An AI-based intelligent risk assessment system and method for scientific and technological equipment

Abstract

The present invention relates to the technical field of equipment risk identification, specifically to an AI-based intelligent risk assessment system and method for scientific and technological equipment. The system includes: a multidimensional data and model building module to collect multidimensional parameters of equipment operation, perform correlation analysis on the relationship between parameters, extract the characteristics of each parameter, and build an equipment operation status model. In the present invention, by normalizing the multidimensional parameters of the equipment and analyzing the correlation between parameters, and based on the entropy fluctuation of the parameter time series, it is possible to accurately identify abnormal points in the operation process, and by conducting an in-depth analysis of the trend and degree of the entropy mutation points, an efficient abnormality detection method is provided, so that potential risks can be exposed in advance. By analyzing the extreme value deviation during the operation of the equipment and predicting the future trend of the parameters, the critical state of the equipment can be judged in advance, providing forward-looking information for risk management and avoiding the operational risks caused by extreme deviations.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment risk identification, and in particular to an AI-based intelligent risk assessment system and method for scientific and technological equipment. Background Art

[0002] The field of equipment risk identification technology integrates a variety of technologies, including but not limited to artificial intelligence, machine learning, big data analysis, and Internet of Things technology for risk assessment of the entire life cycle of equipment. It supports continuous monitoring of equipment performance to identify risk factors before risks occur, thereby providing decision support. It is widely used in manufacturing, energy, transportation and other industries to dynamically evaluate the health status of equipment to ensure the safety and reliability of operations.

[0003] Among them, the intelligent risk assessment system for scientific and technological equipment uses artificial intelligence and machine learning technologies to intelligently analyze and evaluate the operating status of scientific and technological equipment. The data collected by the equipment is processed and analyzed through algorithms to identify potential risks and failure points, including the construction of assessment models, so that it can accurately analyze equipment failures and performance degradation. In addition, the system reduces human intervention by automating and optimizing risk management processes.

[0004] Existing technologies for analyzing equipment status rely more on static or single-dimensional parameter evaluation, which makes it difficult to fully capture the potential complex relationships between parameters and identify subtle changes in equipment operation. At the same time, it also ignores the complexity of parameter evolution over time. When capturing equipment anomalies, some early fault signals may be missed, resulting in the failure to provide timely warnings of equipment hazards. In addition, when equipment parameters are close to the limit state, existing technologies are difficult to accurately predict future risks, limiting the preventive maintenance capabilities during equipment operation. In terms of fault propagation path analysis, existing technologies are usually difficult to deeply explore the dependencies between components, resulting in difficulty in taking effective measures in a timely manner after the fault spreads, increasing equipment maintenance costs and downtime risks. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an AI-based intelligent risk assessment system and method for scientific and technological equipment.

[0006] In order to achieve the above purpose, the present invention adopts the following technical solution: an AI-based scientific and technological equipment risk intelligent assessment system includes:

[0007] The multi-dimensional data and model building module collects multi-dimensional parameters of equipment operation, performs correlation analysis on the relationship between parameters, extracts the characteristics of each parameter, and builds a model of equipment operation status;

[0008] The entropy value monitoring and abnormality identification module analyzes the entropy value fluctuation of the parameter in the current time period based on the equipment operation status model, and generates the abnormal status characteristic information of the equipment by determining the change trend and mutation degree of the abnormal point;

[0009] The extreme value and risk assessment module refers to the abnormal state characteristic information of the equipment, analyzes the current maximum deviation state and fluctuation of the equipment parameters, predicts the future change trend of the parameters under various conditions, determines the risk of extreme deviation to the operation of the equipment, and generates dynamic prediction and analysis results of equipment deviation;

[0010] The fault propagation and risk analysis module is based on the dynamic prediction analysis results of the equipment offset and combines the dependencies between equipment components to evaluate the propagation path of the fault between components, extract the correlation information between key components and other components, calculate the possibility of fault propagation, and generate an equipment risk propagation path diagram.

[0011] As a further solution of the present invention, the step of acquiring the equipment operation status model is specifically as follows:

[0012] Based on the collection of multi-dimensional parameters of equipment operation, the time series data of the parameters are processed, abnormal data are removed and interpolated, and the data is normalized to generate a time series data set;

[0013] Based on the time series data set, the parameter values ​​at time points are compared using the formula:

[0014]

[0015] parameter and parameters The linear correlation coefficient between , and obtain the analysis results of the correlation of parameters in equipment operation;

[0016] in, It's time point The next parameter values, It's time point The next parameter values, It is The average value of the parameters, It is The average value of the parameters, is the total number of time series data, is the index of a time point;

[0017] According to the analysis results of the parameter correlation during the operation of the equipment, statistical features and frequency domain features are extracted, and the feature data is reduced in dimension. The reduced data is input into the machine learning algorithm for model training and cross-validation to establish an equipment operation status model.

