A log file processing method and system for obstetrics and gynecology medical equipment data

Abstract

The present invention relates to the field of data processing technology, and in particular to a log file processing method and system for obstetrics and gynecology medical equipment data. The method comprises the following steps: S1: collecting and preprocessing log files of obstetrics and gynecology medical equipment, extracting warning information data and error alarm data from the preprocessed log files; S2: initializing a balance stack based on the warning information data and the error alarm data, and pressing the warning information data and the error alarm data into the balance stack; determining the conversion point between online log files and offline log files. The present invention preprocesses the log files, predicts low-frequency events using an LSTM model, introduces balance stack management and dynamically adjusts capacity, and combines fuzzy reasoning to deal with unpredicted low-frequency events. The present invention improves the automation and reliability of log file processing, enhances the predictability and responsiveness of the system, and provides a strong guarantee for the safe operation of obstetrics and gynecology medical equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a log file processing method and system for obstetrics and gynecology medical equipment data. Background Art

[0002] Log file processing methods provide real-time monitoring and fault diagnosis by automatically extracting and analyzing large amounts of data, optimizing system performance and security. However, they also face privacy and security risks, challenges in efficient processing of large amounts of data, and difficulty in parsing due to complex log formats. These problems are particularly prominent in the data processing of obstetric and gynecological medical equipment: complex log formats increase the difficulty of parsing, highly sensitive data require strict security measures, and existing methods are prone to performance bottlenecks and security risks when processing large amounts of data. Therefore, there is an urgent need for a smarter and more efficient log file processing solution that can accurately predict and quickly respond to low-frequency events while ensuring data security and privacy, thereby improving the reliability and intelligence of the system. Summary of the invention

[0003] In order to overcome the shortcoming of low efficiency in log file processing, the present invention provides a log file processing method and system for obstetrics and gynecology medical equipment data.

[0004] The technical solution of the present invention is: a log file processing method for obstetrics and gynecology medical equipment data, comprising the following steps:

[0005] S1: Collect and preprocess the log files of obstetrics and gynecology medical equipment, and extract warning information data and error alarm data from the preprocessed log files;

[0006] S2: Initialize the balance stack based on the warning information data and the error alarm data, and push the warning information data and the error alarm data into the balance stack; determine the switching point between the online log file and the offline log file; the balance stack is composed of two sub-stacks with the same capacity, the left stack: defined as a sub-stack that supports data insertion from the left end, all new elements are pushed in from the left end of the stack, following the LIFO principle; the right stack: defined as a sub-stack that supports data insertion from the right end, all new elements are pushed in from the right end of the stack, also following the LIFO principle; the switching point refers to the timing of switching between the online log file and the offline log file determined according to the difference in the upper limit of the capacity of the left stack and the right stack;

[0007] S3: Based on the conversion point between the online log file and the offline log file, the total capacity of the balancing stack is adjusted; based on the capacity change rate of the balancing stack, the impact of low-frequency events on the log file is obtained using warning information data and error alarm data; the low-frequency events are rare or uncommon events.

[0008] Preferably, the collecting and preprocessing of log files of obstetrics and gynecology medical equipment, and extracting warning information data and error alarm data from the preprocessed log files, include:

[0009] Collect raw log files from obstetrics and gynecology medical equipment and perform preprocessing, including data cleaning, missing value supplementation and standardization, to generate a preprocessed log file dataset;

[0010] Use the preprocessed log file dataset to build an LSTM model, and input the log file dataset into the LSTM model for training to predict low-frequency events in the log file;

[0011] According to the prediction results of the LSTM model, warning information data and error alarm data are extracted from the log files of low-frequency events.

[0012] Preferably, the extracting warning information data and error alarm data from the log file of the low-frequency event according to the prediction result of the LSTM model includes:

[0013] Collect the prediction results of the LSTM model for low-frequency events in log files;

[0014] Identify low-frequency events that the LSTM model fails to predict, and extract warning information data and error alarm data from the log files corresponding to the low-frequency events;

[0015] Construct a fuzzy reasoning model, and use the fuzzy reasoning model to prioritize the low-frequency events predicted and not predicted by the LSTM model to obtain the classification results;

[0016] Based on the results of priority division, the time dependency and quantity ratio of warning information data and false alarm data are analyzed when low-frequency events occur.

