A hydraulic station status monitoring method and system based on cloud data

Abstract

The present invention discloses a method and system for monitoring the state of a hydraulic station based on cloud data, and relates to the field of hydraulic stations. The method includes collecting the operating data of the hydraulic station; using data mining technology to analyze the mode and trend of the operating state of the hydraulic station in the cloud, and integrating the state prediction model into the data communication process; maintaining the hydraulic station in combination with the real-time monitoring data and the prediction results of the state prediction model; diagnosing the predicted potential faults based on the association rule algorithm to find out the potential causes of the faults; using collaborative filtering technology to learn from historical cases and obtain recommended maintenance measures; the system includes an acquisition encryption module, a prediction module, an early warning module, a diagnosis module and a recommendation module. The present invention diagnoses the potential causes of faults through the association rule algorithm, provides guidance for maintenance, and uses collaborative filtering technology to learn from historical cases to provide targeted maintenance measures recommendations, thereby improving the quality of maintenance decisions.

Description

Technical Field

[0001] The present invention relates to the field of hydraulic stations, and in particular to a method and system for monitoring the status of a hydraulic station based on cloud data. Background Art

[0002] Cloud data refers to data stored in a cloud computing environment, which users can access through the Internet. The advantage of cloud data is that it provides remote storage, backup and recovery functions, while supporting large-scale data processing and analysis, facilitating data sharing and collaboration. The hydraulic station is an important part of the hydraulic system, which provides the power source for the hydraulic system. The hydraulic station usually includes components such as oil tanks, pumps, valves, pipelines and controllers to generate, regulate and transmit the flow and pressure of hydraulic oil to drive hydraulic machinery to perform work.

[0003] Hydraulic station condition monitoring refers to the real-time tracking and detection of the operating status of the hydraulic station to ensure stable operation of the system and prevent possible failures. This usually includes monitoring the pressure, temperature, flow of hydraulic oil and the performance of key components such as pumps and valves. Cloud-based hydraulic station condition monitoring is a modern monitoring method that combines Internet of Things technology and cloud computing platform. Data is collected by installing sensors in the hydraulic station and transmitted to the cloud server via the Internet.

[0004] In the prior art, for example, Chinese patent 202111047763.8 discloses a self-propelled tail hydraulic system status monitoring and fault diagnosis system and method, which includes using a local area network to transmit physical signals to a monitoring and diagnosis computer, and performing visual data monitoring and fault diagnosis on the monitoring and diagnosis computer. However, the above system still has the following deficiencies in specific applications: its intelligence level is insufficient, it cannot provide more accurate fault prediction and intelligent diagnosis, and it is only effective within a local area network and cannot provide more extensive remote access capabilities.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In view of the problems in the related technology, the present invention proposes a hydraulic station status monitoring method and system based on cloud data to overcome the above-mentioned technical problems existing in the existing related technology.

[0007] To this end, the specific technical solution adopted by the present invention is as follows:

[0008] According to one aspect of the present invention, a hydraulic station status monitoring method based on cloud data is provided, and the hydraulic station status monitoring method based on cloud data comprises the following steps:

[0009] S1. Collect the operating data of the hydraulic station, encrypt the operating data, and upload the encrypted operating data to the cloud server through the Internet of Things technology.

[0010] S2. Use data mining technology to analyze the patterns and trends of the hydraulic station's operating status in the cloud, build a status prediction model, and predict potential failures of the hydraulic station. At the same time, integrate the status prediction model into the data communication process.

[0011] S3. Combine the real-time monitoring data and the prediction results of the status prediction model to maintain the hydraulic station, set the warning threshold, and automatically trigger the maintenance process when the warning threshold is reached.

[0012] S4. Diagnose the predicted potential faults based on the association rule algorithm to find out the potential causes of the faults.

[0013] S5. Use collaborative filtering technology to learn from historical cases and obtain recommended maintenance measures.

[0014] Furthermore, using data mining technology to analyze the patterns and trends of the hydraulic station operating status in the cloud includes the following steps:

[0015] Calculate the mean of the hydraulic station's operating data and reflect the average operating status of the hydraulic station;

[0016] Calculate the variance of the hydraulic station's operating data and evaluate the stability of the hydraulic station's operating status;

[0017] Use time series analysis to smooth the data and obtain the long-term trend of hydraulic station operation;

[0018] Monitor how operating data changes over time by drawing control charts.

