Remote monitoring method, system and equipment and storage medium

Abstract

The invention relates to the technical field of remote monitoring, in particular to a remote monitoring method, system and device and a storage medium. The monitoring method comprises the following steps: acquiring multi-dimensional real-time data and historical data of a monitored object; calculating a comprehensive state index CS I of the monitoring object according to the real-time data and the historical data; judging whether the monitored object is in an abnormal state or not based on the comprehensive state index CS I; and if the monitoring object is judged to be in the abnormal state, multi-level alarm information is generated. According to the remote monitoring method provided by the invention, the comprehensive state index CS I is introduced, so that weighted fusion of multi-dimensional parameters is realized, and the accuracy and sensitivity of anomaly detection are improved.

Description

Technical Field

[0001] The present invention relates to the field of remote monitoring technology, and in particular to a remote monitoring method, system, equipment and storage medium. Background Art

[0002] In modern industrial and infrastructure management, remote monitoring systems have become a key technology to ensure the safe and efficient operation of equipment and systems. However, traditional remote monitoring methods often have problems such as single data dimension and low anomaly detection accuracy, which makes it difficult to meet the real-time monitoring needs of complex systems. Especially when faced with massive multidimensional data, existing technologies generally lack the ability to comprehensively analyze these data and cannot effectively identify and predict potential abnormal conditions. Therefore, the problem that needs to be solved in current remote monitoring technology is: how to achieve effective integration of multidimensional data to improve the accuracy of abnormal state detection. Summary of the invention

[0003] The present invention provides a remote monitoring method, system, device and storage medium to solve the technical problem of low detection accuracy of abnormal states in existing remote monitoring technologies.

[0004] The technical solution of the present invention to solve the above technical problems is as follows:

[0005] In one aspect, a remote monitoring method is provided, the monitoring method comprising the following steps:

[0006] Obtain multi-dimensional real-time and historical data of the monitored object;

[0007] Calculating a comprehensive status index CSI of the monitored object according to the real-time data and the historical data;

[0008] Based on the comprehensive status index CSI, determining whether the monitored object is in an abnormal state;

[0009] If it is determined that the monitored object is in an abnormal state, a multi-level alarm message is generated;

[0010] Sending the multi-level alarm information to a remote terminal device;

[0011] receiving a processing instruction sent by the remote terminal device;

[0012] According to the processing instruction, corresponding automatic adjustment or manual intervention operation is performed.

[0013] In another aspect, a remote monitoring system is provided, the monitoring system comprising:

[0014] Data acquisition module, used to obtain multi-dimensional real-time data and historical data of the monitored object;

[0015] A state calculation module, used to calculate the comprehensive state index CSI of the monitored object according to the real-time data and historical data;

[0016] An abnormality judgment module, used to judge whether the monitored object is in an abnormal state based on the comprehensive state index CSI;

[0017] An alarm generation module is used to generate multi-level alarm information when it is determined that the monitored object is in an abnormal state;

[0018] A communication module, used for sending the multi-level alarm information to a remote terminal device and receiving a processing instruction sent by the remote terminal device;

[0019] An execution module, used to execute corresponding automatic adjustment or manual intervention operations according to the processing instructions;

[0020] A prediction module is used to perform time series analysis on the comprehensive status index CSI to predict the status trend of the monitored object in the future;

[0021] An analysis module, used to perform cluster analysis on the comprehensive state indexes CSI of multiple monitored objects to obtain the state distribution of the monitored objects; and

[0022] The machine learning module is used to train anomaly detection models based on historical data to assist in determining whether the monitored object is in an abnormal state.

[0023] On the other hand, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the remote monitoring method as described above is implemented.

[0024] The beneficial effects of the present invention are:

[0025] The remote monitoring method proposed in the present invention realizes the weighted fusion of multi-dimensional parameters by introducing the comprehensive state index CSI, thereby improving the accuracy and sensitivity of anomaly detection. This method can not only detect abnormal states in real time, but also predict future trends based on time series analysis, providing managers with timely and effective decision support. Therefore, the present invention effectively solves the shortcomings of traditional remote monitoring methods in multi-dimensional data integration and abnormal state identification, greatly improves the efficiency and reliability of remote monitoring, and provides strong technical support for safe operations in the fields of industrial production, infrastructure management, etc.

[0026] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and implement it according to the contents of the specification, the following is a detailed description of the preferred embodiments of the present invention in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1A flowchart of a remote monitoring method in one embodiment of the present invention;

[0028] Figure 2 A logic diagram of a remote monitoring method according to an embodiment of the present invention;

[0029] Figure 3 A diagram showing the steps of calculating the comprehensive state index CSI of a monitored object in one embodiment of the present invention;

[0030] Figure 4 The process of expanding the steps of the remote monitoring method in one embodiment of the present invention is as follows Figure 1 ;

[0031] Figure 5 A flowchart of a prediction model predicting a monitoring object in one embodiment of the present invention;

[0032] Figure 6 The process of expanding the steps of the remote monitoring method in one embodiment of the present invention is as follows Figure 2 ;

[0033] Figure 7 A flowchart of an analysis model analysis method according to an embodiment of the present invention;

[0034] Figure 8 This is a flow chart of abnormality determination of a monitored object in one embodiment of the present invention;

[0035] Fig. 9 A flowchart of training an anomaly detection model in one embodiment of the present invention;

[0036] Fig.10 A block diagram of a remote monitoring system in one embodiment of the present invention;

[0037] In the accompanying drawings, the components represented by the reference numerals are listed as follows:

[0038] 100. Data acquisition module; 200. State calculation module; 300. Abnormal judgment module; 400. Alarm generation module; 500. Communication module; 600. Execution module; 700. Prediction module; 800. Analysis module; 900. Machine learning module. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. In addition, it should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0040] The disclosure below provides many different embodiments or examples for implementing different structures of the present invention. In order to simplify the disclosure of the present invention, the components and settings of specific examples are described below. Of course, they are merely examples and are not intended to limit the present invention. In addition, the present invention may repeat reference numbers and / or reference letters in different examples, and such repetition is for the purpose of simplicity and clarity, and does not itself indicate the relationship between the various embodiments and / or settings discussed. In addition, the present invention provides examples of various specific brands, models and series, but those of ordinary skill in the art may be aware of the use scenarios of other brands, models and series.