[0017] Preferably, the step of initializing the balancing stack based on the warning information data and the error alarm data, and pushing the warning information data and the error alarm data into the balancing stack; and determining the conversion point between the online log file and the offline log file comprises:

[0018] The specific method of building a balanced stack is as follows:

[0019] Determine a fixed total capacity, which is shared by two sub-stacks, and each sub-stack has a fixed upper capacity limit;

[0020] The left stack pushes elements from the left, and the stack top pointer moves to the right; the right stack pushes elements from the right, and the stack top pointer moves to the left;

[0021] When any substack reaches its upper capacity limit, it stops accepting new elements until an element is popped out; the two substacks always maintain the same capacity, and the same operations are performed on the two substacks while adjusting the capacity;

[0022] Push operation: elements are pushed into the left stack from the left, and elements are pushed into the right stack from the right; pop operation: remove the elements at the top of each stack.

[0023] Preferably, the initializing the balancing stack, pushing the warning information data and the error alarm data into the balancing stack; and determining the conversion point between the online log file and the offline log file include:

[0024] Push the warning information data into the left stack of the balance stack, and push the error alarm data into the right stack of the balance stack;

[0025] Determine the switching point between the online log file and the offline log file according to the difference between the upper capacity limits of the left stack and the right stack;

[0026] Calculate the first conversion frequency and the second conversion frequency: The time when the left stack reaches the upper limit is recorded as , refers to the time from the beginning of recording to the first time the stack on the left reaches the upper limit; the time when the stack on the right reaches the upper limit is recorded as , refers to the time from the beginning of recording to the first time the right stack reaches the upper limit; when the left stack capacity reaches the upper limit before the right stack, that is < , calculate the time difference from when the left stack is full to when the right stack is also full, and determine the first conversion frequency; when the capacity of the right stack reaches the upper limit before the left stack, that is, < , the same logic applies to calculating the second switching frequency; the calculation formulas for the first switching frequency and the second switching frequency are as follows, in, is the first switching frequency, is the second switching frequency, is the time when the left stack reaches its capacity limit, The time when the right stack reaches its upper limit; the priority of the warning information data stored in the left stack is greater than the priority of the error alarm data stored in the right stack, and the system design ensures > .

[0027] Preferably, adjusting the total capacity of the balancing stack based on the conversion point of the online log file and the offline log file includes:

[0028] Extract warning information data and error alarm data under low-frequency events from log files;

[0029] Analyze the data to determine the time dependency, i.e. the trend over time, and the quantity ratio, i.e. the proportion of the two in the total amount of data;

[0030] According to the conversion points of the online log file and the offline log file, combined with the above time dependency and quantity ratio relationship, the switching time between the online mode and the offline mode is determined;

[0031] Based on the determined transition point, the total capacity of the balancing stack is dynamically adjusted; the new total capacity of the balancing stack is calculated using a formula.

[0032] Preferably, the total capacity of the balancing stack is dynamically adjusted based on the determined conversion point; and a new total capacity of the balancing stack is calculated using a formula, including: the formula is calculated as follows, in, is the total capacity of the balanced stack after adjustment, To balance the total capacity of the stack, is the current capacity of the left stack, is the current capacity of the right stack, is the time dependency coefficient of the left stack, is the time dependence coefficient of the right stack, is the ratio of the number of stacks on the left, is the ratio of the number of stacks on the right, is the weight parameter, is the scaling factor.

[0033] Preferably, the capacity change rate of the balancing stack is based on the use of warning information data and error alarm data to obtain the impact of low-frequency events on the log file, including:

[0034] Record the change of each balance stack capacity and calculate the capacity change rate. The capacity change rate is defined as the ratio of two adjacent capacities, that is, ,in It is The capacity of times, It is times the capacity;

[0035] Calculate two differences: collect the difference between the warning information data in the LSTM model prediction state and the warning information data in the unpredicted state, and define it as the first difference ; Collect the difference between the error alarm data in the LSTM model prediction state and the error alarm data in the unpredicted state, and define it as the second difference ;

[0036] According to the capacity change rate of the balanced stack, the first difference and the second difference , evaluate the impact of low-frequency events on log files, and use the formula to calculate the impact .

[0037] Preferably, the capacity change rate of the balance stack, the first difference and the second difference , evaluate the impact of low-frequency events on log files, and use the formula to calculate the impact , including: The formula is as follows, in, is the impact of low-frequency events on log files, For the The capacity of the secondary balance stack, The difference between the warning information data in the LSTM model prediction state and the warning information data in the unpredicted state, that is, the first difference , The difference between the error alarm data in the LSTM model prediction state and the error alarm data in the unpredicted state, that is, the second difference , and is the weight parameter, is the weight parameter, The number of recorded capacity changes.