[0019] Furthermore, constructing a state prediction model and predicting the potential failure of the hydraulic station includes the following steps:

[0020] Features are extracted from the operating data using feature selection technology, and merged with the long-term trend feature data of the hydraulic station operation, and fault records are used as labels to obtain a data set;

[0021] Select a state prediction model, and train and evaluate the state prediction model using a data set;

[0022] The trained state prediction model is applied to real-time data to predict potential failures of the hydraulic station.

[0023] Furthermore, based on the association rule algorithm, the predicted potential faults are diagnosed and the potential causes of the faults are found, including the following steps:

[0024] S41. Extract entities related to potential faults using natural language processing technology to obtain a set of extracted fault-related entities;

[0025] S42, generating a candidate item set from the extracted fault-related entity set, including a single entity and a combination of two entities, and outputting a candidate item set set;

[0026] S43, using an improved association rule mining algorithm to find frequent item sets from the candidate item set set, and obtain a frequent item set set;

[0027] S44, generating association rules from the frequent item set, and obtaining an association rule set;

[0028] S45, defining a distance metric between fault entities, and clustering the association rules using a clustering algorithm to obtain a clustered fault mode set;

[0029] S46. Analyze each failure mode, identify potential failure causes and influencing factors, and match real-time data with association rules and failure modes to obtain prediction results and diagnostic suggestions for potential failures.

[0030] Furthermore, using the improved association rule mining algorithm to find frequent item sets from the candidate item set set includes the following steps:

[0031] S431, minimum support threshold, minimum confidence threshold and minimum lift threshold;

[0032] S432, calculating the lift for each frequent item set, where the lift is the ratio of the probability of the item set appearing as a whole to the probability of each item appearing individually;

[0033] S433, for each candidate 2-item set, calculating the distance measure between the candidate 2-item set and the identified frequent 1-item set;

[0034] S434. Identify frequent itemsets based on the minimum support threshold, and give priority to itemsets with smaller distance measures based on the distances between itemsets.

[0035] Furthermore, generating association rules from frequent item sets and obtaining an association rule set includes the following steps:

[0036] S441. For frequent item sets, calculate the confidence and lift of the association rules;

[0037] S442, using the minimum confidence threshold and the minimum lift threshold to filter the association rules, and retain the rules that meet the conditions;

[0038] When filtering association rules, all association rules are traversed, and for each association rule, if the confidence of the association rule is less than the minimum confidence threshold, the association rule is discarded;

[0039] If the confidence of the association rule meets the requirements, the lift is checked again. If the lift is less than the minimum lift threshold, the association rule is discarded.

[0040] Furthermore, for each candidate 2-item set, when calculating the distance measure between it and the identified frequent 1-item set, the calculation formula is:

[0041]

[0042] In the formula, S 1 represents a frequent 1-item set, S 2 represents a candidate 2-item set, |S 1 ∣ represents the total number of frequent 1-item sets, ∣S 2 ∣ represents the total number of candidate 2-item sets;

[0043] S 1i represents the i-th item set in the frequent 1-item set;

[0044] w j Represents the candidate 2-item set S 2 The weight of the jth item in ;

[0045] S 2j Represents the candidate 2-item set S 2 The jth item in ;

[0046] ∣S 2j ∩S 1i ∣ represents the item set S 2j With item set S 1i The size of the intersection.

[0047] Furthermore, using collaborative filtering technology to learn from historical cases and obtain the best recommended maintenance measures includes the following steps:

[0048] S51. Collect failure cases of hydraulic stations;

[0049] S52, taking the fault cause as the user and the maintenance measure as the project, creating a fault cause and maintenance measure feedback matrix;

[0050] S53, defining a trust degree for each combination of a maintenance measure and a fault cause in the fault cause and maintenance measure feedback matrix, and quantifying the trust degree into a numerical value;

[0051] S54, using a similarity measurement method to calculate the similarity between different fault causes, and selecting a group of neighbor fault causes for each fault cause according to the similarity and the trust value;

[0052] S55: using the maintenance measure feedback of the neighbor's fault cause and the corresponding trust value, output the maintenance measures applicable to the potential fault cause, and generate a recommended list of maintenance measures for the potential fault cause.

[0053] Furthermore, when the trust metric is quantified into a numerical value, the calculation formula is:

[0054]

[0055] In the formula, represents the trust value between the fault cause f and the maintenance measure m at time t;

[0056] Sim f，i represents the similarity between the fault cause f and the operation feature i;

[0057] represents the trust value between operation feature i and maintenance measure m at time t;

[0058] N represents the total number of features considered.