[0041] The present invention provides the following preferred embodiments:

[0042] Embodiment 1

[0043] In order to solve the shortcomings of traditional remote monitoring systems in processing multi-dimensional data and anomaly detection, this embodiment proposes a remote monitoring method. The method first obtains multi-dimensional real-time data and historical data of the monitored object to lay the foundation for subsequent analysis. These data may include but are not limited to various physical quantities such as temperature, pressure, vibration, current, as well as system logs, operation records and other information.

[0044] like Figure 1 , Figure 2 As shown, the monitoring method includes the following steps:

[0045] S100. Acquire multi-dimensional real-time data and historical data of the monitored object.

[0046] S200: Calculate the comprehensive status index CSI of the monitored object based on real-time data and historical data.

[0047] S300: Based on the comprehensive status index CSI, determine whether the monitored object is in an abnormal state.

[0048] S400: If it is determined that the monitored object is in an abnormal state, a multi-level alarm message is generated.

[0049] S500. Send multi-level alarm information to a remote terminal device.

[0050] S600: Receive a processing instruction sent by a remote terminal device.

[0051] S700: Execute corresponding automatic adjustment or manual intervention operations according to the processing instruction.

[0052] Furthermore, this method introduces the concept of comprehensive status index (CSI), which is an index that can fully reflect the current status of the monitored object by performing complex mathematical calculations and statistical analysis on real-time data and historical data. It should be understood that the calculation of CSI may involve the weighted fusion of multiple parameters, and the weight of each parameter can be dynamically adjusted according to its impact on the overall status of the system.

[0053] Furthermore, based on the calculated CSI, the method can determine whether the monitored object is in an abnormal state. It is understandable that this judgment process may involve multiple technical means such as setting thresholds, using statistical models or machine learning algorithms. When an abnormal state is detected, the system will generate multiple levels of alarm information. These alarm information may be graded according to factors such as the severity of the abnormality and the potential impact range, so that managers can quickly identify and deal with the most urgent problems.

[0054] Furthermore, the method sends the alarm information to a remote terminal device. This step ensures that managers can obtain monitoring information anytime and anywhere, greatly improving response speed and decision-making efficiency. The remote terminal device can be a smart phone, tablet computer or dedicated monitoring terminal, which receives real-time alarms through a secure network connection.

[0055] Furthermore, the method also includes receiving processing instructions sent by the remote terminal device. This two-way communication mechanism enables the management personnel to respond in time and send corresponding processing instructions based on the received alarm information. The processing instructions may include various operations such as adjusting equipment parameters, starting backup systems, and arranging on-site maintenance.

[0056] Furthermore, the system performs corresponding automatic adjustments or manual intervention operations according to the received processing instructions. It should be understood that automatic adjustments may involve complex control algorithms and actuators, while manual intervention may require the cooperation of on-site personnel. This step closely combines remote monitoring with actual operations to form a closed-loop management system.

[0057] The benefit of this embodiment is that, through the technical solution of this embodiment, a comprehensive and accurate assessment and timely response to the monitored object are achieved. The remote monitoring method of this embodiment not only improves the accuracy of anomaly detection, but also enhances the controllability of the system through remote operation. The monitoring solution of the present invention is particularly suitable for complex industrial environments, critical infrastructure and other fields that require high reliability and real-time response, providing a strong guarantee for improving operational efficiency and safety.

[0058] Embodiment 2

[0059] In order to solve the problem that traditional monitoring systems lack comprehensive evaluation capabilities when processing multi-dimensional data, this embodiment further optimizes the calculation method of the comprehensive status index CSI and provides a more accurate and comprehensive monitoring object status evaluation mechanism. This method achieves a quantitative description of the current status of the monitored object by integrating the real-time data of multiple monitoring parameters and considering their historical performance and relative importance.

[0060] like Figure 3 As shown, calculating the comprehensive state index CSI of the monitored object includes the following steps:

[0061] S201. Obtain real-time values ​​Pi, reference values ​​Bi, standard deviations Si, and weights Wi of multiple monitoring parameters.

[0062] S202. For each monitoring parameter, calculate its normalized deviation value (Pi-Bi) / Si.

[0063] S203: Multiply the normalized deviation value of each monitoring parameter by its corresponding weight Wi.

[0064] S204: Sum all weighted normalized deviation values ​​to obtain a comprehensive state index CSI.

[0065] The formula is:

[0066] Wherein, Wi is the weight of the ith monitoring parameter, Pi is the real-time value of the ith monitoring parameter, Bi is the reference value of the ith monitoring parameter, and Si is the standard deviation of the ith monitoring parameter.

[0067] Furthermore, this embodiment elaborates on the calculation process of CSI. First, the system obtains the key values ​​of multiple monitoring parameters, including real-time value Pi, reference value Bi, standard deviation Si and weight Wi. It should be understood that the real-time value Pi reflects the current state of the parameter, the reference value Bi represents the normal operating level of the parameter, the standard deviation Si quantifies the fluctuation range of the parameter, and the weight Wi reflects the degree of influence of different parameters on the overall state.

[0068] Further, for each monitored parameter, the system calculates its normalized deviation value (Pi-Bi) / Si. It can be understood that this step achieves the standardization between parameters of different dimensions, so that they can be compared and integrated on the same scale. By dividing the difference between the real-time value and the baseline value by the standard deviation, a dimensionless indicator is obtained, which reflects the degree to which the current parameter deviates from the normal state.

[0069] Furthermore, the system multiplies the normalized deviation value of each parameter by its corresponding weight Wi. This step reflects the differentiated contribution of different parameters in the overall evaluation. It should be understood that the setting of weights may be based on experience, historical data analysis, or the output of machine learning algorithms, which reflects the degree of influence of each parameter on the overall state of the system.