[0038] A log file processing system for obstetrics and gynecology medical equipment data, comprising:

[0039] The log file collection and preprocessing module collects log files from obstetrics and gynecology medical equipment and performs data cleaning, missing value supplementation and standardization processing;

[0040] LSTM model prediction module, builds an LSTM model, puts the preprocessed log file dataset into the LSTM model, and predicts low-frequency events in the log file;

[0041] Fuzzy reasoning module, builds a fuzzy reasoning model, prioritizes the low-frequency events predicted and not predicted by the LSTM model, and obtains the time dependency and quantity ratio relationship;

[0042] The balanced stack management module defines and manages a balanced stack consisting of two sub-stacks of the same volume, supporting data entry from the left and right sides respectively, and following the LIFO principle;

[0043] A conversion point determination module determines the conversion point between the online log file and the offline log file according to the difference between the upper capacity limits of the left stack and the right stack;

[0044] Dynamic adjustment module, dynamically adjusts the total capacity of the balancing stack based on the conversion points between online log files and offline log files;

[0045] The impact assessment module records the capacity change rate of each balancing stack, and calculates the impact of low-frequency events on log files by combining the difference between the warning information data predicted by the LSTM model and the warning information data not predicted, as well as the difference between the false alarm data.

[0046] Beneficial effects: First, by carrying out detailed data preprocessing of log files, including data cleaning, missing value supplementation and standardization, the quality and reliability of the data are ensured, laying a solid foundation for subsequent analysis. Secondly, the use of the LSTM model to predict low-frequency events not only improves the accuracy of the prediction, but also fills the gap in the existing log management system for insufficient processing of rare events, and enhances the predictability and responsiveness of the system. Then, the introduction of the balance stack management warning information data and error alarm data realizes the intelligent conversion of online and offline log files, effectively avoids data loss, and further optimizes resource utilization by dynamically adjusting the balance stack capacity. In addition, the application of the fuzzy reasoning module enables the system to flexibly respond to unpredicted low-frequency events, improving the adaptability and robustness of the system. Finally, by formulating the evaluation of the impact of low-frequency events on log files, a scientific and quantitative evaluation method is provided to help the medical team discover and deal with potential problems in a timely manner. In summary, the present invention not only improves the automation level of log file processing, but also significantly enhances the reliability and intelligence of the system, providing a strong guarantee for the safe operation of obstetrics and gynecology medical equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A flow chart of a log file processing method for obstetrics and gynecology medical equipment data according to the present invention;

[0048] Figure 2 The diagram is a structural diagram of a log file processing system for obstetrics and gynecology medical equipment data according to the present invention. DETAILED DESCRIPTION

[0049] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Embodiment 1: A log file processing method for obstetrics and gynecology medical equipment data, such as Figure 1 As shown, the following steps are included:

[0051] S1: Collect and preprocess the log files of obstetrics and gynecology medical equipment, and extract warning information data and error alarm data from the preprocessed log files;

[0052] S2: Initialize the balance stack based on the warning information data and the error alarm data, and push the warning information data and the error alarm data into the balance stack; determine the switching point between the online log file and the offline log file; the balance stack is composed of two sub-stacks with the same capacity, the left stack: defined as a sub-stack that supports data insertion from the left end, all new elements are pushed in from the left end of the stack, following the LIFO principle; the right stack: defined as a sub-stack that supports data insertion from the right end, all new elements are pushed in from the right end of the stack, also following the LIFO principle; the switching point refers to the timing of switching between the online log file and the offline log file determined according to the difference in the upper limit of the capacity of the left stack and the right stack;

[0053] S3: Based on the conversion point between the online log file and the offline log file, the total capacity of the balancing stack is adjusted; based on the capacity change rate of the balancing stack, the impact of low-frequency events on the log file is obtained using warning information data and error alarm data; the low-frequency events are rare or uncommon events.

[0054] Collect and preprocess the log files of obstetrics and gynecology medical equipment, and extract warning information data and error alarm data from the preprocessed log files, including:

[0055] Collect raw log files from obstetrics and gynecology medical equipment and perform preprocessing, including data cleaning, missing value supplementation and standardization, to generate a preprocessed log file dataset;

[0056] Use the preprocessed log file dataset to build an LSTM model, and input the log file dataset into the LSTM model for training to predict low-frequency events in the log file;

[0057] According to the prediction results of the LSTM model, warning information data and error alarm data are extracted from the log files of low-frequency events.