[0059] According to another aspect of the present invention, a hydraulic station status monitoring system based on cloud data is provided, which includes an acquisition encryption module, a prediction module, an early warning module, a diagnosis module and a recommendation module; wherein the acquisition encryption module is connected to the prediction module, the prediction module is connected to the early warning module and the diagnosis module, and the diagnosis module is connected to the recommendation module.

[0060] The collection and encryption module is used to collect the operating data of the hydraulic station, encrypt the operating data, and upload the encrypted operating data to the cloud server through the Internet of Things technology.

[0061] The prediction module is used to analyze the patterns and trends of the hydraulic station's operating status in the cloud using data mining technology, build a status prediction model, and predict potential failures of the hydraulic station, while integrating the status prediction model into the data communication process.

[0062] The early warning module is used to maintain the hydraulic station by combining the real-time monitoring data and the prediction results of the status prediction model, set the early warning threshold, and automatically trigger the maintenance process when the early warning threshold is reached.

[0063] The diagnosis module is used to diagnose the predicted potential faults based on the association rule algorithm and find out the potential causes of the faults.

[0064] The recommendation module is used to learn from historical cases using collaborative filtering technology to obtain recommended maintenance measures.

[0065] The beneficial effects of the present invention are:

[0066] (1) The present invention provides a method and system for monitoring the state of a hydraulic station based on cloud data. By uploading the operating data of the hydraulic station using the Internet of Things technology, real-time monitoring and remote access to the data are achieved, making it convenient to monitor the state of the hydraulic station anytime and anywhere. Data mining technology is used to reveal patterns and trends in the data, helping to understand the complexity of the operating state of the hydraulic station. The constructed state prediction model can accurately predict the potential failures of the hydraulic station, take countermeasures in advance, and reduce unexpected downtime and maintenance costs. The state prediction model is integrated into the data communication process to realize the automation of data analysis and state monitoring, thereby improving monitoring efficiency.

[0067] (2) The present invention uses association rule algorithms to diagnose potential causes of failures, provide guidance for maintenance, and improve repair efficiency. It also provides references for future design improvements and preventive measures by discovering subtle failure modes and causes. It uses collaborative filtering technology to learn from historical cases, provide targeted maintenance measure recommendations, and improve the quality of maintenance decisions.

[0068] (3) The present invention diagnoses predicted potential faults based on association rule algorithms. When finding out the potential causes of faults, defining distance metrics between fault entities can help accurately measure the similarities or differences between different faults. Through distance metrics, personalized maintenance strategies can be customized for each fault mode based on the similarities between faults. Clustering algorithms can group similar fault modes, thereby more accurately identifying different types of faults. Clustered fault modes can help formulate maintenance plans for specific types of faults and improve maintenance efficiency. At the same time, the improved association rule mining algorithm can effectively filter out frequent item sets from a large number of candidate item sets, reducing computational complexity.

[0069] (4) The present invention uses collaborative filtering technology to learn from historical cases. When obtaining recommended maintenance measures, the relationship between the cause of the failure and the maintenance measures is organized, and the assignment of trust values ​​helps to prioritize reliable feedback. By quantifying trust, the influence of different user feedback in the recommendation process can be weighted. Identifying similar causes of failure can group related problems. Similarity indicators help reveal patterns and commonalities between different cases. By using similar cases, effective maintenance measures are recommended when dealing with similar problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0071] Figure 1is a flow chart of a hydraulic station status monitoring method based on cloud data according to an embodiment of the present invention;

[0072] Figure 2 It is a principle block diagram of a hydraulic station status monitoring system based on cloud data according to an embodiment of the present invention.

[0073] In the figure:

[0074] 1. Collection and encryption module; 2. Prediction module; 3. Early warning module; 4. Diagnosis module; 5. Recommendation module. DETAILED DESCRIPTION

[0075] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in the field should be able to understand other possible implementations and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0076] According to an embodiment of the present invention, a method and system for monitoring the status of a hydraulic station based on cloud data are provided.

[0077] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to one embodiment of the present invention, a hydraulic station status monitoring method based on cloud data is provided, and the hydraulic station status monitoring method based on cloud data includes the following steps:

[0078] S1. Collect the operating data of the hydraulic station, perform pre-processing operations such as cleaning and formatting on the collected data, encrypt the operating data, and upload the encrypted operating data to the cloud server through the Internet of Things technology. Specifically, the following steps are included:

[0079] Use sensors to collect operating parameters of the hydraulic station, such as pressure, flow, temperature, etc.

