Hydropower station equipment state-based data prediction model subdivision method and system

Abstract

The invention relates to the technical field of equipment state monitoring, in particular to a data prediction model subdivision method and system based on hydropower station equipment states, and the method comprises the following steps: collecting the operation data of hydropower station equipment, sorting the operation data according to a time sequence, and removing the noise and abnormal values of the data; and obtaining a processed time sequence data set. According to the method, trend and periodic change in data and statistical characteristic data input into the hidden Markov model can be effectively identified by calculating the average value and the standard deviation of equipment operation, the dynamic relation between the operation states of the equipment is disclosed through state transition analysis, cognition of equipment early warning and fault stages is enhanced, and the reliability of the equipment early warning and fault stages is improved. The calculation of the real-time state probability not only predicts the health condition of the equipment in the next week, but also can accurately evaluate the severity of the alarm stage, provides a feasible maintenance plan, prolongs the service life of the equipment, and facilitates the formulation of an energy-saving operation strategy.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment status monitoring, and in particular to a data prediction model subdivision method and system based on the equipment status of a hydropower station. Background Art

[0002] The field of equipment condition monitoring technology mainly involves the use of sensors and monitoring systems to track and evaluate the performance and status of industrial equipment. This technology is widely used in predictive maintenance, fault diagnosis and performance optimization. By collecting equipment operation data (such as temperature, pressure, vibration and other parameters) in real time or periodically, continuous monitoring of equipment health can be achieved. By analyzing the data, potential problems and signs of failure can be discovered in a timely manner, allowing timely maintenance or replacement to avoid equipment downtime and production losses, which is of great significance to improving the safety and efficiency of industrial operations.

[0003] Among them, the data prediction model segmentation method for the status of hydropower station equipment refers to building a mathematical model for predicting the future status of equipment by analyzing the equipment operation data in the hydropower station. This model can help hydropower station managers understand the problems of equipment in advance, carry out timely maintenance and adjustments, optimize equipment operation efficiency, extend equipment service life, and reduce maintenance costs. This prediction model can also predict the energy consumption of equipment to a certain extent and help formulate a more energy-saving operation strategy.

[0004] Existing equipment condition monitoring technology has some deficiencies in actual operation. Although it is able to collect parameters such as temperature, pressure, and vibration, it lacks effective data preprocessing, resulting in noise and outliers interfering with the accuracy of data analysis. Existing technologies ignore in-depth analysis of data trends and periodic changes, limiting the possibility of a deep understanding of the health status of equipment. The lack of such in-depth analysis results in the prediction of potential equipment problems not being timely or accurate enough, increasing the risk of sudden failures. For example, if equipment failure is not accurately predicted, it will cause serious production stoppages and economic losses. The deficiencies show that although existing technologies provide basic monitoring functions, there is still room for improvement in data processing and prediction accuracy, especially in long-term health management and efficiency optimization of equipment. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings existing in the prior art and to propose a data prediction model segmentation method and system based on the status of hydropower station equipment.

[0006] In order to achieve the above object, the present invention adopts the following technical solution, a data prediction model segmentation method based on the status of hydropower station equipment, comprising the following steps: S1: Collect the operation data of the hydropower station equipment, sort the operation data according to the time series, and remove the noise and outliers of the data to obtain the processed time series data set; S2: Based on the processed time series data set, the operating average and standard deviation of the hydropower station equipment are calculated, and the trend and periodic changes in the data are identified through statistical indicators to obtain statistical characteristic data; S3: inputting the statistical feature data into a hidden Markov model, setting the number of hidden states of the model, calculating the state transition probability, the state including normal operation, warning and fault stages, and obtaining the state transition analysis result; S4: Based on the state transition analysis results, the real-time probability of the equipment state at the target time point is calculated, and the health status of the hydropower station equipment in the next week is predicted by using the data prediction model to obtain the real-time state probability calculation result; S5: extracting the key alarm stage from the real-time state probability calculation result, evaluating the severity of the impact of the alarm stage on the equipment health, providing maintenance plans and measures, and obtaining key alarm analysis results; S6: Integrate the key alarm analysis results with the data prediction model, adjust the penalty coefficient and kernel parameters, and predict the health status of the equipment through cross-validation to obtain the equipment health monitoring results.

[0007] As a further solution of the present invention, the processed time series data set includes stability data points arranged in chronological order, filtered continuous monitoring data and data sequences marked with timestamps, the statistical feature data includes the moving average of the time series, the sliding window standard deviation and the identified data growth trend and seasonal pattern, the state transition analysis results include the probability distribution diagram between differentiated states, the average duration of differentiated states and the frequency table of state transitions, the real-time state probability calculation results include the predicted probability value of each state, the time distribution of state probability and the probability diagram of state changes within the prediction period, the key alarm analysis results include the time point of the key alarm, the time period of the alarm and the evaluation results of the impact of the alarm on the overall operation of the equipment, and the equipment health monitoring results include the predicted health status trend, the confidence interval of the health status and the equipment performance changes updated based on the monitoring data.