[0058] Further explanation is that, first, preprocessing ensures the accuracy and consistency of the data, laying a solid foundation for subsequent analysis. Secondly, the LSTM model can effectively capture complex patterns in time series data, improve the prediction accuracy of rare events (such as sudden temperature changes or sensor failures), and enhance the predictability of the system. Finally, the reasons for extracting warning information data and error alarm data are as follows: 1. Accurately identify potential problems. Through these data, the system can detect and handle anomalies earlier to avoid the expansion of faults; 2. Optimize maintenance and response, help technicians quickly locate problems, and reduce downtime and costs; 3. Improve system reliability, real-time monitoring and early warning to improve equipment safety and continuity; 4. Support decision-making, provide scientific basis, ensure that the medical team responds quickly in emergencies, and ensure patient safety. These four points together improve the overall performance and safety of the system.

[0059] According to the prediction results of the LSTM model, warning information data and error alarm data are extracted from the log files of low-frequency events, including:

[0060] Collect the prediction results of the LSTM model for low-frequency events in log files;

[0061] Identify low-frequency events that the LSTM model fails to predict, and extract warning information data and error alarm data from the log files corresponding to the low-frequency events;

[0062] Construct a fuzzy reasoning model, and use the fuzzy reasoning model to prioritize the low-frequency events predicted and not predicted by the LSTM model to obtain the classification results;

[0063] Based on the results of priority division, the time dependency and quantity ratio of warning information data and false alarm data are analyzed when low-frequency events occur.

[0064] Further explanation is that the fuzzy inference model is built to handle low-frequency events that the LSTM model fails to predict and prioritize these events. The fuzzy inference model can cope with uncertainty and complexity in the data and enhance the flexibility and robustness of the system. It can handle imprecise or incomplete data to ensure that the system can still make reasonable decisions in complex situations and optimize resource allocation. The fuzzy inference model prioritizes low-frequency events predicted and not predicted by the LSTM model according to the urgency and scope of the event. High-priority events are handled first to ensure that the most serious problems are solved first, thereby achieving rapid response and resource optimization, and avoiding waste on low-priority events. Based on the results of the prioritization, the time dependency (such as the concentrated time period) and quantity ratio (the proportion of the two in the total data volume) of the warning information data and the error alarm data when low-frequency events occur are analyzed. This helps to identify patterns, evaluate the impact of different events, predict and prepare countermeasures in advance, focus on solving major problems, and improve the system's predictability and responsiveness.

[0065] Based on the warning information data and the error alarm data, the balance stack is initialized, and the warning information data and the error alarm data are pushed into the balance stack; the conversion point between the online log file and the offline log file is determined, including:

[0066] The specific method of building a balanced stack is as follows:

[0067] Determine a fixed total capacity, which is shared by two sub-stacks, and each sub-stack has a fixed upper capacity limit;

[0068] The left stack pushes elements from the left, and the stack top pointer moves to the right; the right stack pushes elements from the right, and the stack top pointer moves to the left;

[0069] When any substack reaches its upper capacity limit, it stops accepting new elements until an element is popped out; the two substacks always maintain the same capacity, and the same operations are performed on the two substacks while adjusting the capacity;

[0070] Push operation: elements are pushed into the left stack from the left, and elements are pushed into the right stack from the right; pop operation: remove the elements at the top of each stack.

[0071] To further explain, the balanced stack achieves: Efficient resource utilization: ensuring balanced resource allocation between the two sub-stacks to prevent one-sided overload. Enhanced flexibility: supporting bidirectional data input, improving the adaptability of the system. Stability and reliability: dynamic capacity management and the LIFO principle ensure stable operation under high load and reduce the risk of data loss. Fast response: efficient push and pop operations reduce data processing delays and improve the system's response speed.

[0072] Initialize the balance stack and push the warning information data and error alarm data into the balance stack; determine the conversion point between the online log file and the offline log file, including:

[0073] Push the warning information data into the left stack of the balance stack, and push the error alarm data into the right stack of the balance stack;

[0074] Determine the switching point between the online log file and the offline log file according to the difference between the upper capacity limits of the left stack and the right stack;

[0075] Calculate the first conversion frequency and the second conversion frequency: The time when the left stack reaches the upper limit is recorded as , refers to the time from the beginning of recording to the first time the stack on the left reaches the upper limit; the time when the stack on the right reaches the upper limit is recorded as , refers to the time from the beginning of recording to the first time the right stack reaches the upper limit; when the left stack capacity reaches the upper limit before the right stack, that is < , calculate the time difference from when the left stack is full to when the right stack is also full, and determine the first conversion frequency; when the capacity of the right stack reaches the upper limit before the left stack, that is, < , the same logic applies to calculating the second switching frequency; the calculation formulas for the first switching frequency and the second switching frequency are as follows,

[0076] in, is the first switching frequency, is the second switching frequency, is the time when the left stack reaches its capacity limit, The time when the right stack reaches its upper limit; the priority of the warning information data stored in the left stack is greater than the priority of the error alarm data stored in the right stack, and the system design ensures > .