[0070] Furthermore, the system sums all weighted normalized deviation values ​​to obtain the comprehensive status index CSI. This step combines the status information of multiple parameters into a single, comprehensive indicator. It should be understood that the calculation formula of CSI not only takes into account the current status of each parameter, but also reflects the relative importance and comparability of parameters through weighting and normalization.

[0071] Furthermore, by adjusting the weight Wi, the system can adapt to different types of monitoring objects and different operating scenarios. For example, in some cases, the temperature parameter may be more critical than the pressure parameter. At this time, the importance can be reflected by increasing the weight of the temperature parameter. Through the CSI calculation method described in this embodiment, accurate quantification and comprehensive evaluation of the state of the monitored object are achieved. This method can not only reflect the abnormal state of the monitored object in a timely manner, but also capture subtle change trends, providing strong support for predictive maintenance and risk management. In addition, since CSI is a standardized indicator, it is convenient for comparison and analysis across systems and equipment, and provides a unified evaluation standard for large-scale industrial monitoring and management.

[0072] The benefit of this embodiment is that it provides a flexible and comprehensive status assessment method that can effectively improve the accuracy and reliability of the monitoring system. By converting complex multi-dimensional data into a single comprehensive indicator, the decision-making process is greatly simplified, allowing managers to quickly and accurately judge the system status and take necessary intervention measures in a timely manner.

[0073] Embodiment 3

[0074] In order to solve the problem that traditional remote monitoring systems can only respond to current anomalies but lack predictive capabilities, this embodiment further optimizes the monitoring method and introduces a time series analysis and prediction mechanism based on the comprehensive status index CSI. This innovative improvement enables the system to not only monitor the current status in real time, but also to proactively predict possible abnormal situations in the future, thereby providing managers with more time and space for decision-making.

[0075] like Figure 4 As shown, the monitoring method also includes the following steps:

[0076] S10. Perform time series analysis on the comprehensive status index CSI, and predict the status trend of the monitored object in the future through a prediction model.

[0077] S20: If it is predicted that the monitored object may be abnormal in the future, a warning message is generated.

[0078] S30: Send the warning information to the remote terminal device.

[0079] Furthermore, this embodiment elaborates on the process of time series analysis of the comprehensive status index CSI. It should be understood that time series analysis is a statistical method used to study data series that change over time and extract meaningful statistical features and patterns from them. In this embodiment, the CSI value is the main research object of the time series, and by analyzing its historical change trend, periodic fluctuations, abnormal fluctuations and other characteristics, it provides key input for the subsequent prediction model.

[0080] It is understandable that the selection and construction of the prediction model is one of the core links of this embodiment. According to different application scenarios and data characteristics, a variety of prediction algorithms can be used, such as autoregressive integrated moving average model (ARIMA), exponential smoothing method, neural network model, etc. These models can predict the CSI value in the future by learning the historical data pattern of CSI. It should be noted that the parameters of the prediction model need to be optimized and adjusted according to the actual situation to ensure the accuracy and reliability of the prediction results.

[0081] Furthermore, the mechanism for generating warning information based on the prediction results of this embodiment will continuously compare the predicted CSI value with the preset normal range threshold, and once the predicted value exceeds this range, the warning mechanism will be triggered. It can be understood that the generation of warning information not only considers the absolute size of the predicted value, but may also include multi-dimensional information such as abnormal duration and change rate, so as to more comprehensively describe potential abnormal situations.

[0082] Furthermore, this embodiment sends the warning information to the remote terminal device in a timely manner to ensure that the management personnel can obtain relevant information before the abnormal event occurs, so that there is enough time to formulate a response strategy. It should be understood that the transmission of warning information should adopt a safe and reliable communication protocol to ensure the timeliness and integrity of the information. At the same time, the remote terminal device may need to be equipped with a special warning information processing module to classify, prioritize and visualize the received information.

[0083] The benefit of this embodiment is that it greatly improves the predictability and initiative of the remote monitoring system. By performing time series analysis and prediction on CSI, the system can identify potential abnormal trends in advance, providing strong support for preventive maintenance and risk management. This forward-looking monitoring method can not only reduce the occurrence of sudden failures, but also optimize maintenance plans and reduce operating costs. In addition, by sending early warning information to remote terminal devices, this method realizes the seamless connection between the monitoring system and management decision-making, enabling managers to make more timely and accurate decisions based on data-driven prediction results.

[0084] Embodiment 4

[0085] In order to solve the problem of insufficient accuracy and poor adaptability of traditional time series prediction models in complex industrial environments, this embodiment further optimizes the prediction method based on the prediction model and provides a more accurate and flexible monitoring object state trend prediction mechanism. This method achieves high-precision prediction of the future state of the monitored object through detailed parameter determination and model building process.

[0086] like Figure 5 As shown, predicting the state trend of the monitored object in the future by using the prediction model includes the following steps:

[0087] S11, determining the order p, d, q of the prediction model;

[0088] S12, obtaining historical time series data Yt-1, Yt-2, ..., Yt-p;

[0089] S13, calculating the time series after difference;

[0090] S14, estimating model parameters, including the constant term c, autoregressive parameters φ1, φ2, ..., φp, and moving average parameters θ1, θ2, ..., θq;

[0091] S15, calculating white noise terms εt, εt-1, εt-2, ..., εt-q;

[0092] S16, substituting the above parameters into the equation of the prediction model to calculate the predicted value Yt at time t;

[0093] The equation of the prediction model is:

[0094] Yt=c+φ1Yt-1+φ2Yt-2+...+φpYt-p+εt+θ1εt-1+θ2εt-2+...+θqεt-q,

[0095] Among them, Yt is the predicted value at time t, c is a constant term, φp and θq are autoregressive and moving average parameters, and εt is white noise.

[0096] Further, this embodiment elaborates on the construction and application process of the prediction model. First, the system needs to determine the three parameters of the prediction model: the number of autoregressive terms p, the number of differences d, and the number of moving average terms q. It should be understood that the selection of these parameters directly affects the performance and prediction accuracy of the model. Generally, the optimal p, d, and q values ​​can be determined by combining statistical indicators such as AIC (Akaike Information Criterion) or BIC (Bayesian Information Criterion) through autocorrelation function (ACF) and partial autocorrelation function (PACF) analysis.