[0018] As a further solution of the present invention, the step of obtaining the entropy value fluctuation of the analysis parameter in the current time period is specifically as follows:

[0019] Based on the parameter information of the equipment operation status model, the formula is used:

[0020]

[0021] Calculation parameters The entropy value , generate the entropy value analysis results of the device;

[0022] in, Indicates the time series The probability of a state, Indicates The probability of a state The logarithm of Representation parameters The state index in the time series, Represents the total number of states in the time series;

[0023] According to the entropy analysis result of the device, by comparing with the historical baseline entropy value of the device, the entropy fluctuation of the current time period is analyzed, fluctuations higher than the baseline value are identified, potential device anomalies are marked, and entropy fluctuation analysis results are generated.

[0024] As a further solution of the present invention, the acquisition step of determining the change trend and mutation degree of the abnormal point is specifically as follows:

[0025] According to the entropy fluctuation analysis results, the time series is segmented, the entropy value changes in each time period are compared, the abnormal points are screened, and an abnormal point information set is generated;

[0026] According to the abnormal point information set, the formula is adopted:

[0027]

[0028] Calculate the change in entropy , combined with the amplitude and frequency of entropy value mutation, the current random fluctuations are eliminated, abnormal points are monitored, and characteristic information of abnormal status of equipment is generated;

[0029] in, Indicates time point The entropy value of Indicates time point The entropy value of .

[0030] As a further solution of the present invention, the step of obtaining the risk of extreme deviation to the operation of the equipment is specifically as follows:

[0031] According to the abnormal state characteristic information of the equipment, the fluctuation amplitude and deviation value of the parameter are compared, and smoothing is performed to generate the maximum deviation data of the current parameter;

[0032] According to the maximum deviation data of the current parameters, the formula is adopted:

[0033]

[0034] Calculation time Time equipment parameters Status , and obtain the dynamic prediction and analysis results of equipment offset;

[0035] in, Indicates the initial time The following device parameters, Indicates device parameters The speed of change over time, Indicates device parameters The acceleration over time, is the adjustment factor, and It indicates the start time and the future prediction time. represents the time increment, Indicates device parameters The amount of change.

[0036] As a further solution of the present invention, the step of obtaining the propagation path of the evaluation fault between components is specifically:

[0037] According to the dynamic prediction and analysis results of the device offset, the operating characteristics of the components are analyzed, the reaction behaviors of the components under the offset state are tracked, and correlation analysis is performed to generate dependency information between the device components;

[0038] According to the dependency information between the equipment components, the fault propagation path between the key components and other components is analyzed, and the risk propagation nodes are extracted by combining the historical fault data and the current status of the equipment to generate a fault propagation path diagram for the equipment.

[0039] As a further solution of the present invention, the steps of obtaining the equipment risk propagation path map are specifically as follows:

[0040] According to the information in the fault propagation path diagram of the device, historical interaction data of the key component and other related components are extracted, the device operation status record is called, and the analysis and classification are performed to generate the correlation information between the key component and other components;

[0041] According to the correlation information between the key component and other components, the formula is adopted:

[0042]

[0043] Calculate faults from components Propagate to components Probability , obtain the fault propagation evaluation results of key components;

[0044] in, Representation Components Fault pair component The direct impact Representation Components The weight, weight Indicates the corresponding component The importance of Represents the index of all components that may be affected by fault propagation. Indicates the total number of components in the device;

[0045] According to the fault propagation assessment results of the key components, the risk propagation path is marked and an equipment risk propagation path diagram is generated.

[0046] An AI-based intelligent risk assessment method for scientific and technological equipment, the AI-based intelligent risk assessment method for scientific and technological equipment is executed based on the above-mentioned AI-based intelligent risk assessment system for scientific and technological equipment, and comprises the following steps:

[0047] S1: Based on the multi-dimensional parameter data of equipment operation, normalization processing and correlation analysis are performed, and the equipment operation status model is constructed using machine learning algorithms;

[0048] S2: Based on the equipment operation status model, entropy values ​​are calculated for the parameter time series, abnormal points with sudden entropy changes are identified, and the changing trends of the abnormal points are analyzed to generate characteristic information of abnormal status of the equipment;

[0049] S3: Based on the abnormal state characteristic information of the equipment, analyze the deviation state of the multi-dimensional parameters of the equipment, predict the future critical state of the parameters, evaluate the risk of extreme deviation to the operation of the equipment, and generate dynamic prediction analysis results of equipment deviation;

[0050] S4: Based on the dynamic prediction analysis result of the device offset, analyze the dependencies between device components, calculate the fault propagation path and its possibility, and generate component fault propagation path information;

[0051] S5: Based on the component fault propagation path information, visualize the fault propagation chain, mark the risk path, and generate a device risk propagation path diagram.