[0008] As a further solution of the present invention, the steps of collecting the operation data of the hydropower station equipment, sorting the operation data according to the time series, and removing the noise and outliers of the data to obtain the processed time series data set are specifically as follows: S101: Collecting the operation data of the hydropower station equipment, synchronizing the operation data of the equipment to check the integrity of the information, uploading and storing the data in the data buffer, and generating an original data set; S102: Based on the original data set, the equipment operation data is sorted using timestamp tags, the data sequence is adjusted to verify the continuity of the timeline, and the timestamp integrity is checked to avoid data missing, so as to generate time series data; S103: According to the time series data, identify and remove statistical outliers and noise data that deviates from the normal state, iteratively optimize data quality, and generate a processed time series data set.

[0009] As a further solution of the present invention, based on the processed time series data set, the operating average value and standard deviation of the hydropower station equipment are calculated, and the trend and periodic changes in the data are identified by statistical indicators to obtain the statistical characteristic data. Specifically, the steps are as follows: S201: calculating the average operating data of the hydropower station equipment through the processed time series data set, determining the average operating level at the target time point by using accumulation and division operations, and generating an equipment operating average data table; S202: Calculate the standard deviation of the hydropower station equipment operation data based on the equipment operation average value data table, determine the fluctuation range of the data using the difference and square method, and generate a standard deviation calculation result; S203: Analyze the trend and periodic changes in the data using the standard deviation calculation result, identify the repetitive patterns and change trends of the data by recording the peaks and valleys in the data, and generate statistical feature data.

[0010] As a further solution of the present invention, the statistical feature data is input into a hidden Markov model, the number of hidden states of the model is set, and the state transition probability is calculated. The states include normal operation, early warning and fault stages. The steps of obtaining the state transition analysis result are specifically as follows: S301: inputting the statistical feature data into a hidden Markov model, setting the number of hidden states in the model according to the equipment operation status data, including normal operation, warning and fault stages, and generating model initialization settings; S302: Based on the model initialization setting, frequency analysis is used to count the number of state transitions in the statistical feature data, and the probability of each state transition is calculated using the statistical results to generate a state transition probability matrix; S303: Using the state transition probability matrix and a hidden Markov model, a dynamic simulation of the operating state of the equipment is performed, and the state transition trend within the next week is predicted through the model, potential fault development paths are identified, and a state transition analysis result is generated.

[0011] As a further solution of the present invention, the formula of the hidden Markov model is as follows:

[0012] in, is the state transition trend value, Represents at time From the status To status The transition probability, Represents the intermediate state To status The basic transition probability, Represents the state To status The transfer weight, is the number of states.

[0013] As a further solution of the present invention, based on the state transition analysis results, the real-time probability of the equipment state at the target time point is calculated, and the health status of the hydropower station equipment in the next week is predicted by using the data prediction model. The steps of obtaining the real-time state probability calculation result are specifically as follows: S401: According to the state transition analysis result, a target time point is selected, the real-time probability of the differentiated state of the device is calculated, the state transition probability is multiplied by the real-time state probability, the probability of the differentiated state is determined, and real-time state probability estimation data is generated; S402: Based on the real-time status probability estimation data, using the data prediction model, adopting the time series analysis method to predict the health status of the equipment in the next week, analyzing the potential fault warning and the probability change of normal operation, and generating a short-term health prediction result; S403: Using the short-term health prediction results, evaluate and verify the status changes of the hydropower station equipment in the next week, provide a scientific basis for maintenance and operation decisions, and generate real-time status probability calculation results.

[0014] As a further solution of the present invention, the steps of extracting the key alarm stage from the real-time state probability calculation result, evaluating the severity of the impact of the alarm stage on the equipment health, providing maintenance plans and measures, and obtaining the key alarm analysis result are specifically as follows: S501: Based on the real-time status probability calculation result, by comparing the probability values ​​of the differentiated alarm stages, identifying the key alarm stages that affect the health of the equipment, and generating key alarm status data; S502: Evaluate the severity of the critical alarm state through the critical alarm state data, analyze the potential impact on the health of the equipment with reference to the alarm state duration and probability value, calculate the damage degree, and generate an alarm impact assessment result; S503: Based on the alarm impact assessment results, formulate targeted maintenance plans and preventive measures, adjust maintenance frequency and inspection content, verify the maintenance efficiency of measures taken in the alarm stage on equipment damage, and generate key alarm analysis results.

[0015] As a further solution of the present invention, the key alarm analysis results are integrated with the data prediction model, the penalty coefficient and the kernel parameter are adjusted, and the health status of the equipment is predicted through cross-validation to obtain the equipment health monitoring results. Specifically, the steps are as follows: S601: Integrate the key alarm analysis results with the data prediction model, adjust the penalty coefficient and kernel parameters in the model, optimize the prediction response speed by increasing the weight of the data and changing the sensitivity of the kernel, and generate model parameter adjustment data; S602: adjusting the data through the model parameters, applying the adjusted parameter settings, performing cross-validation to evaluate the stability of the model, using the segmented data sets to test the model multiple times, analyzing the consistency of each prediction result with the real-time data, and generating cross-validation results; S603: Based on the cross-validation results, use the optimized data prediction model to predict the health status of the equipment in the next week, evaluate the probability and timing of occurrence, evaluate the health of the hydropower station equipment, and generate equipment health monitoring results.