[0077] Further explanation is that in order to ensure > , that is, the conversion frequency when the left stack is full first is higher than the conversion frequency when the right stack is full first. The system design gives the left stack a higher response priority. This is because the data processed by the left stack (such as warning information data) is usually considered more urgent or frequent, resulting in the capacity limit being reached earlier. < , the system will trigger the switch from online mode to offline mode more quickly to respond to these urgent data quickly, so that On the contrary, the data processed by the right stack (such as error alarm data) is relatively less frequent, even if it is fully loaded first. < , the system response will be slightly delayed, making The value of is small. The conversion point refers to the timing of switching between online log files and offline log files based on the difference in the upper capacity limits of the left stack and the right stack. Specifically: Ensure resource balance: By monitoring the capacity of the two sub-stacks, the system can switch modes at the appropriate time to avoid overloading of one side of the stack. Optimize data processing efficiency: The setting of the conversion point ensures that under high load conditions, the system can switch online log files to offline mode in time for detailed analysis, reducing the pressure of real-time processing. The conversion frequency is used to measure and control the speed of switching between online log files and offline log files. The specific functions are as follows: Adapt to different traffic: Different conversion frequencies dynamically adjust the switching speed according to the actual data traffic to ensure that the system is always in the best state. Improve response speed: A high conversion frequency means faster switching, which helps to deal with emergencies more promptly and reduce potential risks. The first conversion frequency Greater than the second switching frequency The reason is: Warning information is processed first: The left stack stores warning information data, which is usually more urgent and requires a quick response. Therefore, when the left stack reaches its capacity limit first, the system will process it at a higher frequency. The right stack stores error alarm data, which is important but not so urgent. Therefore, when the right stack reaches the upper limit of capacity first, the system will This design ensures that the system can respond quickly when dealing with emergencies, while allocating resources reasonably to maintain system stability and efficiency.

[0078] Based on the transition points between online and offline log files, adjust the total capacity of the balancing stack, including:

[0079] Extract warning information data and error alarm data under low-frequency events from log files;

[0080] Analyze the data to determine the time dependency, i.e. the trend over time, and the quantity ratio, i.e. the proportion of the two in the total amount of data;

[0081] According to the conversion points of the online log file and the offline log file, combined with the above time dependency and quantity ratio relationship, the switching time between the online mode and the offline mode is determined;

[0082] Based on the determined transition point, the total capacity of the balancing stack is dynamically adjusted; the new total capacity of the balancing stack is calculated using a formula.

[0083] Further explanation is: Determine the time dependency: Through time series analysis and trend analysis, identify the change pattern and correlation of warning information data and error alarm data over time. Determine the quantitative proportion relationship: Through frequency distribution analysis and proportion calculation, count and quantify the relative importance and frequency of warning information data and error alarm data in the total data volume. Switching point: The switching point refers to the specific conditions or thresholds for determining the system to switch from online mode to offline mode or from offline mode to online mode based on the status of online log files and offline log files (such as capacity limit, data flow). Function: The switching point determines under what circumstances the mode switching is required to ensure that the system can switch at the appropriate time point to avoid resource overload or data processing delay. In this solution, the switching point is determined based on the difference in the capacity limit of the online log file and the offline log file. For example, when the left stack (warning information data) or the right stack (error alarm data) approaches the capacity limit, the switching point is triggered. Switching time: The switching time refers to the specific moment or time period when the mode switching is actually executed. Function: The switching time is determined based on the transition point and other related factors (such as time dependency and quantity ratio), ensuring that the switching operation occurs at the most appropriate time to optimize resource utilization and data processing efficiency. In this solution, the switching time is determined after the transition point conditions are met, combining the time dependency (trend over time) and quantity ratio (the proportion of the two in the total amount of data). For example, the growth rate of warning information data is significantly higher than that of error alarm data, which will trigger the switching time earlier and switch more log files to offline mode for detailed analysis.

[0084] Based on the determined transition point, the total capacity of the balancing stack is dynamically adjusted; the new total capacity of the balancing stack is calculated using a formula, including: The formula is as follows, in, is the total capacity of the balanced stack after adjustment, To balance the total capacity of the stack, is the current capacity of the left stack, is the current capacity of the right stack, is the time dependency coefficient of the left stack, is the time dependence coefficient of the right stack, is the ratio of the number of stacks on the left, is the ratio of the number of stacks on the right, is the weight parameter, is the scaling factor.