[0080] Cleaning: Remove noise and outliers during the collection process to ensure data accuracy.

[0081] Formatting: Convert raw data into a unified format for easy subsequent processing.

[0082] Standardization: Standardize the data to ensure that data collected by different devices or at different times are comparable.

[0083] Data encryption: In order to protect the security of data during transmission, the collected data is encrypted using common encryption algorithms such as AES, RSA, etc.

[0084] Through IoT technology, such as using wireless communication technologies such as GPRS, 4G, 5G or Wi-Fi, the encrypted data is sent to the cloud server.

[0085] S2. Use data mining technology to analyze the patterns and trends of the hydraulic station's operating status in the cloud, build a status prediction model, and predict potential failures of the hydraulic station. At the same time, integrate the status prediction model into the data communication process to achieve real-time monitoring and early warning.

[0086] In a further embodiment, analyzing the patterns and trends of the hydraulic station operating status in the cloud using data mining technology includes the following steps:

[0087] Calculate the mean of the hydraulic station's operating data and reflect the average operating status of the hydraulic station.

[0088] Calculate the variance of the hydraulic station's operating data and evaluate the stability of the hydraulic station's operating status. The larger the variance, the greater the volatility of the data and the worse the stability.

[0089] Use time series analysis methods (such as moving average or exponential smoothing methods) to smooth the data and obtain the long-term trend of hydraulic station operation;

[0090] Among them, the moving average method is a simple smoothing forecasting technology used to eliminate random fluctuations in time series data to show the development trend of events, including simple moving average (SMA) and weighted moving average (WMA). Exponential smoothing is a weighted average method.

[0091] By drawing control charts, we can monitor the changes of operating data over time. Control charts are tools in quality control that are used to graphically display the time series of process data to monitor process stability. Control charts can be used to monitor whether the operating data is within the control limits in real time, and any abnormalities or trend deviations in the process can be discovered in time, so that timely adjustments or maintenance can be made.

[0092] In a further embodiment, constructing a state prediction model and predicting potential failures of a hydraulic station comprises the following steps:

[0093] According to the feature selection technology, features are extracted from the operating data, and merged with the long-term trend feature data of the hydraulic station operation, and the fault records are used as labels to obtain a data set; among them, feature selection techniques (such as correlation analysis, principal component analysis, and automatic feature selection algorithm) are applied to reduce the number of features. The features most relevant to fault prediction are selected.

[0094] Select a state prediction model (such as decision tree, random forest, support vector machine or neural network), and train and evaluate the state prediction model using the dataset.

[0095] The trained state prediction model is applied to real-time data to predict potential failures of the hydraulic station.

[0096] S3. Combine the real-time monitoring data and the prediction results of the status prediction model to maintain the hydraulic station, set the warning threshold, and automatically trigger the maintenance process when the warning threshold is reached.

[0097] It should be noted that the warning threshold is set according to the failure probability predicted by the state prediction model or the predicted value of certain key performance indicators. The warning threshold is the critical value of one or more indicators. When the actual monitoring data reaches or exceeds this threshold, it is considered that the hydraulic station has a high risk of failure.

[0098] The threshold is determined based on historical data analysis and expert experience. It is directly related to the potential failure results predicted by the model. When real-time data shows that one or some key indicators are close to or exceed the threshold, it indicates that the failure risk predicted by the model is increasing, and the maintenance process will be automatically triggered. This includes sending alarms, shutting down for inspection, replacing parts, or other necessary maintenance operations.

[0099] For example, normal operating range of pressure: 900-1100psi. Warning threshold: 1200psi.

[0100] Normal flow rate range: 40-60L / min. Warning threshold: 70L / min.

[0101] Normal operating range of temperature: 50-70℃. Warning threshold: 80℃.

[0102] The state prediction model analyzed data from the past year and determined the following warning thresholds:

[0103] If the pressure exceeds 1200 psi for 5 consecutive minutes, the probability of failure increases to 15%.

[0104] If the flow rate suddenly increases to more than 70L / min within 10 minutes, the failure probability increases to 20%.

[0105] If the temperature exceeds 80°C for 15 minutes, the probability of failure increases to 25%.

[0106] In real-time monitoring, the following occurs:

[0107] At 9:00 AM, the pressure reading rose to 1215 psi and remained there for 5 minutes.

[0108] At 9:10AM, the flow reading suddenly increased to 75L / min.