[0016] A data prediction model segmentation system based on the status of hydropower station equipment, the data prediction model segmentation system based on the status of hydropower station equipment is used to execute the data prediction model segmentation method based on the status of hydropower station equipment, the system comprises: The data processing module collects the operating data of the hydropower station equipment, sorts them by time series, removes noise and outliers, and generates a processed time series data set; The feature analysis module calculates the running mean and standard deviation based on the processed time series data set, identifies trends and periodic changes, and obtains statistical feature data; The transition probability calculation module inputs the statistical feature data into the hidden Markov model, sets the number of hidden states, calculates the state transition probability, including normal operation, warning and fault stages, and obtains the state transition analysis result; The state prediction module calculates the real-time probability of the device state at the target time point based on the state transition analysis result, predicts the health status of the device within the next week, and obtains the real-time state probability calculation result; The alarm stage evaluation module extracts the key alarm stage from the real-time status probability calculation results, evaluates the impact on equipment health, integrates the key alarm analysis results with the data prediction model, adjusts the penalty coefficient and kernel parameters, predicts the health status of the equipment through cross-validation, and obtains the equipment health monitoring results.

[0017] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, the prediction of the health status of the equipment is optimized by collecting and advanced processing the operating data of the hydropower station equipment. After cleaning the noise and outliers in the data, the obtained time series data set is more accurate, providing a high-quality foundation for subsequent data analysis. By calculating the mean and standard deviation of the equipment operation, the trend and periodic changes in the data can be effectively identified, and the statistical characteristic data provides the monitoring system with important information that can reflect the actual operating status. The data input into the hidden Markov model further reveals the dynamic relationship between the operating states of the equipment through state transition analysis, and enhances the understanding of the equipment early warning and failure stages. The calculation of real-time state probability not only predicts the health status of the equipment in the next week, but also can accurately evaluate the severity of the alarm stage, and provide a practical maintenance plan, which greatly improves the equipment operation efficiency and safety, reduces maintenance costs, extends the life of the equipment, and helps to formulate energy-saving operation strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the workflow of the present invention; Figure 2 This is a detailed flow chart of S1 of the present invention; Figure 3 This is a detailed flow chart of S2 of the present invention; Figure 4 This is a detailed flow chart of S3 of the present invention; Figure 5 This is a detailed flow chart of S4 of the present invention; Figure 6 This is a detailed flow chart of S5 of the present invention; Figure 7 This is a detailed flow chart of S6 of the present invention; Figure 8 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0019] 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.

[0020] 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.

[0021] See also Figure 1 The present invention provides a technical solution, a data prediction model segmentation method based on the status of hydropower station equipment, comprising the following steps: S1: Collect the operating data of the hydropower station equipment, including voltage, current, temperature and vibration level, sort the operating data according to time series, and remove the noise and outliers of the data to obtain the processed time series data set; S2: Based on the processed time series data set, the operating average and standard deviation of the hydropower station equipment are calculated, and the trend and periodic changes in the data are identified through statistical indicators to obtain statistical characteristic data; S3: Input the statistical feature data into the hidden Markov model, set the number of hidden states of the model, calculate the state transition probability, the state corresponds to the normal operation, warning and failure stages of the equipment, determine the transition path of the equipment state by training the model, and obtain the state transition analysis result; S4: Based on the state transition analysis results, analyze and calculate the real-time probability of the equipment status at the target time point, use the data prediction model to predict the health status of the hydropower station equipment in the next week, and obtain the real-time state probability calculation results; S5: Extract the key alarm stage from the real-time status probability calculation results, evaluate the severity of the impact of the alarm stage on equipment health, provide maintenance plans and measures, and obtain key alarm analysis results; S6: Integrate the key alarm analysis results with the data prediction model, adjust the penalty coefficient and kernel parameters, check the generalization ability of the model through cross-validation, classify the equipment operation status, predict the health status of the equipment, and obtain the equipment health monitoring results.

[0022] The processed time series data set includes stability data points arranged in chronological order, filtered continuous monitoring data, and data series with timestamps. The statistical feature data includes the moving average of the time series, the sliding window standard deviation, and the identified data growth trend and seasonal pattern. The state transition analysis results include the probability distribution diagram between differentiated states, the average duration of differentiated states, and the frequency table of state transitions. The real-time state probability calculation results include the predicted probability value of each state, the time distribution of state probability, and the probability diagram of state changes during the prediction period. The key alarm analysis results include the time points of key alarms, the time period of alarm duration, and the evaluation results of the impact of alarms on the overall operation of the equipment. The equipment health monitoring results include the predicted health status trend, the confidence interval of the health status, and the changes in equipment performance based on the monitoring data update.

[0023] See also Figure 2 , collect the operation data of the hydropower station equipment, sort the operation data according to the time series, and remove the noise and outliers of the data. The specific steps to obtain the processed time series data set are: S101: Collect the operation data of the hydropower station equipment, synchronize the operation data of the equipment to check the integrity of the information, upload and store the data in the data buffer, and generate the execution process of the original data set as follows; The task of collecting the operating data of the hydropower station equipment is to ensure that the information of each device is correctly recorded and transmitted by synchronously checking the integrity of the information. The operating data of each device is integrated through the steps of uploading and storing the data in the data buffer. The data set contains the equipment operation status at each time point, which provides a basis for subsequent data processing and analysis. Through a series of processes, the integrity and accuracy of the data are guaranteed. During the data uploading and caching process, the integrity and storage of the data need to be checked at each step to ensure that the data can be accurately recorded and used to generate the original data set.