[0052] Compared with the prior art, the advantages and positive effects of the present invention are:

[0053] In the present invention, by normalizing the multi-dimensional parameters of the equipment and analyzing the correlation between the parameters, and according to the entropy fluctuation of the parameter time series, it is possible to accurately identify abnormal points in the operation process, and by conducting in-depth analysis of the trend and degree of the entropy mutation points, an efficient abnormality detection method is provided, so that potential risks can be exposed in advance. By analyzing the extreme value deviations during the operation of the equipment and predicting the future change trends of the parameters, the critical state of the equipment can be judged in advance, providing forward-looking information for risk management and avoiding operational risks caused by extreme deviations. The analysis of the fault propagation path combined with the dependencies between the equipment components deeply reveals the propagation path of the fault from the key components to other components, which helps to identify key fault nodes in advance, and reduces the impact of equipment failures by optimizing decisions, thereby improving the system's global risk management capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a system flow chart of the present invention;

[0055] Figure 2 A flowchart of the steps for obtaining the equipment operation status model of the present invention;

[0056] Figure 3 A flow chart of the steps for obtaining the entropy value fluctuation of the analysis parameter in the current time period of the present invention;

[0057] Figure 4 A flowchart of the steps for determining the change trend and mutation degree of abnormal points in the present invention;

[0058] Figure 5 A flow chart of the steps for determining the risk of extreme deviation to the operation of the equipment according to the present invention;

[0059] Figure 6 A flow chart of the acquisition steps for evaluating the propagation path of a fault between components of the present invention;

[0060] Figure 7 This is a flow chart of the steps for obtaining the equipment risk propagation path map of the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. 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.

[0062] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0063] See also Figure 1 The present invention provides a technical solution: an AI-based intelligent risk assessment system for scientific and technological equipment, comprising:

[0064] The multi-dimensional data and model building module collects multi-dimensional parameters of equipment operation, including temperature, pressure, speed, and vibration data, processes and normalizes the data, and performs correlation analysis on the relationship between the parameters, extracts the characteristics of each parameter, and builds a model of equipment operation status through machine learning algorithms;

[0065] The entropy monitoring and anomaly identification module calculates the entropy of the time series of parameters during the operation of the equipment based on the equipment operation status model, analyzes the entropy fluctuation of the parameters in the current time period, extracts the abnormal points with sudden changes in entropy values ​​in the specified time period, and generates characteristic information of the abnormal state of the equipment by determining the change trend and degree of mutation of the abnormal points;

[0066] The extreme value and risk assessment module refers to the abnormal state feature information of the equipment, analyzes the current maximum deviation state and fluctuation of the equipment parameters for the multi-dimensional parameter data during the equipment operation process, and predicts the future change trend of the parameters under various conditions by evaluating the possibility of extreme deviation of the equipment parameters, that is, the critical state that will be reached within a time period, determines the risk of extreme deviation to the equipment operation, and generates dynamic prediction and analysis results of equipment deviation;

[0067] The fault propagation and risk analysis module is based on the dynamic prediction analysis results of equipment deviation, combined with the dependencies between equipment components. Equipment components are mechanical or electronic components that independently perform specific functions in equipment. It evaluates the propagation path of faults between components, extracts the correlation information between key components and other components. Key components are components that play a key role in the normal operation of equipment, calculates the possibility of fault propagation, and generates an equipment risk propagation path diagram;

[0068] The equipment operation status model includes the normalized values ​​of multi-dimensional parameters, the correlation matrix between parameters and the characteristic weight of each parameter. The characteristic information of the equipment abnormal status includes the timestamp of the abnormal point, the entropy value change amplitude of the abnormal parameter and the equipment status when the abnormality occurs. The dynamic prediction analysis results of equipment deviation include the parameter deviation range, potential extreme value locations and the duration of parameter deviation from normal values ​​within the future prediction time. The equipment risk propagation path diagram includes the failure impact range of key components, the dependency structure between components and the potential path of fault propagation.

[0069] See also Figure 2 , the specific steps for obtaining the equipment operation status model are:

[0070] Based on the collection of multi-dimensional parameters of equipment operation, the time series data of the parameters are processed, abnormal data are removed and interpolated, and the data is normalized to generate a time series data set;

[0071] Collect multi-dimensional parameters of equipment operation, including time series data of temperature, pressure, speed, and vibration. First, collect various types of data. Temperature parameters are collected through the temperature sensor module, and the collection interval is set to once per second. Pressure parameters are continuously collected through the pressure monitoring module, and the sampling frequency is three times per second. Speed ​​data is obtained through the encoder module, and vibration data is sampled at a high frequency through the vibration sensor module. After the collection of various parameters is completed, the data will be centrally stored in a unified data storage unit, and then the data will be processed. First, the abnormal points in the data are detected and eliminated. The moving average method is used to eliminate the abnormal points in the data that do not conform to the physical characteristics. Then the data is interpolated and supplemented to ensure that the values ​​lost due to interference during the collection process can be reasonably supplemented to form a complete time series. After that, the data is normalized. By calculating the minimum and maximum values ​​of each parameter, the minimum-maximum normalization method is used to map the data to the range of 0 to 1 to ensure that the data of each physical parameter are compared and analyzed at the same scale to facilitate subsequent correlation analysis and modeling.

