry data, calculating the residual between the predicted value and the observed value, and using the Dixon criterion to determine and remove outliers to obtain processed telemetry data; The primary fusion module is responsible for primary fusion of the processed telemetry data and telesignaling data to obtain primary fusion data; The feature extraction module is responsible for extracting visual feature vectors from remote viewing data using a convolutional neural network model; The advanced fusion module is responsible for performing advanced fusion of primary fusion data and visual feature vectors to obtain advanced fusion data; The report generation module is responsible for evaluating the equipment status based on advanced fusion data and finally generating an inspection report.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the intelligent inspection method based on multidimensional data fusion analysis described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the intelligent inspection method based on multidimensional data fusion analysis described in any one of claims 1 to 7 are implemented.

Images