[0024] S102: Based on the original data set, the equipment operation data is sorted using timestamp tags, the data sequence is adjusted to verify the continuity of the timeline, and the timestamp integrity is checked to avoid data missing. The execution process of generating time series data is as follows; Based on the original data set, sort the equipment operation data according to the formula , calculate the time series data set. In the formula, represents the original timestamped dataset, Represents the operation of sorting by timestamp. Formula explanation and formula calculation derivation process: Consider the timestamp data set , where each Indicates a specific point in time. Sorting operation Arrange the data in chronological order to ensure the continuity and integrity of the data timeline. , after sorting ,The process ensures the correct sequence of the time data and provides preparation for further data processing.

[0025] S103: According to the time series data, identify and remove statistical outliers and noise data that deviates from the normal state, iteratively optimize the data quality, and generate the processed time series data set. The execution process is as follows; According to the time series data, the data quality is iteratively optimized by identifying and removing statistical anomalies and noise data that deviates from the normal state. In this way, the data is cleaned, the noise and anomalies that do not conform to the normal operating range are removed, and the reliability and validity of the data set are enhanced. Through the iterative process, the data processing strategies and parameters are continuously adjusted to ensure that the actual operating status of the equipment is reflected, providing accurate reference for subsequent data analysis and equipment maintenance. Through continuous data quality optimization, equipment management and operation and maintenance personnel can more effectively formulate equipment monitoring and maintenance strategies and generate processed time series data sets.

[0026] See also Figure 3 , based on the processed time series data set, the operating average and standard deviation of the hydropower station equipment are calculated, and the trends and periodic changes in the data are identified through statistical indicators. The specific steps to obtain statistical characteristic data are as follows: S201: Calculate the average value of the operating data of the hydropower station equipment through the processed time series data set, use the accumulation and division operations to determine the average operating level at the target time point, and generate the equipment operating average value data table. The execution process is as follows; The average operating data of the hydropower station equipment is calculated through the processed time series data set, according to the formula , calculate the equipment operation average data table. In the formula, Represents the equipment operation data at each time point, represents the total number of data points, Represents the cumulative sum of all running data. Detailed explanation of the formula and formula calculation process: Consider a time series data set with data points, with specific values The accumulation operation calculates the sum of all data points, that is, . This sum is divided by the total number of data points , get the average value of the equipment operation For example, if and ,but ,and then The calculation display determines the average operating level of the equipment at the target time point through averaging processing, providing basic data for further analysis.

[0027] S202: Based on the equipment operation average data table, the standard deviation of the hydropower station equipment operation data is calculated, and the fluctuation range of the data is determined using the difference and square method. The execution process of generating the standard deviation calculation result is as follows; Based on the equipment operation average value data table, the fluctuation range of the hydropower station equipment operation data is estimated by calculating the difference between each data point and the average value and then squaring the difference. The method of calculating the standard deviation not only shows the average fluctuation size of the data, but also provides a quantitative assessment of the data volatility. The standard deviation value can be accurately obtained by summing the square difference and dividing the number of data points. The standard deviation can accurately reflect the stability and abnormal fluctuations of the data. It is an important parameter for evaluating the equipment operation status and making maintenance decisions, and generates the standard deviation calculation results.

[0028] S203: Using the standard deviation calculation result, analyzing the trend and periodic changes in the data, identifying the repetitive patterns and change trends of the data by recording the peaks and valleys in the data, and generating statistical feature data. The execution process is as follows; The standard deviation calculation results are used to analyze the trends and periodic changes in the operating data of hydropower station equipment. By recording and analyzing the peaks and troughs in the data in detail, not only can the repeated patterns of the data be identified, but also the changing trends of the data can be observed, helping operation and maintenance personnel understand the performance of the equipment under different operating conditions, predict the time when equipment fails or requires maintenance, and further provide data-driven insights and decision support, making equipment management more scientific and accurate, and generating statistical characteristic data.

[0029] See also Figure 4 , input the statistical feature data into the hidden Markov model, set the number of hidden states of the model, calculate the state transition probability, the state includes normal operation, warning and failure stages, and the steps to obtain the state transition analysis results are as follows: S301: Inputting statistical feature data into the hidden Markov model, setting the number of hidden states in the model according to the equipment operation status data, including normal operation, warning and fault stages, and generating the execution flow of the model initialization setting as follows; In the process of inputting statistical feature data into the hidden Markov model, it is necessary to set the number of hidden states in the model according to the equipment operation status data. This not only involves the initialization setting of the model, but also includes the subdivision of the equipment operation status, including three states: normal operation, warning and fault stage. The model can simulate and predict the operation status of the equipment more accurately. The initialization setting of the model is a key step in the successful application of the hidden Markov model, which directly affects the accuracy of subsequent state identification and prediction. Through reasonable hidden state settings, the model can effectively capture various dynamic changes in equipment operation, provide a scientific basis for equipment maintenance and fault prevention, and generate model initialization settings.