[0109] At 9:20 AM, the temperature stabilized at 82°C.

[0110] Based on the warning threshold, the following actions are automatically performed:

[0111] 9:05AM: The system sends a pressure alert because the pressure exceeds the warning threshold, suggesting checking for possible leaks or blockages.

[0112] 9:10AM: The flow rate exceeds the warning threshold and the system prompts to check the pump performance or possible pipe rupture.

[0113] 9:20AM: The temperature exceeds the warning threshold and the system recommends checking the condition of the cooling system or hydraulic oil.

[0114] S4. Diagnose the predicted potential faults based on the association rule algorithm to find out the potential causes of the faults.

[0115] In a further embodiment, diagnosing the predicted potential fault based on the association rule algorithm to find out the potential fault cause includes the following steps:

[0116] S41. Use a Transformer model or other NLP (natural language processing) technology to extract entities related to potential faults (such as component names, fault types, environmental conditions, etc.) to obtain a set of extracted fault-related entities.

[0117] S42. Generate a candidate item set from the extracted fault-related entity set, including a single entity (1-item set) and a combination of two entities (2-item sets), and output the candidate item set set.

[0118] S43. Use the improved association rule mining algorithm to find frequent itemsets from the candidate itemset set and obtain the frequent itemset set, considering support, confidence and lift. The frequent itemsets show obvious patterns, which may be the factors causing the failure.

[0119] S44. Generate association rules from frequent item sets, determine which rules are meaningful for fault prediction, and obtain a set of association rules, that is, derive potential causal relationships from frequent item sets. These rules can help predict and understand the occurrence of faults.

[0120] S45. Define the distance metric between fault entities and use a clustering algorithm (such as DBSCAN) to cluster the association rules to obtain a clustered set of fault patterns. Group similar fault rules together to form a broader fault pattern. Clustered fault patterns can help understand the relationship between different faults and their connection with specific conditions.

[0121] S46, analyze each fault mode, identify potential fault causes and influencing factors, and match real-time data with association rules and fault modes to obtain prediction results and diagnostic suggestions for potential faults. Finally, analyze and determine the cause of the fault, match the real-time data with the determined mode, and perform fault prediction and diagnosis.

[0122] In a further embodiment, using the improved association rule mining algorithm to find frequent itemsets from the candidate itemset set includes the following steps:

[0123] S431, minimum support threshold, minimum confidence threshold and minimum lift threshold; the confidence and lift of association rules are two key indicators used to measure the importance of rules.

[0124] S432. Calculate the lift for each frequent item set, where the lift is the ratio of the probability of the item set appearing as a whole to the probability of each item appearing individually.

[0125] S433. For each candidate 2-item set, calculate a distance metric between the candidate 2-item set and the identified frequent 1-item set, where the distance metric is based on the number of times they co-occur.

[0126] S434. According to the minimum support threshold, frequent item sets are identified. At the same time, based on the distance between item sets, item sets with smaller distance metric are preferentially selected to reduce redundancy.

[0127] In a further embodiment, generating association rules from frequent item sets and obtaining an association rule set includes the following steps:

[0128] S441. For frequent item sets, calculate the confidence and lift of the association rule; confidence represents the probability of the occurrence of the consequent under the condition that the antecedent occurs. Lift is an indicator of the interestingness of the rule, which represents the degree of association between the antecedent and the consequent. A lift greater than 1 indicates a strong association, equal to 1 indicates no association, and less than 1 indicates a negative association.

[0129] S442: Use the minimum confidence threshold and the minimum lift threshold to filter the association rules and retain the rules that meet the conditions.

[0130] When filtering association rules, all association rules are traversed, and for each association rule, if the confidence of the association rule is less than the minimum confidence threshold, the association rule is discarded;

[0131] If the confidence of the association rule meets the requirements, the lift is checked again. If the lift is less than the minimum lift threshold, the association rule is discarded.

[0132] For example, set the minimum confidence threshold (minConf) = 30%, and the minimum lift threshold (minLift) = 2.

[0133] Rule 1: Confidence = 25%, lift = 3.

[0134] Rule 2: Confidence = 35%, lift = 1.5.

[0135] Rule 3: Confidence = 40%, lift = 2.5.

[0136] According to the set thresholds, the filtering process is as follows.

[0137] Rule 1: Confidence 25% < minConf 30%, discard.

[0138] Rule 2: Confidence 35% >= minConf 30%, lift 1.5 < minLift 2, discard.