[0085] Further explanation is, It is the result of the formula calculation, which represents the new total capacity of the balanced stack after adjustment based on the current state and various coefficients. is the total capacity of the current balance stack, which serves as an initial reference value. Indicates the trend of the data in the left stack over time. For example, if the data in the left stack grows fast, It will be big, otherwise it will be small. Similar to , but for the right stack. It reflects the trend of the data in the right stack changing over time. Reflects the importance or frequency of the data in the left stack. It reflects the importance or frequency of the data in the right stack. Weight parameters are used to adjust the influence of different factors. Adjust according to actual needs to make some factors occupy a more important position in the calculation. The scaling factor is used to adjust the size of the calculation result as a whole to ensure that the final adjusted capacity is within a reasonable range. The formula comprehensively considers the time dependency, quantity ratio, conversion frequency, weight and scaling factor to ensure that the adjusted total capacity of the balance stack can adapt to the current data traffic and priority requirements while maintaining the stability and efficiency of the system. In this way, the system can switch modes at appropriate time points, flexibly adjust resource allocation, and optimize the processing efficiency of log files. The specific values ​​of the time dependency coefficient and quantity ratio in the formula are dynamically adjusted and regularly updated through historical data analysis and real-time statistics, combined with the experience of field experts.

[0086] Based on the capacity change rate of the balance stack, the impact of low-frequency events on the log file is obtained using warning information data and error alarm data, including:

[0087] Record the change of each balance stack capacity and calculate the capacity change rate. The capacity change rate is defined as the ratio of two adjacent capacities, that is, ,in It is The capacity of times, It is times the capacity;

[0088] Calculate two differences: collect the difference between the warning information data in the LSTM model prediction state and the warning information data in the unpredicted state, and define it as the first difference ; Collect the difference between the error alarm data in the LSTM model prediction state and the error alarm data in the unpredicted state, and define it as the second difference ;

[0089] According to the capacity change rate of the balanced stack, the first difference and the second difference , evaluate the impact of low-frequency events on log files, and use the formula to calculate the impact .

[0090] According to the capacity change rate of the balanced stack, the first difference and the second difference , evaluate the impact of low-frequency events on log files, and use the formula to calculate the impact , including: The formula is as follows, in, is the impact of low-frequency events on log files, For the The capacity of the secondary balance stack, The difference between the warning information data in the LSTM model prediction state and the warning information data in the unpredicted state, that is, the first difference , The difference between the error alarm data in the LSTM model prediction state and the error alarm data in the unpredicted state, that is, the second difference , and is the weight parameter, is the weight parameter, The number of recorded capacity changes.

[0091] Further explanation is, A comprehensive evaluation indicator of the overall system performance, combining the stability of capacity changes and the impact of prediction errors. and These two weight parameters adjust the capacity change stability term and the prediction error term to the final index respectively. degree of impact. Indicates the importance of adjusting for capacity change stability terms. Indicates the importance of adjusting the forecast error term. Indicates that after The total capacity of the balance stack after adjustment. is the capacity after the next adjustment. Used to evaluate the effects of multiple adjustments. Indicates The change rate of the new capacity relative to the old capacity after the adjustment. The square is taken to amplify the impact of a larger change rate and ensure that large capacity fluctuations have a significant impact on the comprehensive index. The impact is greater. It means summing up all the adjustments and evaluating the overall stability of multiple adjustments. It reflects the change in the amount of warning information data when the LSTM model is not predicted. It reflects the change in the amount of false alarm data when the LSTM model is not predicted. For adjustment and The degree of influence on the final forecast error term. If the value is large, the change of warning information data will have a greater impact on the prediction error; otherwise, more attention will be paid to the false alarm data.

[0092] Embodiment 2: Based on Embodiment 1, a log file processing system for obstetrics and gynecology medical equipment data includes:

[0093] The log file collection and preprocessing module collects log files from obstetrics and gynecology medical equipment and performs data cleaning, missing value supplementation and standardization processing;

[0094] LSTM model prediction module, builds an LSTM model, puts the preprocessed log file dataset into the LSTM model, and predicts low-frequency events in the log file;

[0095] Fuzzy reasoning module, builds a fuzzy reasoning model, prioritizes the low-frequency events predicted and not predicted by the LSTM model, and obtains the time dependency and quantity ratio relationship;

[0096] The balanced stack management module defines and manages a balanced stack consisting of two sub-stacks of the same volume, supporting data entry from the left and right sides respectively, and following the LIFO principle;

[0097] A conversion point determination module determines the conversion point between the online log file and the offline log file according to the difference between the upper capacity limits of the left stack and the right stack;

[0098] Dynamic adjustment module, dynamically adjusts the total capacity of the balancing stack based on the conversion points between online log files and offline log files;

[0099] The impact assessment module records the capacity change rate of each balancing stack, and calculates the impact of low-frequency events on log files by combining the difference between the warning information data predicted by the LSTM model and the warning information data not predicted, as well as the difference between the false alarm data.