[0030] S302: Based on the model initialization settings, the frequency analysis is used to count the number of state transitions in the statistical feature data, and the probability of each state transition is calculated using the statistical results to generate the state transition probability matrix. The execution process is as follows; Based on the model initialization settings, the frequency analysis of the number of state transitions in the statistical feature data is performed according to the formula , calculate the state transition probability matrix. In the formula, Represents the state Transfer to state The number of times, Representation status The total number of occurrences. Detailed explanation of the formula and the process of formula calculation: Consider the model has states, let the state set be ,in It is recorded in the statistical feature data from the state To status The number of transitions. Each state Total number of occurrences is the sum of the number of times the state is transferred to the remaining states, that is, . Calculate from the state To status The transition probability As follows: If in state Next, there are 5 transitions to state ,state Appears 10 times, then The formula reflects the state transition probability based on statistical data, which is crucial for understanding the pattern of device state changes and predicting future device state changes.

[0031] S303: Use the state transition probability matrix and the hidden Markov model to dynamically simulate the operation state of the equipment, and use the model to predict the state transition trend in the next week, identify potential fault development paths, and generate the state transition analysis results. The execution process is as follows; The formula of hidden Markov model is as follows:

[0032] in, is the state transition trend value, Represents at time From the status To status The transition probability, Represents the intermediate state To status The basic transition probability, Represents the state To status The transfer weight, is the number of states.

[0033] The formula is used to calculate the state To status At the next time step The calculation of the transition probability includes three main parameters: , ,as well as .

[0034] It's in time From the status To the intermediate state The probability is obtained by statistical methods based on the operating data and reflects the transition probability of the device from state to state. arrive For example, if in the past 100 observations, the transition frequency from state To status There are 20 transfers, is 0.2.

[0035] From any intermediate state To the target state The basic transition probability is obtained through long-term operation data analysis. To status Transfers account for all transfers from 30% of the outgoing transfers, then is 0.3.

[0036] is an adjustment factor used to adjust the Directly to the status The transfer weight is customized according to the specific operating conditions of the equipment or external environmental factors. For example, if the state To status The transfer is more frequent, Set to 1.1.

[0037] To calculate , Substitute the above parameters into the formula. Assuming the device has three states, calculate the transition probability from state 1 to state 3, given the following data: , , .

[0038] Calculation formula:

[0039]

[0040]

[0041]

[0042] The results show that at the next time step ,The probability of the equipment transition from state 1 to state 3 is 0.66, which shows the state transition trend of the equipment after considering certain environmental factors.,A higher probability indicates that the equipment has a higher risk of transitioning from state 1 to state 3, prompting the maintenance team to pay attention and take preventive measures.

[0043] See also Figure 5 Based on the state transition analysis results, the real-time probability of the equipment state at the target time point is calculated. The health status of the hydropower station equipment in the next week is predicted using the data prediction model. The specific steps to obtain the real-time state probability calculation results are as follows: S401: According to the state transition analysis result, a target time point is selected, the real-time probability of the differentiated state of the device is calculated, the state transition probability is multiplied by the real-time state probability, the probability of the differentiated state is determined, and the execution process of generating the real-time state probability estimation data is as follows; Select the target time point, multiply the state transition probability by the real-time state probability, and use the formula , calculate the real-time probability of the device's differentiated state. In the formula, Represents at time The state transition probability, Represents at time The current state probability. Set at time point , state transition probability is 0.8, which represents the probability of the device transferring from the current state to another state, the real-time state probability is 0.5, indicating the probability that the device is currently in a specific state. The probability of being in the differentiated state is: The calculation shows how to use state probability to estimate the real-time state of the equipment at a specific point in time, which is crucial for responding to equipment changes and adjusting operating strategies in real time.

[0044] S402: Based on the real-time status probability estimation data, using the data prediction model, adopting the time series analysis method to predict the health status of the equipment in the next week, analyzing the probability changes of potential fault warnings and normal operation, and generating the short-term health prediction results. The execution process is as follows; Based on real-time status probability prediction data, the health status of the equipment in the next week is predicted through time series analysis methods. The process involves in-depth analysis of the current status probability data to predict potential fault warnings and probability changes in normal operation. The application of time series analysis enables the data prediction model to estimate future status change trends based on existing data, helping the operation and maintenance team to identify potential problems in advance, optimize maintenance strategies, ensure the continuity and safety of equipment operation, and generate short-term health prediction results.

[0045] S403: Using the short-term health prediction results, evaluate and verify the status changes of the hydropower station equipment in the next week, provide a scientific basis for maintenance and operation decisions, and generate the execution process of real-time status probability calculation results as follows; Using the short-term health prediction results, the status changes of the hydropower station equipment in the next week are evaluated and verified. The process is not only based on the data of the prediction model, but also involves scientific support for actual operation and maintenance decisions. By analyzing the matching degree between the prediction results and the actual operation data, the accuracy and reliability of the prediction model can be verified, and specific data basis can be provided for the maintenance and operation decisions of the equipment. Through the real-time status probability calculation results, the operation and maintenance team can more accurately adjust the maintenance plan and operation strategy to ensure the long-term stable operation of the equipment and generate real-time status probability calculation results.