[0097] Next, the system obtains historical time series data Yt-1, Yt-2, ..., Yt-p. It is understandable that these historical data are the basis for model training, and their quality and quantity directly affect the accuracy of prediction. In order to improve the stability of the model, it may be necessary to preprocess the original data, such as removing outliers and filling missing values.

[0098] Furthermore, an important step of this embodiment is to calculate the time series after differentiation. The purpose of the differentiation operation is to make the time series reach a stable state, which is the basic assumption of the prediction model. It should be noted that the selection of the number of differences d needs to be cautious. Excessive differentiation may lead to information loss, while insufficient differentiation may not eliminate the non-stationarity of the series.

[0099] Further, the system estimates the model parameters, including the constant term c, the autoregressive parameters φ1, φ2, ..., φp, and the moving average parameters θ1, θ2, ..., θq. This step usually uses statistical methods such as maximum likelihood estimation or conditional least squares. It should be understood that the accuracy of parameter estimation directly affects the predictive ability of the model, so it may be necessary to use an iterative optimization algorithm to obtain the optimal parameter values.

[0100] Furthermore, the system calculates white noise terms εt, εt-1, εt-2, …, εt-q. White noise terms represent random fluctuations that cannot be explained by the model, and their distribution characteristics are crucial for the diagnosis and validation of the model. It is understandable that, ideally, white noise terms should be normally distributed and independent of each other.

[0101] Furthermore, the system substitutes the above parameters into the equation of the prediction model to calculate the predicted value Yt at time t. The equation of the prediction model comprehensively considers the autoregressive effect of historical data, the trend change after differential processing, and the moving average effect of random fluctuations, so as to make a comprehensive prediction of the future state.

[0102] Furthermore, this embodiment introduces a model diagnosis and dynamic adjustment mechanism. By analyzing the distribution characteristics of the prediction error, the system can continuously evaluate the applicability of the model and automatically adjust the model parameters or reselect the model order when necessary. This adaptive capability enables the prediction system to cope with dynamic changes in the state of the monitored object and maintain long-term prediction accuracy.

[0103] The benefit of this embodiment is that it provides a highly accurate and adaptive state trend prediction method. Through detailed parameter estimation and model building process, this method can capture subtle change trends in complex industrial environments and provide reliable data support for predictive maintenance and risk management. At the same time, the adaptive mechanism of the model ensures the stability and reliability of the prediction system in long-term operation, greatly reduces the need for manual intervention, and improves the automation level of the monitoring system.

[0104] Embodiment 5

[0105] In order to solve the problem that traditional remote monitoring systems lack the ability to identify systematic problems when handling large-scale multi-object monitoring, this embodiment further optimizes the monitoring method and introduces a multi-object state distribution analysis mechanism based on the comprehensive state index CSI. This innovative improvement enables the system to not only evaluate the state of each monitored object individually, but also identify potential systematic problems from an overall perspective, thereby providing more comprehensive and in-depth insights for management decisions.

[0106] like Figure 6 As shown, the monitoring method also includes the following steps:

[0107] S50, analyzing the comprehensive state index CSI of multiple monitored objects through an analysis model to obtain state distribution of the monitored objects.

[0108] S60. Identify potential systemic problems based on state distribution.

[0109] Furthermore, this embodiment elaborates on the process of analyzing the comprehensive status index CSI of multiple monitored objects. It should be understood that the analysis model here involves a variety of advanced statistical and machine learning techniques, such as principal component analysis (PCA), cluster analysis, anomaly detection algorithms, etc. The selection and combination of these techniques depends on the specific application scenario and data characteristics. For example, when processing high-dimensional CSI data, PCA can be used to reduce the dimension first to extract the most representative features; then, K-means or hierarchical clustering algorithms can be applied to group the reduced-dimensional data to identify a set of objects with similar status characteristics.

[0110] It is understandable that the acquisition of state distribution is one of the core links of this embodiment. This distribution not only includes the statistical distribution of CSI values ​​(such as mean, variance, skewness, kurtosis, etc.), but may also involve the distribution characteristics of CSI values ​​in time and space dimensions. For example, the system may analyze the time series pattern of CSI values ​​to identify periodic fluctuations or sudden changes; at the same time, it may also consider the positional relationship of the monitored objects in the physical space or logical topology, and analyze the spatial correlation and propagation pattern of CSI values.

[0111] Furthermore, the present embodiment is a mechanism for identifying potential systemic problems based on state distribution. The "systemic problems" here refer to common factors or potential risks that affect the performance of multiple monitored objects or the entire system. It should be understood that the identification of systemic problems usually requires a combination of domain knowledge and advanced analysis techniques. For example, by analyzing the spatial clustering pattern of CSI values, it may be found that multiple devices in a certain geographic area have performance degradation at the same time, which may indicate the influence of environmental factors; for example, by analyzing the time synchronization of CSI values, it may be found that multiple seemingly independent devices have abnormalities at the same time at a specific time point, which may indicate potential problems in the upstream supply chain or control system.

[0112] Furthermore, this embodiment emphasizes the importance of converting identified systemic problems into actionable insights. The system may automatically generate problem reports containing problem descriptions, impact scope, possible causes, and recommended actions. These reports not only provide quantitative analysis of the problem, but may also be presented to decision makers in an easily understandable form through natural language generation technology.

[0113] The benefit of this embodiment is that it greatly improves the analysis depth and decision support capabilities of the remote monitoring system. By systematically analyzing the CSI of multiple monitored objects, managers can obtain a more comprehensive view of the system's operating status, thereby making more accurate and timely decisions. This method can not only identify anomalies of individual objects, but also discover potential systemic risks, providing strong support for preventive maintenance and system optimization.

[0114] Embodiment 6

[0115] In order to solve the problem that traditional monitoring systems lack effective clustering analysis capabilities when processing large-scale multi-object comprehensive status index (CSI) data, this embodiment further optimizes the analysis model. This improvement not only improves the depth of understanding of the status distribution of monitored objects, but also provides strong data support for identifying potential systemic problems.