[0072] Based on the time series data set, the parameter values ​​at each time point are compared using the formula:

[0073]

[0074] parameter and parameters The linear correlation coefficient between , and obtain the analysis results of the correlation of parameters in equipment operation;

[0075] in, It is used to quantify the degree of linear dependence between two parameters. The value range is -1 to 1. A value close to 1 indicates a strong positive correlation, a value close to -1 indicates a strong negative correlation, and a value close to 0 indicates no correlation. It's time point The next The parameter values ​​are collected in real time by the device's sensors. For example, the temperature parameter can be collected by the temperature sensor at different time points to obtain a set of data. It's time point The next parameter values, and Similarly, it is also obtained through sensors. It is The average value of the parameters is obtained by Sum the values ​​collected at all time points and divide by the number of time points Calculated, the calculation method is , which is used to standardize the parameter values ​​to make their mean zero in order to better analyze the fluctuation part, It is The average value of the parameters is calculated using the same method as same, It is the total number of time series data, indicating the number of different time points. In this example, the specific value can be determined by specifying the sampling time interval and sampling duration. It is the index of a time point, indicating the collection point of the parameter value at a certain moment. The value is determined by the timing recording module in the acquisition system.

[0076] If The first parameter is temperature, The parameter is pressure, and data is collected at three time points: temperature ,

[0077] pressure .

[0078] Calculate the average values ​​of temperature and pressure:

[0079]

[0080]

[0081] Enter the formula to calculate:

[0082]

[0083] Compute the difference term by term:

[0084] : ;

[0085] : ;

[0086] : ;

[0087] Calculate the numerator part:

[0088]

[0089] Calculate the denominator:

[0090]

[0091]

[0092] Calculate the correlation coefficient:

[0093]

[0094] The results show that temperature and pressure show a very strong positive correlation in this data set, close to 1. By comparing with historical benchmark data, if the correlation coefficient in the historical data is about 0.8, it means that the current coupling relationship between temperature and pressure is closer than before, which may indicate that the equipment is currently under a higher workload or there are some changes in operating conditions. Therefore, it is necessary to pay further attention to the changes in these two parameters to determine whether equipment maintenance or adjustment of operating conditions is required.

[0095] According to the analysis results of the correlation of parameters during equipment operation, statistical features and frequency domain features are extracted, and the feature data is reduced in dimension. The reduced-dimensional data is input into the machine learning algorithm for model training and cross-validation to establish the equipment operation status model.

[0096] Extract the characteristics of each parameter. Perform feature analysis on the time series data. First, perform time domain analysis on the data to extract the statistical characteristics of each parameter, including mean, variance, skewness and kurtosis, etc. The mean is used to represent the overall level of the data, the variance is used to measure the fluctuation amplitude of the data, the skewness is used to judge the distribution symmetry of the data, and the kurtosis is used to reflect the sharpness of the data. Next, perform frequency domain analysis to convert the time series data into frequency domain features through frequency domain transformation, extract the main frequency components and their corresponding amplitude information, and then perform dimensionality reduction on the extracted time domain and frequency domain features. Use feature selection methods to retain features that are highly related to the equipment status and reduce irrelevant or redundant features to reduce computational complexity. Next, input the reduced dimensionality features into the machine learning algorithm for training. Use supervised learning algorithms such as support vector machines or random forests to learn historical equipment data. Establish a mapping relationship based on features and corresponding equipment status labels. Use cross-validation methods to adjust hyperparameters during model training to optimize model performance. Finally, the training model obtained is used to judge the equipment operation status in real time.

[0097] See also Figure 3 , the specific steps for obtaining the entropy fluctuation of the analysis parameters in the current time period are:

[0098] Based on the parameter information of the equipment operation status model, the formula is used:

[0099]

[0100] Calculation parameters The entropy value , generate the entropy value analysis results of the device;

[0101] in, It is used to measure the amount of information or uncertainty in a parameter time series. A high entropy value means that the parameter changes more randomly, and a low entropy value means that the parameter is more stable. This value is used to evaluate the complexity of different states during the operation of the equipment. Indicates the time series The probability of a state, indicating the parameter Take value in time period The frequency of is obtained by counting the frequency of occurrence of each state in the time series. Indicates The probability of a state The logarithm of , usually with base 2, is used to calculate the actual amount of information (i.e., number of bits) in the state. Representation parameters The state index in the time series obtains different states through data collection at different time points. It represents the total number of different states in the time series, which is obtained by statistically analyzing the actual data points of the parameter through the data acquisition system.

[0102] If the device collects a time series of a parameter (such as temperature), the sampling frequency is once per second, and a total of 5 seconds of data is collected. During this time period, the temperature has three different states, namely , and , and their frequencies are , , , obtained by the ratio of the number of occurrences of each state to the total number of occurrences in the statistical data.

[0103] Calculate the amount of information in each state:

[0104]

[0105]

[0106]

[0107] Calculate the entropy value during this period:

[0108]

[0109] The results show that entropy It means that the temperature change of the equipment during this time period has a certain uncertainty.