[0046] See also Figure 6 , extract the key alarm stage from the real-time state probability calculation results, evaluate the severity of the impact of the alarm stage on equipment health, and provide maintenance plans and measures. The specific steps to obtain the key alarm analysis results are: S501: Based on the real-time status probability calculation result, by comparing the probability values ​​of the differentiated alarm stages, the critical alarm stages that affect the health of the equipment are identified, and the execution process of generating the critical alarm status data is as follows; Based on the real-time status probability calculation results, by comparing the probability values ​​of different alarm stages, the key alarm stages that affect equipment health can be identified. The process involves data analysis and probability comparison to determine the alarm stage that causes equipment performance degradation or damage. The alarm stage with the greatest impact on equipment health can be accurately selected from the many alarm stages. It not only indicates the specific alarm stage, but also provides the specific probability value of the stage, which is very important for subsequent maintenance and decision-making processes, ensuring the pertinence and effectiveness of equipment maintenance work and generating key alarm status data.

[0047] S502: The execution process of evaluating the severity of the critical alarm status through the critical alarm status data, analyzing the potential impact on the health of the equipment with reference to the duration and probability value of the alarm status, calculating the degree of damage, and generating the alarm impact assessment result is as follows; Based on the critical alarm status data, the severity of the critical alarm status is evaluated according to the formula , calculate the alarm impact assessment results. Where, represents the probability of the alarm state over time, Represents the differential element of time. Detailed explanation of the formula and the derivation process of the formula calculation: Setting the alarm state probability function Indicates from time arrive The probability of an alarm state is calculated by integrating the function to obtain the total impact of the alarm state in a given time period. For example, if Keep it as a constant 0.5 in the time interval [0,1], then The calculation process shows how to evaluate the potential impact on equipment health through the probability value and duration of the alarm state, providing a quantitative basis for taking appropriate maintenance measures.

[0048] S503: Based on the alarm impact assessment results, formulate targeted maintenance plans and preventive measures, adjust the maintenance frequency and inspection content, verify the maintenance efficiency of the measures in the alarm stage on equipment damage, and generate the execution process of key alarm analysis results as follows; Develop targeted maintenance plans and preventive measures based on the alarm impact assessment results. The process involves adjusting maintenance frequency and inspection content to address the identified critical alarm stages. Through adjustments, the efficiency of maintenance of potential damage to equipment can be increased. Verifying measures at the alarm stage ensures the timeliness and effectiveness of equipment maintenance, providing a practical basis for equipment operation and maintenance. By systematically analyzing alarm impacts and maintenance efficiency, equipment management strategies can be optimized, potential future equipment failures and maintenance costs can be reduced, and critical alarm analysis results can be generated.

[0049] See also Figure 7 , integrate the key alarm analysis results with the data prediction model, adjust the penalty coefficient and kernel parameters, and predict the health status of the equipment through cross-validation. The specific steps to obtain the equipment health monitoring results are as follows: S601: Integrate the key alarm analysis results with the data prediction model, adjust the penalty coefficient and kernel parameters in the model, and optimize the response speed of the prediction by increasing the weight of the data and changing the sensitivity of the kernel. The execution process of generating the model parameter adjustment data is as follows; In the process of integrating key alarm analysis results with the data prediction model, adjusting the penalty coefficient and kernel parameters in the model is a crucial step. By increasing the weight of the data and changing the sensitivity of the kernel to optimize the response speed of the prediction, adjusting the parameters can make the model more accurately reflect small changes and abnormal conditions in the data, improve the model's prediction accuracy for future changes in equipment operating status, adjust the new penalty coefficient and kernel parameter values, reflect the specific impact of the adjustment on the response speed and accuracy of the prediction model, provide an important basis for further optimization and application of the model, and generate model parameter adjustment data.

[0050] S602: adjust the data through model parameters, apply the adjusted parameter settings, perform cross-validation to evaluate the stability of the model, use the segmented data set to test the model multiple times, analyze the consistency of each prediction result with the real-time data, and generate the cross-validation results. The execution process is as follows; By adjusting the data through model parameters, cross-validation is performed to evaluate the stability of the model. According to the formula , calculate the cross validation results. In the formula, Represents the number of parts the data set is divided into. Representative in the The model used in the test, Represents the prediction accuracy of the model. Detailed explanation of the formula and the process of formula calculation: The model is tested 5 times, and the prediction accuracy of each test is 0.8, 0.85, 0.82, 0.78, and 0.86 respectively. The average accuracy of cross validation is The calculation is as follows: The calculation process shows how to evaluate the stability and reliability of the model through multiple independent tests, ensuring that the model can show high accuracy and consistency on different data sets and improve the generalization ability of the model.