[0116] like Figure 7 As shown, analyzing the comprehensive state index CSI of multiple monitoring objects through the analysis model includes the following steps:

[0117] S51. Initialize K cluster centers μj.

[0118] S52: For each data point xi, calculate its distance to the center of each cluster and assign it to the cluster with the closest distance.

[0119] S53, recalculate the center μj of each cluster.

[0120] S54. Calculate the sum of the squares of the distances from all data points to the centers of their clusters.

[0121] S55. Repeat the above steps until the cluster division does not change or the maximum number of iterations is reached.

[0122] S56. Output the final cluster division result and objective function value J.

[0123] The function of the analytical model is expressed as:

[0124] J=∑∑||xi-μj|| 2 , where xi is the i-th data point and μj is the center of the j-th cluster.

[0125] Furthermore, this embodiment elaborates on the application process of the analysis model in CSI data analysis. First, the system needs to initialize K cluster centers μj. It should be understood that the selection of the initial cluster center has an important influence on the convergence speed and final result of the algorithm. In order to optimize this step, an improved initialization method can be used to select the initial center point through probability distribution to improve the quality and stability of clustering.

[0126] Further, the algorithm enters the iteration phase. For each data point xi (representing the CSI value of each monitored object in this embodiment), the system calculates its distance to the center of each cluster. It can be understood that the choice of distance metric has a significant impact on the clustering results. Although Euclidean distance is the most commonly used metric, in some application scenarios, alternatives such as Manhattan distance or Mahalanobis distance may need to be considered to better capture the characteristics of the data.

[0127] Furthermore, recalculating the center μj of each cluster ensures that the cluster center can be dynamically adjusted to better reflect the distribution characteristics of the data. It should be noted that when processing large-scale data, batch processing or incremental update can be considered to improve computational efficiency.

[0128] Furthermore, the system calculates the sum of the squares of the distances from all data points to the centers of their clusters, which constitutes the objective function J of the analysis model.

[0129] Furthermore, this embodiment emphasizes that the iterative process algorithm will repeat the allocation and update steps until the cluster division no longer changes or the preset maximum number of iterations is reached. It should be understood that the setting of the convergence condition requires a trade-off between computational efficiency and clustering accuracy. In order to improve the stability of the algorithm, an early stopping strategy can be introduced, that is, the iteration is terminated early when the improvement of the objective function is lower than a certain threshold for multiple consecutive times.

[0130] Furthermore, after outputting the final clustering results and the objective function value J, the system will further analyze the characteristics of each cluster, including the statistical distribution of the CSI values ​​within the cluster, the time change trend, and the relative relationship with other clusters. This in-depth analysis helps to identify groups of monitored objects with similar status characteristics, providing important clues for the diagnosis of systemic problems.

[0131] Furthermore, this embodiment considers the problem of K value selection. The selection of K value directly affects the effect and interpretability of clustering. In order to optimize the K value, the system may use methods such as the Elbow Method or the Silhouette Coefficient to determine the optimal number of clusters by running clustering with different K values ​​multiple times and comparing the results.

[0132] The benefit of this embodiment is that it provides an efficient and interpretable method to analyze the state distribution of large-scale monitoring objects. Through the analysis model, the system can automatically identify groups of monitoring objects with similar state characteristics, providing managers with a concise and comprehensive view of the system operation status. This cluster analysis can not only help identify monitoring objects in abnormal states, but also reveal potential systemic problems and operating modes, providing data-driven decision support for preventive maintenance and system optimization.

[0133] Embodiment 7

[0134] In order to solve the problem that traditional remote monitoring systems lack accuracy and foresight in abnormal state judgment, this embodiment further optimizes the abnormal state judgment mechanism and introduces an innovative method that combines digital twin technology with machine learning algorithms. This method not only improves the accuracy of abnormality detection, but also gives the system the ability to predict potential problems, thereby achieving a major shift from passive response to active prevention.

[0135] like Figure 8 As shown, determining whether the monitored object is in an abnormal state includes the following steps:

[0136] S301. Establish a digital twin model of the monitored object.

[0137] S302: Input real-time data into the digital twin model to simulate the operating status of the monitored object.

[0138] S303, comparing the simulation result with the actual operating status.

[0139] S304. Use a machine learning algorithm to train an anomaly detection model based on historical data.

[0140] S305: Use an anomaly detection model to assist in determining whether the monitored object is in an abnormal state.

[0141] Furthermore, this embodiment elaborates on the process of establishing a digital twin model of a monitored object. It should be understood that a digital twin model is a high-fidelity virtual representation of a physical entity in the digital world. The construction of such a model usually involves the integration of multidisciplinary knowledge, including but not limited to physical modeling, mathematical simulation, data analysis, etc. For example, for complex industrial equipment, it may be necessary to combine advanced simulation technologies such as computational fluid dynamics (CFD) and finite element analysis (FEA) to capture the dynamic behavior of the equipment. At the same time, the model also needs to consider multi-dimensional information such as the geometric characteristics, material properties, and operating parameters of the equipment to ensure that the virtual model can accurately reflect the characteristics and behavior of the physical entity.

[0142] Furthermore, a key link in this embodiment is to input real-time data into the digital twin model to simulate the operating status of the monitored object. It is understandable that this step requires the establishment of an efficient data transmission and processing mechanism. The system may need to adopt edge computing technology to achieve real-time data processing and rapid response of the model. At the same time, in order to ensure the accuracy of the simulation, data preprocessing and calibration steps may need to be implemented, such as denoising, outlier processing, data completion, etc.

[0143] Furthermore, the system compares the simulation results with the actual operating status. This step involves not only simple numerical comparisons, but may also include complex time series analysis and pattern recognition. For example, the dynamic time warping (DTW) algorithm can be used to evaluate the similarity between the simulated time series and the actual time series, or the spectrum analysis technique can be used to compare the vibration characteristics. It should be noted that the comparison process should take into account the measurement error and the model error, and avoid overly sensitive judgments by setting an appropriate tolerance range.