[0139] Rule 3: Confidence 40% >= minConf 30%, lift 2.5 >= minLift 2, retain.

[0140] Finally, only Rule 3 meets the filtering conditions and will be retained in the final association rule set.

[0141] In a further embodiment, when calculating the distance metric between each candidate 2-itemset and the identified frequent 1-itemset, the calculation formula is:

[0142]

[0143] In the formula, S 1 represents the frequent 1-itemset, S 2 represents the candidate 2-itemset, |S 1 | represents the total number of frequent 1-itemsets, |S 2 | represents the total number of candidate 2-itemsets;

[0144] S 1i represents the i-th itemset in the frequent 1-itemset;

[0145] w j represents the weight of the j-th item in the candidate 2-itemset S 2 , which is adjusted according to the importance of the item;

[0146] S 2j represents the j-th item in the candidate 2-itemset S 2 ;

[0147] |S 2j ∩S1i ∣ represents the item set S 2j With item set S 1i The size of the intersection.

[0148] This formula calculates the average of the weighted overlap measures between the candidate 2-item set and each frequent 1-item set, and subtracts this value from one to get the distance measure. The closer the value is to one, the greater the difference between the candidate 2-item set and the frequent 1-item set, that is, the greater the distance. Weights are given different importance to different items and are set according to actual business needs or domain knowledge.

[0149] The formula calculates a distance metric for each candidate 2-item set, and then uses this value to determine whether the candidate 2-item set is different enough from any of the frequent 1-item sets to decide whether to retain it as a non-redundant itemset.

[0150] In association rule mining, S 1 It is a set of identified frequent itemsets, each of which contains only a single item. 1-itemset is the basis for building more complex itemsets (such as 2-itemset). Candidate 2-itemset is the set of all possible 2-itemsets composed of items in 1-itemset. In order to reduce redundancy and improve efficiency, the distance measure between candidate 2-itemset and frequent 1-itemset is calculated. If a candidate 2-itemset has a high overlap with items in frequent 1-itemset (i.e., they contain many of the same items), it is considered redundant and will not be considered for generating association rules. In this way, the efficiency of association rule mining is improved and the quality of generated association rules is improved.

[0151] S5. Use collaborative filtering technology to learn from historical cases and obtain recommended maintenance measures.

[0152] In a further embodiment, using collaborative filtering technology to learn from historical cases and obtain recommended optimal maintenance measures includes the following steps:

[0153] S51. Collect failure cases of hydraulic stations, including the causes of failures and maintenance measures taken, and clean the data to remove incomplete or inaccurate records.

[0154] S52. Taking the fault cause as the user and the maintenance measure as the project, a fault cause and maintenance measure feedback matrix is ​​created to record the maintenance measures and their effects corresponding to each fault cause; in the matrix, rows represent different fault causes and columns represent different maintenance measures. Each element in the matrix represents the effect evaluation of a maintenance measure corresponding to a specific fault cause, which can be a binary value, a score or other forms of feedback.

[0155] S53. Define a degree of trust for each combination of maintenance measures and fault causes in the fault cause and maintenance measure feedback matrix. This can be based on the frequency of historical successful fault resolution, the long-term effects of maintenance measures, professional evaluations of engineers, etc., and quantify the trust into a numerical value.

[0156] S54. Use a similarity measurement method to calculate the similarities between different fault causes, and select a group of neighbor fault causes for each fault cause based on the similarities and the trust values, where the neighbor fault causes are the cases most relevant to the current problem.

[0157] S55. Using the maintenance measure feedback and corresponding trust values ​​of neighbor fault causes, output maintenance measures applicable to potential fault causes, and generate a recommended list of maintenance measures for potential fault causes. The recommended list is generated based on successful maintenance measures and trust values ​​of similar cases in historical data, providing maintenance personnel with a data-driven list of measures that may solve the problem.

[0158] In a further embodiment, when the trust metric is quantified into a numerical value, the calculation formula is:

[0159]

[0160] In the formula, represents the trust value between the fault cause f and the maintenance measure m at time t;

[0161] Sim f，i represents the similarity between the fault cause f and the operation feature i;

[0162] represents the trust value between operation feature i and maintenance measure m at time t;

[0163] N represents the total number of features considered.

[0164] It should be noted that operational features refer to various specific attributes or parameters related to fault causes and maintenance measures. These features are used to describe the characteristics of fault conditions and maintenance measures, so that the algorithm can calculate the similarity between fault causes and the applicability of maintenance measures based on these features, including the frequency of faults, the severity of faults, etc. In the collaborative filtering algorithm, these similarity calculation results will be used in the recommendation system to predict and recommend maintenance measures suitable for the current fault condition.