[0051] S603: Based on the cross-validation results, the optimized data prediction model is used to predict the health status of the equipment in the next week, evaluate the probability and timing of occurrence, evaluate the health of the hydropower station equipment, and generate the execution process of the equipment health monitoring results as follows; Based on the cross-validation results, the optimized data prediction model is used to predict the health status of the equipment in the next week. The process includes evaluating the probability and timing of occurrence. The optimized model can more accurately predict the various health problems faced by hydropower station equipment, and intervene and maintain them in advance. It not only lists in detail the various states of the equipment in the next week, but also evaluates the specific probability and expected timing of the state, providing scientific prediction and decision-making basis for equipment management, and generating equipment health monitoring results.

[0052] See also Figure 8A data prediction model segmentation system based on the status of hydropower station equipment, the data prediction model segmentation system based on the status of hydropower station equipment is used to execute the above-mentioned data prediction model segmentation method based on the status of hydropower station equipment, the system includes: The data processing module collects the operating data of the hydropower station equipment, sorts them by time series, removes noise and outliers, and generates a processed time series data set; The feature analysis module calculates the running mean and standard deviation based on the processed time series data set, identifies trends and periodic changes, and obtains statistical feature data; The transition probability calculation module inputs the statistical feature data into the hidden Markov model, sets the number of hidden states, calculates the state transition probability, including normal operation, warning and fault stages, and obtains the state transition analysis results; The state prediction module calculates the real-time probability of the equipment state at the target time point based on the state transition analysis results, predicts the equipment health status in the next week, and obtains the real-time state probability calculation results; The alarm stage evaluation module extracts the key alarm stages from the real-time status probability calculation results, evaluates the impact on equipment health, integrates the key alarm analysis results with the data prediction model, adjusts the penalty coefficient and kernel parameters, predicts the health status of the equipment through cross-validation, and obtains the equipment health monitoring results.

[0053] 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. A data prediction model segmentation method based on the status of hydropower station equipment, characterized in that: The following steps are involved: Collect the operation data of the hydropower station equipment, sort the operation data according to the time series, and remove the noise and outliers of the data to obtain the processed time series data set; Based on the processed time series data set, the operating average and standard deviation of the hydropower station equipment are calculated, and the trend and periodic changes in the data are identified through statistical indicators to obtain statistical characteristic data; Inputting the statistical characteristic data into a hidden Markov model, setting the number of hidden states of the model, calculating the state transition probability, the state including normal operation, early warning and fault stages, and obtaining the state transition analysis result; Based on the state transition analysis results, the real-time probability of the equipment state at the target time point is calculated, and the health status of the hydropower station equipment in the next week is predicted by using the data prediction model to obtain the real-time state probability calculation result; Extracting the key alarm stage from the real-time state probability calculation result, evaluating the severity of the impact of the alarm stage on the health of the equipment, providing maintenance plans and measures, and obtaining key alarm analysis results; The key alarm analysis results are integrated with the data prediction model, the penalty coefficient and kernel parameters are adjusted, and the health status of the equipment is predicted through cross-validation to obtain the equipment health monitoring results.

2. The data prediction model segmentation method based on the hydropower station equipment status according to claim 1 is characterized in that: The processed time series data set includes stability data points arranged in chronological order, filtered continuous monitoring data and data sequences marked with timestamps. The statistical feature data includes the moving average of the time series, the sliding window standard deviation and the identified data growth trend and seasonal pattern. The state transition analysis results include the probability distribution diagram between differentiated states, the average duration of differentiated states and the frequency table of state transitions. The real-time state probability calculation results include the predicted probability value of each state, the time distribution of state probability and the probability diagram of state changes within the prediction period. The key alarm analysis results include the time point of the key alarm, the time period of the alarm and the evaluation results of the impact of the alarm on the overall operation of the equipment. The equipment health monitoring results include the predicted health status trend, the confidence interval of the health status and the equipment performance changes updated based on the monitoring data.

3. The data prediction model segmentation method based on the hydropower station equipment status according to claim 1 is characterized in that: The steps to collect the operation data of the hydropower station equipment, sort the operation data according to the time series, and remove the noise and outliers of the data to obtain the processed time series data set are as follows: Collect the operating data of the hydropower station equipment, synchronize the equipment operating data to check the integrity of the information, upload and store the data in the data buffer, and generate the original data set; Based on the original data set, the equipment operation data is sorted using timestamp tags, the data sequence is adjusted to verify the continuity of the timeline, and the timestamp integrity is checked to avoid data missing, so as to generate time series data; According to the time series data, statistical outliers and noise data that deviates from the normal state are identified and removed, data quality is iteratively optimized, and a processed time series data set is generated.

4. The data prediction model segmentation method based on the hydropower station equipment status according to claim 1 is characterized in that: Based on the processed time series data set, the operation average value and standard deviation of the hydropower station equipment are calculated, and the trend and periodic changes in the data are identified through statistical indicators to obtain the statistical characteristic data. Specifically, the steps are as follows: Calculate the average operating data of the hydropower station equipment through the processed time series data set, determine the average operating level at the target time point using accumulation and division operations, and generate an equipment operating average data table; Based on the equipment operation average value data table, calculate the standard deviation of the hydropower station equipment operation data, use the difference and square method to determine the fluctuation range of the data, and generate the standard deviation calculation result; The standard deviation calculation result is used to analyze the trends and periodic changes in the data, and by recording the peaks and troughs in the data, the repetitive patterns and changing trends of the data are identified to generate statistical feature data.