[0144] Furthermore, this embodiment uses a machine learning algorithm to train an anomaly detection model based on historical data. The machine learning algorithm here may include but is not limited to support vector machines (SVM), random forests, deep learning networks, etc. It should be understood that the choice of model should be based on the characteristics of the data and the complexity of the problem. For example, for high-dimensional nonlinear data, it is possible to consider using a deep autoencoder or a variational autoencoder to learn the implicit representation of the data, so as to more effectively detect anomalies. At the same time, in order to improve the generalization ability and robustness of the model, an integrated learning method such as Bagging or Boosting technology can be used.

[0145] Furthermore, the system uses the trained anomaly detection model to assist in determining whether the monitored object is in an abnormal state. This step not only includes the judgment of the current state, but may also involve the prediction of the future state. For example, a recurrent neural network structure such as a long short-term memory network (LSTM) or a gated recurrent unit (GRU) can be used to capture the long-term dependencies of time series data, thereby predicting potential abnormal trends.

[0146] Furthermore, this embodiment realizes the collaborative work of the digital twin model and the machine learning model. The digital twin model provides simulation results based on physical principles, while the machine learning model supplements data-driven insights. This combination not only improves the accuracy of anomaly detection, but also enhances the system's ability to identify new or rare anomalies.

[0147] The benefit of this embodiment is that it provides a comprehensive, accurate and predictive abnormal state judgment method. Through digital twin technology, the system can simulate and predict the behavior of the monitored object in a virtual environment, while the machine learning algorithm gives the system the ability to learn from historical data. This method can not only accurately identify known types of abnormalities, but also discover potential unknown abnormal patterns, providing strong support for preventive maintenance and risk management.

[0148] Embodiment 8

[0149] In order to solve the limitations of traditional anomaly detection models in processing high-dimensional nonlinear data and improve the generalization and robustness of the model in complex monitoring environments, this embodiment further optimizes the anomaly detection mechanism based on support vector machines (SVM). This improvement not only improves the accuracy and efficiency of anomaly detection, but also enhances the model's ability to identify new or rare anomalies, thereby providing more reliable decision support for remote monitoring systems.

[0150] like Fig. 9 As shown, using machine learning algorithms, training anomaly detection models based on historical data includes the following steps:

[0151] S3041. Obtain the parameters of the trained anomaly detection model, including the Lagrange multiplier αi, the category label yi, the kernel function K(xi,x) and the bias term b.

[0152] S3042. For the input sample x, calculate the kernel function value K(xi,x) between it and all support vectors xi.

[0153] S3043. Multiply the kernel function value with the corresponding αi and yi, and sum all the results.

[0154] S3044. Add the bias term b to the summation result.

[0155] S3045. Take the symbol function of the final result to obtain the classification result. The decision function of the machine learning algorithm is expressed as:

[0156] f(x)=sign(∑αi*yi*K(xi,x)+b).

[0157] Furthermore, this embodiment elaborates on the training and application process of the SVM anomaly detection model. First, the system needs to obtain the key parameters of the trained anomaly detection model. It should be understood that these parameters include the Lagrange multiplier αi, the category label yi, the kernel function K(xi,x) and the bias term b, which together constitute the core of the SVM model. The Lagrange multiplier αi reflects the contribution of each support vector to the decision boundary, while the category label yi indicates the category (normal or abnormal) to which each support vector belongs. The kernel function K(xi,x) is the key to SVM in dealing with nonlinear problems. It maps the input space to a high-dimensional feature space, making the originally linearly inseparable data linearly separable in the high-dimensional space.

[0158] Further, the kernel function value K(xi,x) of the input sample x and all support vectors xi is calculated. It can be understood that the choice of kernel function has a significant impact on the performance of the model. Commonly used kernel functions include linear kernel, polynomial kernel, radial basis function (RBF) kernel, etc. In practical applications, RBF kernel is widely used due to its good versatility and ability to handle nonlinear problems. The mathematical expression of the kernel function is: K(xi,x).

[0159] Furthermore, the system multiplies the kernel function value with the corresponding αi and yi, and sums all the results. This step actually calculates the position of the input sample x in the feature space relative to the decision hyperplane. It should be noted that only support vectors (i.e. samples with αi>0) will affect the final decision, which reflects the sparsity feature of SVM and helps improve the computational efficiency and generalization ability of the model.

[0160] An innovative point of this embodiment is to add a bias term b to the summation result. The introduction of the bias term b allows the decision hyperplane to be freely translated in the feature space, thereby finding the optimal classification position. It can be understood that the value of b is determined by the optimization algorithm during the model training process, and it determines the final decision boundary together with the support vector and the Lagrange multiplier.

[0161] Furthermore, the system takes the sign function of the final result to obtain the classification result. The role of the sign function is to convert the continuous output value into a discrete category label, usually represented by sign(x). When the result is greater than 0, the sample is classified as normal; when the result is less than 0, the sample is classified as abnormal; when the result is equal to 0, the sample falls exactly on the decision boundary, which is rare in practical applications.

[0162] Furthermore, the decision function of the machine learning algorithm integrates the information of all support vectors and realizes linear classification in high-dimensional feature space, thereby effectively solving the nonlinear problem in anomaly detection.

[0163] The benefit of this embodiment is that it provides an efficient, explainable and powerful anomaly detection method. Through SVM technology, the system can construct an optimal classification hyperplane in a high-dimensional feature space and effectively handle complex nonlinear data distribution. This method can not only accurately identify known types of anomalies, but also discover potential unknown anomaly patterns through nonlinear mapping of kernel functions, providing a more comprehensive and reliable anomaly detection capability for the monitoring system.

[0164] Embodiment 9

[0165] In order to solve the limitations of traditional remote monitoring systems in data processing, anomaly detection, predictive analysis, and intelligent decision-making, this embodiment further optimizes an integrated and intelligent remote monitoring system architecture. This innovative architecture not only improves the real-time and accuracy of the system, but also enhances the system's predictability and adaptability, thereby achieving a shift from passive response to active prevention.