[0110] According to the entropy analysis results of the equipment, by comparing with the historical baseline entropy value of the equipment, analyze the entropy fluctuation of the current time period, identify fluctuations higher than the baseline value, mark potential equipment anomalies, and generate entropy fluctuation analysis results;

[0111] According to the entropy value , analyze the entropy fluctuations in the current time period. First, extract the baseline entropy value of the parameter (such as temperature) from the historical data of the device. For example, the entropy value of the device during normal operation is 1.2. The baseline value is obtained through long-term historical data analysis, indicating that the parameter fluctuation is small and stable during normal operation of the device. Then, calculate the entropy value of the data collected in the current time period. The results show that the entropy value of the current time period is It is significantly higher than the baseline value, indicating that the operating status of the equipment has large fluctuations in the current time period and the temperature parameters show higher uncertainty.

[0112] See also Figure 4 , the specific steps for determining the change trend and mutation degree of the abnormal points are:

[0113] According to the results of entropy fluctuation analysis, the time series is segmented, the entropy changes in each time period are compared, anomalies are screened, and anomaly information sets are generated;

[0114] Extract the abnormal points of entropy value mutation within the specified time period. Based on the identified abnormal points of entropy value fluctuation, first further analyze the abnormal entropy values ​​of each time window. Use the sliding window technology to divide the time series into multiple continuous time periods according to the set window size. By comparing the entropy value fluctuation amplitude in each time period, focus on screening out those points that are significantly different from the conventional entropy value. Then use the predefined entropy value mutation threshold to filter these abnormal points to ensure that only points with significant mutation amplitudes are retained. Then mark these mutation points as abnormal points that need to be paid attention to during the operation of the equipment. In order to ensure the accuracy of these abnormal points, further compare the changing trends of other related parameters (such as pressure, speed, etc.) to exclude possible short-term random fluctuations. Finally, integrate the filtered entropy value mutation abnormal points into a final abnormal point set to provide input data for the next step of state judgment and analysis.

[0115] According to the abnormal point information set, the formula is used:

[0116]

[0117] Calculate the change in entropy , combined with the amplitude and frequency of entropy value mutation, the current random fluctuations are eliminated, abnormal points are monitored, and characteristic information of abnormal status of equipment is generated;

[0118] in, Represents the mutation of a parameter in a time series between consecutive time points. Indicates time point The entropy value of the equipment parameters obtained through the previous analysis is used to quantify the randomness of the equipment operating state at that moment. Indicates time point The entropy value represents the uncertainty of the device's state at the next time point, which is usually calculated based on the device's operating data. It indicates the absolute difference in entropy between two time points. It is used to quantify the entropy mutation of the device in this short period of time. The larger the mutation, the more obvious the change in the device's operating status.

[0119] If the device is at time The entropy value , time point The entropy value , obtained through equipment operation data analysis, substituted into the formula:

[0120]

[0121] Compared with the historical entropy mutation threshold (for example, 0.1), it can be determined that this mutation value is far beyond the normal range, indicating that the device has a significant state change during this period. Result analysis: This result shows that the entropy value of the device is at the time point and A significant mutation occurred between 2009 and 2011, with a mutation amplitude of 0.215, which was consistent with the historical benchmark mutation threshold. By comparison, it is found that the current state fluctuation of the equipment exceeds the normal range. Combined with the change trend and mutation degree of the abnormal point, the operation state of the equipment has shown an unstable trend, and the characteristic information of the abnormal state of the equipment is generated to provide a basis for further maintenance decisions.

[0122] See also Figure 5 ,The specific steps for determining the risk of extreme deviation to equipment operation are:

[0123] According to the characteristic information of the abnormal state of the equipment, the fluctuation amplitude and deviation value of the parameters are compared and smoothed to generate the maximum deviation data of the current parameters;

[0124] Referring to the characteristic information of the abnormal status of the equipment, the historical benchmark data of each parameter is obtained for the multi-dimensional parameter data during the operation of the equipment. The standard deviation range of the normal operation of the equipment is defined by the benchmark data. Next, the fluctuation amplitude and maximum deviation value of each parameter under the current operating state of the equipment are extracted based on the statistical analysis method. These data are compared with the historical benchmark data item by item, and the moving average method is used to smooth the various parameters of the equipment to eliminate the errors caused by short-term fluctuations. By comparing the standard deviation with the threshold, the part of the current equipment operating parameters that deviates from the benchmark range is identified, with special attention paid to those parameters with large deviations and long durations. These parameters with large deviations are then processed separately, and the sliding window technology is used to divide the time period, and the deviation data of these parameters in different time periods are extracted. Finally, a report on the current maximum deviation status of the equipment is generated, with emphasis on marking the parameter deviations that may cause equipment risks.