[0165] By analyzing the similarity between the fault cause and each feature, as well as the trust relationship between these features and specific maintenance measures, the trust of taking a maintenance measure for the current fault cause is comprehensively calculated. This trust is used as part of the recommendation system to help the maintenance team decide which maintenance measure to take.

[0166] In order to facilitate understanding of the above technical solutions of the present invention, the working principle or operation mode of the present invention in the actual process is described in detail below.

[0167] The following operating parameters are collected using sensors in the hydraulic station:

[0168] Pressure: 1000psi.

[0169] Flow rate: 50L / min.

[0170] Temperature: 60℃.

[0171] Remove noise, such as a sudden jump to -5°C in the temperature reading. Convert all readings to SI units. Ensure data comparability across time periods, such as uniformly expressing pressure in psi. Process data using the AES encryption algorithm. Send encrypted data to a cloud server via 4G network.

[0172] Calculate the mean and variance.

[0173] Average pressure: 1000psi.

[0174] Average flow rate: 50L / min.

[0175] Average temperature: 60℃.

[0176] Pressure variance: 15psi 2 .

[0177] Flow rate variance: 2(L / min)2.

[0178] Temperature variance: 3℃2.

[0179] Time series analysis: The simple moving average method was used to calculate the average temperature trend over the past week. A temperature control chart was drawn and it was found that the temperature was stable between 60±5℃.

[0180] Extract key features from the data, such as the pressure rise rate. Select the random forest model for state prediction. Use historical data to train the model with an accuracy of 95%.

[0181] When mining association rules, item set A: high temperature, item set B: high pressure, item set C: high flow.

[0182] Frequent itemsets: {high temperature, high pressure}.

[0183] Association rule: If both temperature and pressure are high at the same time, there is a possibility of failure.

[0184] Cluster analysis: It was found that high temperature and high pressure were often associated with filter clogging.

[0185] Collect failure cases: Filter blockage causes temperature and pressure to increase.

[0186] Feedback matrix: records the causes of filter blockage and the effectiveness of maintenance measures (such as filter replacement).

[0187] Confidence calculation: Based on past successful failure resolution cases, a confidence level of 0.9 is defined for the action of replacing the filter.

[0188] Similarity calculation: Calculate the similarity between the current fault and historical cases.

[0189] Maintenance action recommendation: Based on similarity and confidence, filter replacement is recommended as a maintenance action.

[0190] like Figure 2 As shown, according to another embodiment of the present invention, a hydraulic station status monitoring system based on cloud data is provided, and the hydraulic station status monitoring system based on cloud data includes an acquisition encryption module 1, a prediction module 2, an early warning module 3, a diagnosis module 4 and a recommendation module 5; wherein the acquisition encryption module 1 is connected to the prediction module 2, the prediction module 2 is connected to the early warning module 3 and the diagnosis module 4, and the diagnosis module 4 is connected to the recommendation module 5.

[0191] The collection and encryption module 1 is used to collect the operating data of the hydraulic station, encrypt the operating data, and upload the encrypted operating data to the cloud server through the Internet of Things technology.

[0192] Prediction module 2 is used to analyze the patterns and trends of the operating status of the hydraulic station in the cloud using data mining technology, and to build a status prediction model and predict potential failures of the hydraulic station, while integrating the status prediction model into the data communication process.

[0193] The early warning module 3 is used to maintain the hydraulic station by combining the real-time monitoring data and the prediction results of the state prediction model, and to set the early warning threshold, and automatically trigger the maintenance process when the early warning threshold is reached.

[0194] The diagnosis module 4 is used to diagnose the predicted potential faults based on the association rule algorithm and find out the potential causes of the faults.

[0195] The recommendation module 5 is used to learn from historical cases using collaborative filtering technology to obtain recommended maintenance measures.