[0166] like Fig.10 As shown, the monitoring system includes:

[0167] The data acquisition module 100 is used to obtain multi-dimensional real-time data and historical data of the monitored object.

[0168] The state calculation module 200 is used to calculate the comprehensive state index CSI of the monitored object based on real-time data and historical data.

[0169] The abnormality judgment module 300 is used to judge whether the monitored object is in an abnormal state based on the comprehensive state index CSI.

[0170] The alarm generation module 400 is used to generate multi-level alarm information when it is determined that the monitored object is in an abnormal state.

[0171] The communication module 500 is used to send multi-level alarm information to a remote terminal device and receive a processing instruction sent by the remote terminal device.

[0172] The execution module 600 is used to execute corresponding automatic adjustment or manual intervention operations according to the processing instructions.

[0173] The prediction module 700 is used to perform time series analysis on the comprehensive status index CSI and predict the status trend of the monitored object in the future.

[0174] The analysis module 800 is used to perform cluster analysis on the comprehensive state indexes CSI of multiple monitored objects to obtain the state distribution of the monitored objects.

[0175] The machine learning module 900 is used to train an anomaly detection model based on historical data to assist in determining whether the monitored object is in an abnormal state.

[0176] Furthermore, this embodiment elaborates on the core module composition and functions of the remote monitoring system. First, the data acquisition module 100, as the basis of the system, is responsible for acquiring multi-dimensional real-time data and historical data of the monitored object. It should be understood that this module is not just a simple data collection, but also includes data preprocessing, verification and preliminary screening. For example, edge computing technology can be used to perform preliminary processing at the source of the data to reduce the amount of data transmission and increase the system response speed. At the same time, the module may also integrate data quality control mechanisms, such as outlier detection, data completion, etc., to ensure the reliability of subsequent analysis.

[0177] Furthermore, the state calculation module 200 is responsible for calculating the comprehensive state index CSI of the monitored object based on real-time data and historical data. It is understandable that the calculation of CSI involves complex multi-dimensional data fusion and weight distribution problems. This module may use advanced data fusion algorithms, such as Kalman filtering or Bayesian networks, to achieve effective integration of data from different sources and different characteristics. At the same time, the weight distribution may adopt a dynamic adjustment mechanism to adaptively adjust the importance of each indicator according to the system operation status and environmental changes.

[0178] Furthermore, the abnormality judgment module 300 judges whether the monitored object is in an abnormal state based on the calculated CSI. This module may integrate multiple abnormality detection algorithms, such as statistical process control (SPC), cluster analysis, deep learning, etc., to adapt to different types of abnormal patterns. It should be noted that this module may adopt a multi-model fusion method to improve the accuracy of judgment by integrating the results of multiple abnormality detection models.

[0179] Furthermore, the alarm generation module 400 generates multi-level alarm information when an abnormal state is detected. The innovation of this module may lie in its intelligent alarm classification mechanism. For example, it may use fuzzy logic or decision tree algorithms to dynamically determine the alarm level based on multiple factors such as the severity of the abnormality, the potential impact range, and the urgency, thereby avoiding excessive alarms or missing important alarms.

[0180] Furthermore, the communication module 500 is responsible for sending multi-level alarm information to the remote terminal device and receiving processing instructions. This module may use advanced network communication protocols, such as MQTT or CoAP, to ensure reliable communication even in unstable network environments. At the same time, it may also integrate data encryption and identity authentication mechanisms to ensure the security of communication.

[0181] Further, the execution module 600 performs corresponding automatic adjustment or manual intervention operations according to the received processing instructions. The design of this module may take into account complex control theories, such as model predictive control (MPC) or adaptive control, to achieve precise adjustment of the monitored object. At the same time, it may also include a safety check mechanism to prevent potential dangerous operations.

[0182] Furthermore, the prediction module 700 performs time series analysis on the CSI to predict the future state trend of the monitored object. This module may use advanced time series analysis techniques, such as long short-term memory network (LSTM) or prophet algorithm, to capture long-term trends, periodic changes and emergencies in the data.

[0183] Furthermore, the early warning module generates early warning information when a possible anomaly is predicted. This module may be combined with a risk assessment model to consider not only the probability of anomaly occurrence but also its potential impact, thereby achieving more intelligent and targeted early warning.

[0184] Furthermore, the analysis module 800 analyzes the CSIs of the multiple monitored objects to obtain the state distribution of the monitored objects. This module may use high-dimensional data visualization technology, such as t-SNE or UMAP, to intuitively display complex state distribution.

[0185] The machine learning module 900 trains an anomaly detection model based on historical data to assist in determining abnormal conditions. This module may use advanced transfer learning or federated learning technology to solve the problem of inconsistent data distribution between different monitoring objects and improve the generalization ability of the model.

[0186] The benefit of this embodiment is that it provides a comprehensive, integrated, and intelligent remote monitoring system solution. Through the collaborative work of multiple functional modules, the system can achieve full-process automated management from data collection, status assessment, anomaly detection to predictive analysis and intelligent decision-making. This method not only greatly improves the accuracy and efficiency of monitoring, but also realizes the transformation from passive response to active prevention, providing strong technical support for enterprises to reduce operational risks and optimize resource allocation.

[0187] Embodiment 10

[0188] In order to solve the deficiencies of traditional remote monitoring systems in data processing efficiency, anomaly detection accuracy, and prediction capabilities, this embodiment further optimizes a computer program stored in a computer-readable storage medium, and implements a fully integrated remote monitoring method through the execution of a processor. This method combines advanced data acquisition, state calculation, anomaly detection, and prediction analysis technologies to provide users with an efficient, accurate, and intelligent monitoring solution.

[0189] Through the computer program of this embodiment, a comprehensive, efficient and accurate remote monitoring system can be realized. The program is stored in a computer-readable storage medium, and through the execution of the processor, it integrates multiple functional modules such as data acquisition, state evaluation, anomaly detection, predictive analysis and intelligent decision-making. It not only improves the real-time and accuracy of the system, but also provides users with a means of active prevention and intelligent management, greatly improving the overall performance and reliability of the monitoring system. Through this advanced monitoring method, enterprises can achieve accurate monitoring and predictive maintenance in a complex and changing environment, providing solid technical support for optimizing resource allocation and reducing operational risks.