[0125] According to the maximum deviation data of the current parameters, the formula is used:

[0126]

[0127] Calculation time Time equipment parameters Status , and obtain the dynamic prediction and analysis results of equipment offset;

[0128] in, is the predicted offset value of the future device parameters at this moment, Indicates the initial time The device parameters under, that is, the initial state of the current device, are obtained through device monitoring. Indicates device parameters The speed of change over time is calculated by performing differential processing on the current time series data. Indicates device parameters The acceleration that changes with time reflects the second-order derivative of parameter changes. It is obtained by changing the rate of change of equipment parameters and can better reflect the drastic changes of parameters under sudden conditions. It is an adjustment coefficient that controls the impact of the second-order derivative on the result. It is determined by analyzing the equipment status in historical data. The value is set according to the fluctuation of equipment parameters and the expected risk level, which can improve the prediction accuracy and flexibility. and It represents the starting time and the future prediction time. The future offset of the equipment parameters is calculated by integrating the selected time period. Represents the time increment, which is a small time interval in the integration process. and The time interval is divided and accumulated to calculate the future changes of the parameters. Indicates device parameters A small increment or change in the value of a

[0129] If the device is initially Time Parameters (such as temperature), the rate of temperature change during equipment operation is degrees / second, the temperature change acceleration is , adjustment coefficient , target prediction time Seconds, substitute into the formula:

[0130]

[0131] The calculation results show that the device temperature will shift to 124 degrees in the next 10 seconds. By comparing with the safe temperature range of the device (such as not exceeding 110 degrees), it can be determined that the device is about to reach a critical state, indicating that the device is about to face the risk of extreme deviation. Combined with this prediction result, preventive measures can be taken in advance to avoid equipment failure or damage, and ultimately generate a dynamic prediction analysis report on equipment deviation to provide data support for equipment management.

[0132] See also Figure 6 ,The specific steps for obtaining the propagation path of the fault between components are:

[0133] According to the results of dynamic prediction and analysis of equipment offset, the operating characteristics of components are analyzed, the reaction behavior of components under offset state is tracked, and correlation analysis is performed to generate dependency information between equipment components;

[0134] First analyze the key equipment parameters displayed in the dynamic prediction results The resulting offset is determined by comparing the historical data of the operating status of the equipment components, determining the operating characteristics of each component under the offset state, using the dependency chart analysis method to extract the interaction between the mechanical and electronic components in the equipment, gradually analyzing the degree of dependency of each component, and evaluating the impact of each component on the overall operation of the equipment. Then, by monitoring the current operating data of the equipment, calling the historical correlation records between the equipment components, tracking the response of key components under the offset state, combining the multi-dimensional parameters in the dynamic prediction of the equipment, refining the action mode of the key components and their impact path on other components, clarifying how the dependency relationship between equipment components changes due to parameter offset, and finally generating detailed dependency information of the equipment components, identifying the key components and other affected components associated with them.

[0135] According to the dependency information between equipment components, analyze the fault propagation path between key components and other components, combine historical fault data and the current status of the equipment, extract risk propagation nodes, and generate a fault propagation path diagram for the equipment;

[0136] First, the relationship between key components and other dependent components is sorted out to identify potential fault propagation chains between components. By introducing historical fault data and combining it with the offset status of the current device, the fault paths that may propagate through multiple components are identified. Then, the fault propagation model is used to quantify the fault propagation characteristics of each component in the device, and the components that may become propagation nodes in each propagation path are extracted. The fault transmission path between components is represented by a directed graph. The role of each component in the path in the propagation process is then analyzed. Combined with the actual operation data of the equipment, the most likely high-risk propagation nodes are marked. Combined with the dependency relationship of each component and the probability of failure, the potential fault impact range of the high-risk nodes is expanded and predicted. Finally, a fault propagation path diagram of the device is generated, showing the fault propagation path between key components in the device and its corresponding risk area.

[0137] See also Figure 7 ,The specific steps for obtaining the equipment risk propagation path map are:

[0138] According to the information in the fault propagation path diagram of the equipment, the historical interaction data of the key components and other related components are extracted, the equipment operation status records are called, analyzed and classified, and the correlation information between the key components and other components is generated;

[0139] Through the equipment operation history data and current real-time monitoring data, the key components involved in the fault propagation path are analyzed, and the dependency data of these key components in normal operation and fault state are obtained. Next, the component dependency analysis method is used to refine the fault propagation chain between components, identify the impact of key components on other components in different states, and combine the hierarchical structure in the equipment to confirm the linkage effect of the operating status of each component on other components. Through data comparison, the key components with a high degree of impact on other components during the fault propagation process are recorded. Subsequently, by comparing the historical fault propagation data, the components most closely related to the key component failure are further extracted, and the detailed correlation information between the key components and other components is obtained, which provides a reference for the next step of equipment risk assessment and maintenance strategy.