5. The data prediction model segmentation method based on the hydropower station equipment status according to claim 1 is characterized in that: The statistical feature data is input into the hidden Markov model, the number of hidden states of the model is set, and the state transition probability is calculated. The states include normal operation, warning and failure stages. The specific steps of obtaining the state transition analysis result are: Inputting the statistical feature data into a hidden Markov model, setting the number of hidden states in the model according to the equipment operation status data, including normal operation, warning and fault stages, and generating model initialization settings; Based on the model initialization setting, frequency analysis is used to count the number of state transitions in the statistical feature data, and the probability of each state transition is calculated using the statistical results to generate a state transition probability matrix; The state transition probability matrix is ​​used to dynamically simulate the operation state of the equipment using a hidden Markov model, and the state transition trend within the next week is predicted through the model to identify potential fault development paths and generate state transition analysis results.

6. The data prediction model segmentation method based on the hydropower station equipment status according to claim 5 is characterized in that: The formula of the hidden Markov model is as follows: in, is the state transition trend value, Represents at time From the status To status The transition probability, Represents the intermediate state To status The basic transition probability, Represents the state To status The transfer weight, is the number of states.

7. The data prediction model segmentation method based on the hydropower station equipment status according to claim 1 is characterized in that: Based on the state transition analysis results, the real-time probability of the equipment state at the target time point is calculated, and the health status of the hydropower station equipment in the next week is predicted using the data prediction model. The steps for obtaining the real-time state probability calculation results are as follows: According to the state transition analysis result, a target time point is selected, the real-time probability of the differentiated state of the device is calculated, the state transition probability is multiplied by the real-time state probability, the probability of the differentiated state is determined, and real-time state probability estimation data is generated; Based on the real-time status probability estimation data, a data prediction model is used to adopt a time series analysis method to predict the health status of the equipment in the next week, analyze the probability changes of potential fault warnings and normal operation, and generate short-term health prediction results; The short-term health prediction results are used to evaluate and verify the status changes of hydropower station equipment in the next week, provide a scientific basis for maintenance and operation decisions, and generate real-time status probability calculation results.

8. The data prediction model segmentation method based on the hydropower station equipment status according to claim 1 is characterized in that: The steps of extracting the key alarm stage from the real-time state probability calculation result, evaluating the severity of the impact of the alarm stage on the equipment health, providing maintenance plans and measures, and obtaining the key alarm analysis result are specifically as follows: Based on the real-time status probability calculation result, by comparing the probability values ​​of the differentiated alarm stages, the critical alarm stages that affect the health of the equipment are identified, and the critical alarm status data is generated; By using the critical alarm status data, the severity of the critical alarm status is evaluated, the potential impact on the health of the equipment is analyzed with reference to the duration and probability value of the alarm status, the degree of damage is calculated, and an alarm impact assessment result is generated; Through the alarm impact assessment results, targeted maintenance plans and preventive measures are formulated, the maintenance frequency and inspection content are adjusted, the maintenance efficiency of the measures taken in the alarm stage on equipment damage is verified, and key alarm analysis results are generated.

9. The data prediction model segmentation method based on the hydropower station equipment status according to claim 1 is characterized in that: The steps of integrating the key alarm analysis results with the data prediction model, adjusting the penalty coefficient and the kernel parameter, and predicting the health status of the equipment through cross-validation to obtain the equipment health monitoring results are as follows: Integrate the key alarm analysis results with the data prediction model, adjust the penalty coefficient and kernel parameters in the model, optimize the response speed of the prediction by increasing the weight of the data and changing the sensitivity of the kernel, and generate model parameter adjustment data; Adjusting the data through the model parameters, applying the adjusted parameter settings, performing cross-validation to evaluate the stability of the model, using the segmented data set to test the model multiple times, analyzing the consistency of each prediction result with the real-time data, and generating cross-validation results; Based on the cross-validation results, the optimized data prediction model is used to predict the health status of the equipment in the next week, evaluate the probability and timing of occurrence, evaluate the health of the hydropower station equipment, and generate equipment health monitoring results.

10. A data prediction model segmentation system based on the status of hydropower station equipment, characterized in that: According to the data prediction model segmentation method based on the hydropower station equipment status according to any one of claims 1 to 9, the system comprises: The data processing module collects the operating data of the hydropower station equipment, sorts them by time series, removes noise and outliers, and generates a processed time series data set; The feature analysis module calculates the running mean and standard deviation based on the processed time series data set, identifies trends and periodic changes, and obtains statistical feature data; The transition probability calculation module inputs the statistical feature data into the hidden Markov model, sets the number of hidden states, calculates the state transition probability, including normal operation, warning and fault stages, and obtains the state transition analysis result; The state prediction module calculates the real-time probability of the device state at the target time point based on the state transition analysis result, predicts the health status of the device within the next week, and obtains the real-time state probability calculation result; The alarm stage evaluation module extracts the key alarm stage from the real-time status probability calculation results, evaluates the impact on equipment health, integrates the key alarm analysis results with the data prediction model, adjusts the penalty coefficient and kernel parameters, predicts the health status of the equipment through cross-validation, and obtains the equipment health monitoring results.

Images