[0190] The beneficial effects of the present invention are specifically embodied in that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A remote monitoring method, characterized in that: The monitoring method comprises the following steps: Obtain multi-dimensional real-time and historical data of the monitored object; Calculating a comprehensive status index CSI of the monitored object according to the real-time data and the historical data; Based on the comprehensive status index CSI, determining whether the monitored object is in an abnormal state; If it is determined that the monitored object is in an abnormal state, a multi-level alarm message is generated; Sending the multi-level alarm information to a remote terminal device; receiving a processing instruction sent by the remote terminal device; According to the processing instruction, corresponding automatic adjustment or manual intervention operation is performed.

2. The remote monitoring method according to claim 1, characterized in that: Calculating the comprehensive status index CSI of the monitored object comprises the following steps: Obtain the real-time value Pi, reference value Bi, standard deviation Si and weight Wi of multiple monitoring parameters; For each monitoring parameter, calculate its normalized deviation value (Pi-Bi) / Si; Multiply the normalized deviation value of each monitoring parameter by its corresponding weight Wi; All weighted normalized deviation values ​​are summed to obtain the comprehensive state index CSI; The formula is: Among them, Wi is the weight of the i-th monitoring parameter, Pi is the real-time value of the i-th monitoring parameter, Bi is the reference value of the i-th monitoring parameter, and Si is the standard deviation of the i-th monitoring parameter.

3. The remote monitoring method according to claim 1, characterized in that: The monitoring method further comprises the following steps: Performing time series analysis on the comprehensive status index CSI, and predicting the status trend of the monitored object in the future through a prediction model; If it is predicted that the monitored object may be abnormal in the future, an early warning message will be generated; The warning information is sent to a remote terminal device.

4. The remote monitoring method according to claim 3, characterized in that: The method of predicting the state trend of the monitored object in the future period of time by using the prediction model includes the following steps: Determining the order p, d, q of the prediction model; Obtain historical time series data Yt-1, Yt-2, ..., Yt-p; Calculate the time series after difference; Estimate model parameters, including the constant term c, autoregressive parameters φ1, φ2, …, φp, and moving average parameters θ1, θ2, …, θq; Calculate white noise terms εt, εt-1, εt-2, …, εt-q; Substitute the above parameters into the equation of the prediction model to calculate the predicted value Yt at time t; The equation of the prediction model is: Yt=c+φ1Yt-1+φ2Yt-2+...+φpYt-p+εt+θ1εt-1+θ2εt-2+...+θqεt-q, where Yt is the predicted value at time t, c is a constant term, φp and θq are autoregressive and moving average parameters, and εt is white noise.

5. The remote monitoring method according to claim 1, characterized in that: The monitoring method further comprises the following steps: The comprehensive state index CSI of multiple monitored objects is analyzed through the analysis model to obtain the state distribution of the monitored objects; Based on the status distribution, potential systemic problems are identified.

6. The remote monitoring method according to claim 5, characterized in that: The analyzing of the comprehensive status index CSI of the plurality of monitoring objects by using the analysis model comprises the following steps: Initialize K cluster centers μj; For each data point xi, calculate its distance to the center of each cluster and assign it to the cluster with the closest distance; Recalculate the center μj of each cluster; Calculate the sum of the squared distances of all data points to the center of their cluster; Repeat the above steps until the cluster division does not change or the maximum number of iterations is reached; Output the final clustering result and objective function value J; The functional representation of the analytical model is: J=∑∑||xi-μj|| 2 , where xi is the i-th data point and μj is the center of the j-th cluster.

7. The remote monitoring method according to claim 1, characterized in that: Determining whether the monitored object is in an abnormal state comprises the following steps: Establish a digital twin model of the monitored object; Inputting the real-time data into the digital twin model to simulate the operating status of the monitored object; Compare the simulation results with the actual operating status; Use machine learning algorithms to train anomaly detection models based on historical data; The anomaly detection model is used to assist in determining whether the monitored object is in an abnormal state.

8. The remote monitoring method according to claim 7, characterized in that: The method of using a machine learning algorithm to train an anomaly detection model based on historical data includes the following steps: Obtain the parameters of the trained anomaly detection model, including Lagrange multiplier αi, category label yi, kernel function K(xi,x) and bias term b; For the input sample x, calculate the kernel function value K(xi,x) between it and all support vectors xi; Multiply the kernel function value with the corresponding αi and yi, and sum all the results; Add the bias term b to the sum result; Take the symbol function of the final result to get the classification result; The decision function of the machine learning algorithm is expressed as: f(x)=sign(∑αi*yi*K(xi,x)+b).

9. A remote monitoring system, characterized in that: The monitoring system comprises: Data acquisition module, used to obtain multi-dimensional real-time data and historical data of the monitored object; A state calculation module, used to calculate the comprehensive state index CSI of the monitored object according to the real-time data and historical data; An abnormality judgment module, used to judge whether the monitored object is in an abnormal state based on the comprehensive state index CSI; An alarm generation module is used to generate multi-level alarm information when it is determined that the monitored object is in an abnormal state; A communication module, used for sending the multi-level alarm information to a remote terminal device and receiving a processing instruction sent by the remote terminal device; An execution module, used to execute corresponding automatic adjustment or manual intervention operations according to the processing instructions; A prediction module is used to perform time series analysis on the comprehensive status index CSI to predict the status trend of the monitored object in the future; The early warning module is used to generate early warning information when it is predicted that the monitored object may be abnormal in the future. An analysis module, used to perform cluster analysis on the comprehensive state indexes CSI of multiple monitored objects to obtain the state distribution of the monitored objects; and The machine learning module is used to train anomaly detection models based on historical data to assist in determining whether the monitored object is in an abnormal state.

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 remote monitoring method according to any one of claims 1 to 8 is implemented.

Images