[0140] According to the correlation information between key components and other components, the formula is used:

[0141]

[0142] Calculate faults from components Propagate to components Probability , obtain the fault propagation evaluation results of key components;

[0143] in, Used to quantify the probability of fault propagation between equipment components, Representation Components Fault pair component The direct impact caused by the fault can be quantified through historical data, physical connections between components, and changes in the operating status of components. It can be obtained by analyzing the historical fault propagation paths of each component in the equipment and actual operation monitoring. Representation Components The weight reflects the relative importance of the component in the system. It can be obtained through comprehensive evaluation of multiple indicators such as equipment function analysis, operating time, and failure cost. It is usually obtained through expert scoring, historical data regression analysis, or function-based automatic weight adjustment methods. Representation Components The weighted total impact on all other components, weight Indicates the corresponding component The importance of components is used to comprehensively consider the impact of failures and relative weights between components. The fault history data and fault propagation relationship with other components are combined with weight allocation to obtain: Represents the index of all components that may be affected by fault propagation, From 1 to Represents all possible components in a device. This range can be determined through the device's architecture diagram or dependency analysis. Represents the total number of components in a device, as determined by the device's configuration or system composition.

[0144] For example, a device contains components, respectively , and .

[0145] If there are components in the device , and , where components Fault pair The impact strength ,right The impact strength , components and The weights are , , then we can substitute into the formula:

[0146]

[0147] The calculation process is:

[0148]

[0149] Result presentation component Fault propagation to components The probability is about 81.4%. If historical data shows that the component Fault propagation to components The probability baseline value is 80%, and the current calculation result 81.4% indicates that the probability of propagation has increased slightly, which may reflect the increased dependency between components or the increase in the number of components. The fault response capability of the equipment is reduced, and further analysis can reveal the potential fault propagation path of the equipment and help optimize maintenance strategies and fault prevention mechanisms.

[0150] Based on the fault propagation assessment results of key components, the risk propagation path is marked and the equipment risk propagation path diagram is generated;

[0151] Based on the correlation data between the aforementioned key components and other components, the fault propagation path diagram generation method is used. First, the equipment structure model is called. The model is established by comprehensively analyzing the equipment hardware structure, functional modules, physical connections and logical dependencies. It usually comes from the CAD model, system architecture diagram or engineering blueprint in the equipment design stage, and is modified in combination with the topological structure of the actual operation of the equipment and the updated monitoring data. It can reflect the physical layout of each component of the equipment and its interdependence, and classify the fault propagation risk of each component in the equipment. Then, the interdependence between components is mapped, and the fault propagation sequence between components is analyzed. The key nodes in the propagation chain are further marked through historical data and real-time monitoring data. Subsequently, according to the operating status and fault history of the components, the high-risk paths in each fault propagation chain are marked, and the risk level of the propagation path is distinguished by color. The high-risk components and the length of the propagation chain are clearly displayed, and finally the generation of the equipment risk propagation path diagram is completed, showing the fault propagation path of each key component in the equipment, providing an intuitive reference for subsequent equipment fault management.

[0152] An AI-based intelligent risk assessment method for scientific and technological equipment, the AI-based intelligent risk assessment method for scientific and technological equipment is performed based on the above-mentioned AI-based intelligent risk assessment system for scientific and technological equipment, and includes the following steps:

[0153] S1: Based on the multi-dimensional parameter data of equipment operation, normalization processing and correlation analysis are performed, and the equipment operation status model is constructed using machine learning algorithms;

[0154] S2: Based on the equipment operation status model, entropy values ​​are calculated for the parameter time series, abnormal points with sudden entropy changes are identified, and the changing trends of abnormal points are analyzed to generate characteristic information of abnormal equipment status;

[0155] S3: Based on the abnormal status feature information of the equipment, analyze the deviation status of the multi-dimensional parameters of the equipment, predict the future critical state of the parameters, evaluate the risk of extreme deviation to the operation of the equipment, and generate dynamic prediction and analysis results of equipment deviation;

[0156] S4: Based on the dynamic prediction analysis results of equipment offset, analyze the dependencies between equipment components, calculate the fault propagation path and its possibility, and generate component fault propagation path information;

[0157] S5: Based on the component fault propagation path information, visualize the fault propagation chain, mark the risk path, and generate a device risk propagation path diagram.