[0100] The above is a detailed introduction to the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A log file processing method for obstetrics and gynecology medical equipment data, characterized in that: The following steps are involved: S1: Collect and preprocess the log files of obstetrics and gynecology medical equipment, and extract warning information data and error alarm data from the preprocessed log files; S2: Initialize the balance stack based on the warning information data and the error alarm data, and push the warning information data and the error alarm data into the balance stack; determine the switching point between the online log file and the offline log file; the balance stack is composed of two sub-stacks with the same capacity, the left stack: defined as a sub-stack that supports data insertion from the left end, all new elements are pushed in from the left end of the stack, following the LIFO principle; the right stack: defined as a sub-stack that supports data insertion from the right end, all new elements are pushed in from the right end of the stack, also following the LIFO principle; the switching point refers to the timing of switching between the online log file and the offline log file determined according to the difference in the upper limit of the capacity of the left stack and the right stack; S3: Based on the conversion point between the online log file and the offline log file, the total capacity of the balancing stack is adjusted; based on the capacity change rate of the balancing stack, the impact of low-frequency events on the log file is obtained using the warning information data and the error alarm data; the low-frequency events are rare or uncommon events; The collecting and preprocessing of log files of obstetrics and gynecology medical equipment, and extracting warning information data and error alarm data from the preprocessed log files, include: Collect raw log files from obstetrics and gynecology medical equipment and perform preprocessing, including data cleaning, missing value supplementation and standardization, to generate a preprocessed log file dataset; Use the preprocessed log file dataset to build an LSTM model, and input the log file dataset into the LSTM model for training to predict low-frequency events in the log file; According to the prediction results of the LSTM model, the warning information data and error alarm data are extracted from the log files of low-frequency events; The capacity change rate based on the balance stack and the warning information data and the error alarm data are used to obtain the impact of low-frequency events on the log file, including: Record the change of each balance stack capacity and calculate the capacity change rate. The capacity change rate is defined as the ratio of two adjacent capacities, that is, ,in It is The capacity of times, It is times the capacity; Calculate two differences: collect the difference between the warning information data in the LSTM model prediction state and the warning information data in the unpredicted state, and define it as the first difference ; Collect the difference between the error alarm data in the LSTM model prediction state and the error alarm data in the unpredicted state, and define it as the second difference ; According to the capacity change rate of the balanced stack, the first difference and the second difference , evaluate the impact of low-frequency events on log files, and use the formula to calculate the impact ; The capacity change rate of the balance stack and the first difference and the second difference , evaluate the impact of low-frequency events on log files, and use the formula to calculate the impact , including: The formula is as follows, in, is the impact of low-frequency events on log files, For the The capacity of the secondary balance stack, The difference between the warning information data in the LSTM model prediction state and the warning information data in the unpredicted state, that is, the first difference , The difference between the error alarm data in the LSTM model prediction state and the error alarm data in the unpredicted state, that is, the second difference , and is the weight parameter, is the weight parameter, The number of recorded capacity changes.

2. A log file processing method for obstetrics and gynecology medical equipment data according to claim 1, characterized in that: According to the prediction results of the LSTM model, warning information data and error alarm data are extracted from the log files of low-frequency events, including: Collect the prediction results of the LSTM model for low-frequency events in log files; Identify low-frequency events that the LSTM model fails to predict, and extract warning information data and error alarm data from the log files corresponding to the low-frequency events; Construct a fuzzy reasoning model, and use the fuzzy reasoning model to prioritize the low-frequency events predicted and not predicted by the LSTM model to obtain the classification results; Based on the results of priority division, the time dependency and quantity ratio of warning information data and false alarm data are analyzed when low-frequency events occur.