[0158] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An AI-based intelligent risk assessment system for scientific and technological equipment, characterized in that: The system comprises: The multidimensional data and model building module collects multidimensional parameters of equipment operation, conducts correlation analysis on the relationship between parameters, extracts statistical features and frequency domain features of each parameter from the correlation analysis results, performs dimensionality reduction on the feature data, and inputs the reduced-dimensional data into the machine learning algorithm for model training and cross-validation to build an equipment operation status model; The entropy value monitoring and abnormality identification module analyzes the entropy value fluctuation of the parameter in the current time period based on the equipment operation status model, and generates the abnormal status characteristic information of the equipment by determining the change trend and mutation degree of the abnormal point; The steps for obtaining the entropy value fluctuation of the analysis parameter in the current time period are specifically as follows: Based on the parameter information of the equipment operation status model, the formula is used: Calculation parameters The entropy value , generate the entropy value analysis results of the device; in, Indicates the time series The probability of a state, Indicates The probability of a state The logarithm of Representation parameters The state index in the time series, Represents the total number of states in the time series; According to the entropy value analysis result of the device, by comparing with the historical baseline entropy value of the device, analyzing the entropy value fluctuation in the current time period, identifying fluctuations higher than the baseline value, marking potential device anomalies, and generating entropy value fluctuation analysis results; The steps for obtaining the change trend and mutation degree of the abnormal point are specifically as follows: According to the entropy fluctuation analysis results, the time series is segmented, the entropy value changes in each time period are compared, the abnormal points are screened, and an abnormal point information set is generated; According to the abnormal point information set, the formula is adopted: Calculate the change in entropy , combined with the amplitude and frequency of entropy value mutation, the current random fluctuations are eliminated, abnormal points are monitored, and characteristic information of abnormal status of equipment is generated; in, Indicates time point The entropy value of Indicates time point The entropy value of The extreme value and risk assessment module refers to the abnormal state characteristic information of the equipment, analyzes the current maximum deviation state and fluctuation of the equipment parameters, predicts the future change trend of the parameters under various conditions, determines the risk of extreme deviation to the operation of the equipment, and generates dynamic prediction and analysis results of equipment deviation; The steps for determining the risk of extreme deviation to the operation of the equipment are specifically as follows: According to the abnormal state characteristic information of the equipment, the fluctuation amplitude and deviation value of the parameter are compared, and smoothing is performed to generate the maximum deviation data of the current parameter; According to the maximum deviation data of the current parameters, the formula is adopted: Calculation time Time equipment parameters Status , and obtain the dynamic prediction and analysis results of equipment offset; in, Indicates the initial time The following device parameters, Indicates device parameters The speed of change over time, Indicates device parameters The acceleration over time, is the adjustment factor, and It indicates the start time and the future prediction time. represents the time increment, Indicates device parameters The amount of change; The fault propagation and risk analysis module evaluates the fault propagation path between components based on the dynamic prediction analysis results of the equipment deviation and the dependencies between equipment components, extracts the correlation information between key components and other components, calculates the possibility of fault propagation, and generates an equipment risk propagation path diagram; The steps of obtaining the propagation path of the evaluation fault between components are specifically as follows: According to the dynamic prediction and analysis results of the device offset, the operating characteristics of the components are analyzed, the reaction behaviors of the components under the offset state are tracked, and correlation analysis is performed to generate dependency information between the device components; According to the dependency information between the equipment components, the fault propagation path between the key components and other components is analyzed, and the risk propagation nodes are extracted by combining the historical fault data and the current status of the equipment to generate a fault propagation path diagram of the equipment; The steps for obtaining the equipment risk propagation path map are specifically as follows: According to the information in the fault propagation path diagram of the device, historical interaction data of the key component and other related components are extracted, the device operation status record is called, and the analysis and classification are performed to generate the correlation information between the key component and other components; According to the correlation information between the key component and other components, the formula is adopted: Calculate faults from components Propagate to components Probability , obtain the fault propagation evaluation results of key components; in, Representation Components Fault pair component The direct impact Representation Components The weight, weight Indicates the corresponding component The importance of Represents the index of all components that may be affected by fault propagation. Indicates the total number of components in the device; According to the fault propagation assessment results of the key components, the risk propagation path is marked and an equipment risk propagation path diagram is generated.

2. The AI-based intelligent risk assessment system for scientific and technological equipment according to claim 1 is characterized in that: The steps for obtaining the equipment operation status model are specifically as follows: Based on the collection of multi-dimensional parameters of equipment operation, the time series data of the parameters are processed, abnormal data are removed and interpolated, and the data is normalized to generate a time series data set; Based on the time series data set, the parameter values ​​at time points are compared using the formula: parameter and parameters The linear correlation coefficient between , and obtain the analysis results of the correlation of parameters in equipment operation; in, It's time point The next parameter values, It's time point The next parameter values, It is The average value of the parameters, It is The average value of the parameters, is the total number of time series data, is the index of a time point.

3. An AI-based intelligent risk assessment method for scientific and technological equipment, characterized in that: The AI-based scientific and technological equipment risk intelligent assessment system according to any one of claims 1-2 comprises the following steps: Based on the multi-dimensional parameter data of equipment operation, normalization processing and correlation analysis are performed, and the equipment operation status model is constructed using machine learning algorithms; Based on the equipment operation status model, entropy values ​​are calculated for the parameter time series, abnormal points with sudden entropy changes are identified, and the changing trends of the abnormal points are analyzed to generate characteristic information of abnormal status of the equipment; Based on the abnormal state characteristic information of the equipment, the deviation state of the multi-dimensional parameters of the equipment is analyzed, the critical state of the parameters in the future is predicted, the risk of extreme deviation to the operation of the equipment is evaluated, and the dynamic prediction analysis result of the equipment deviation is generated; Based on the dynamic prediction and analysis results of the equipment offset, the dependencies between equipment components are analyzed, the fault propagation paths and their possibilities are calculated, and component fault propagation path information is generated; Based on the component fault propagation path information, the fault propagation chain is visualized, and the risk path is marked to generate a device risk propagation path diagram.

Images