3. A log file processing method for obstetrics and gynecology medical equipment data according to claim 1, characterized in that: The method of initializing the balance stack based on the warning information data and the error alarm data, and pushing the warning information data and the error alarm data into the balance stack; and determining the conversion point between the online log file and the offline log file includes: The specific method of building a balanced stack is as follows: Determine a fixed total capacity, which is shared by two sub-stacks, and each sub-stack has a fixed upper capacity limit; The left stack pushes elements from the left, and the stack top pointer moves to the right; the right stack pushes elements from the right, and the stack top pointer moves to the left; When any substack reaches its upper capacity limit, it stops accepting new elements until an element is popped out; the two substacks always maintain the same capacity, and the same operations are performed on the two substacks while adjusting the capacity; Push operation: The left stack pushes elements from the left, and the right stack pushes elements from the right; Pop operation: remove the element at the top of each stack.

4. A log file processing method for obstetrics and gynecology medical equipment data according to claim 3, characterized in that: Initializing the balancing stack, pushing the warning information data and the error alarm data into the balancing stack; Determine the transition points between online log files and offline log files, including: Push the warning information data into the left stack of the balance stack, and push the error alarm data into the right stack of the balance stack; Determine the switching point between the online log file and the offline log file according to the difference between the upper capacity limits of the left stack and the right stack; Calculate the first conversion frequency and the second conversion frequency: The time when the left stack reaches the upper limit is recorded as , refers to the time from the beginning of recording to the first time the stack on the left reaches the upper limit; the time when the stack on the right reaches the upper limit is recorded as , refers to the time from the beginning of recording to the first time the right stack reaches the upper limit; when the left stack capacity reaches the upper limit before the right stack, that is < , calculate the time difference from when the left stack is full to when the right stack is also full, and determine the first conversion frequency; when the capacity of the right stack reaches the upper limit before the left stack, that is, < , the same logic applies to calculating the second switching frequency; the calculation formulas for the first switching frequency and the second switching frequency are as follows, , ,in, is the first switching frequency, is the second switching frequency, is the time when the left stack reaches its capacity limit, The time when the right stack reaches its upper limit; the priority of the warning information data stored in the left stack is greater than the priority of the error alarm data stored in the right stack, and the system design ensures > .

5. A log file processing method for obstetrics and gynecology medical equipment data according to claim 1, characterized in that: The adjusting the total capacity of the balancing stack based on the conversion point of the online log file and the offline log file includes: Extract warning information data and error alarm data under low-frequency events from log files; Analyze the data to determine the time dependency, i.e. the trend over time, and the quantity ratio, i.e. the proportion of the two in the total amount of data; According to the conversion points of the online log file and the offline log file, combined with the above time dependency and quantity ratio relationship, the switching time between the online mode and the offline mode is determined; Based on the determined transition point, the total capacity of the balancing stack is dynamically adjusted; the new total capacity of the balancing stack is calculated using a formula.

6. A log file processing method for obstetrics and gynecology medical equipment data according to claim 5, characterized in that: The method dynamically adjusts the total capacity of the balancing stack based on the determined transition point; Use the formula to calculate the new total capacity of the balanced stack, including: The formula is as follows, in, is the total capacity of the balanced stack after adjustment, To balance the total capacity of the stack, is the current capacity of the left stack, is the current capacity of the right stack, is the time dependency coefficient of the left stack, is the time dependence coefficient of the right stack, is the ratio of the number of stacks on the left, is the ratio of the number of stacks on the right, is the weight parameter, is the scaling factor.

7. A log file processing system for obstetrics and gynecology medical equipment data, used in the log file processing method for obstetrics and gynecology medical equipment data as claimed in any one of claims 1 to 6, characterized in that: Included are: The log file collection and preprocessing module collects log files from obstetrics and gynecology medical equipment and performs data cleaning, missing value supplementation and standardization processing; LSTM model prediction module, builds an LSTM model, puts the preprocessed log file dataset into the LSTM model, and predicts low-frequency events in the log file; Fuzzy reasoning module, builds a fuzzy reasoning model, prioritizes the low-frequency events predicted and not predicted by the LSTM model, and obtains the time dependency and quantity ratio relationship; The balanced stack management module defines and manages a balanced stack consisting of two sub-stacks of the same volume, supporting data entry from the left and right sides respectively, and following the LIFO principle; A conversion point determination module determines the conversion point between the online log file and the offline log file according to the difference between the upper capacity limits of the left stack and the right stack; Dynamic adjustment module, dynamically adjusts the total capacity of the balancing stack based on the conversion points between online log files and offline log files; The impact assessment module records the capacity change rate of each balancing stack, and calculates the impact of low-frequency events on log files by combining the difference between the warning information data predicted by the LSTM model and the warning information data not predicted, as well as the difference between the false alarm data.

Images