[0196] In summary, the present invention provides a method and system for monitoring the state of a hydraulic station based on cloud data. By uploading the operating data of the hydraulic station using the Internet of Things technology, real-time monitoring and remote access of the data are realized, and the state monitoring of the hydraulic station is convenient anytime and anywhere. The patterns and trends in the data are revealed by data mining technology to help understand the complexity of the operating state of the hydraulic station. The constructed state prediction model can accurately predict the potential faults of the hydraulic station, take countermeasures in advance, and reduce unexpected downtime and maintenance costs. The state prediction model is integrated into the data communication process to realize the automation of data analysis and state monitoring, and improve the monitoring efficiency. The present invention diagnoses the potential causes of faults through an association rule algorithm, provides guidance for maintenance, and improves the repair efficiency. And by discovering the failure modes and causes that are not easy to detect, it provides a reference for future design improvements and preventive measures. Collaborative filtering technology is used to learn from historical cases, provide targeted maintenance measures recommendations, and improve the quality of maintenance decisions. The present invention diagnoses the predicted potential faults based on the association rule algorithm. When finding the potential causes of the faults, defining the distance metric between the fault entities can help accurately measure the similarities or differences between different faults, and through the distance metric, a personalized maintenance strategy can be customized for each fault mode according to the similarity between the faults. Clustering algorithms can group similar failure modes, thereby more accurately identifying different types of failures. Clustered failure modes can help develop maintenance plans for specific types of failures and improve maintenance efficiency. At the same time, the improved association rule mining algorithm can effectively filter out frequent item sets from a large number of candidate item sets, reducing computational complexity. The present invention uses collaborative filtering technology to learn from historical cases, organize the relationship between failure causes and maintenance measures when obtaining recommended maintenance measures, and assigning trust values ​​helps to prioritize reliable feedback. By quantifying trust, the impact of different user feedback in the recommendation process can be weighted. Identifying similar failure causes can group related problems, and similarity indicators help reveal patterns and commonalities between different cases. By using similar cases, effective maintenance measures are recommended when dealing with similar problems.

[0197] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A hydraulic station status monitoring method based on cloud data, characterized in that: The hydraulic station condition monitoring method based on cloud data includes the following steps: S1. Collect the operating data of the hydraulic station, encrypt the operating data, and upload the encrypted operating data to the cloud server through the Internet of Things technology; S2. Use data mining technology to analyze the patterns and trends of the hydraulic station's operating status in the cloud, build a status prediction model, and predict potential failures of the hydraulic station. At the same time, integrate the status prediction model into the data communication process; S3. Combine the real-time monitoring data and the prediction results of the state prediction model to maintain the hydraulic station, set the warning threshold, and automatically trigger the maintenance process when the warning threshold is reached; S4. Diagnose the predicted potential faults based on the association rule algorithm to find out the potential causes of the faults; The method of diagnosing the predicted potential faults based on the association rule algorithm and finding the potential fault causes includes the following steps: S41. Extract entities related to potential faults using natural language processing technology to obtain a set of extracted fault-related entities; S42, generating a candidate item set from the extracted fault-related entity set, including a single entity and a combination of two entities, and outputting a candidate item set set; S43, using an improved association rule mining algorithm to find frequent item sets from the candidate item set set, and obtain a frequent item set set; S44, generating association rules from the frequent item set, and obtaining an association rule set; S45, defining a distance metric between fault entities, and clustering the association rules using a clustering algorithm to obtain a clustered fault mode set; S46. Analyze each failure mode, identify potential failure causes and influencing factors, and match real-time data with association rules and failure modes to obtain prediction results and diagnostic suggestions for potential failures; The method of finding frequent itemsets from a candidate itemset set using an improved association rule mining algorithm comprises the following steps: S431, minimum support threshold, minimum confidence threshold and minimum lift threshold; S432, calculating the lift for each frequent item set, where the lift is the ratio of the probability of the item set appearing as a whole to the probability of each item appearing individually; S433, for each candidate 2-item set, calculating the distance measure between the candidate 2-item set and the identified frequent 1-item set; S434, identifying frequent itemsets according to the minimum support threshold, and giving priority to itemsets with smaller distance measures based on the distances between the itemsets; For each candidate 2-item set, when calculating the distance measure between it and the identified frequent 1-item set, the calculation formula is: ； In the formula, S 1 indicates a frequent 1-item set, S 2 represents candidate 2-item sets, |S1| represents the total number of frequent 1-item sets, and |S2| represents the total number of candidate 2-item sets; S 1i Indicates the first frequent item in a set i Itemsets; w j Represents a candidate 2-item set S 2 in 1 j The weight of each item; S 2j Represents a candidate 2-item set S 2 of 2 j Items; ∣ S 2j ∩ S 1i ∣ represents item set S 2j With item sets S 1i The size of the intersection of S5. Use collaborative filtering technology to learn from historical cases and obtain recommended maintenance measures; The method of using collaborative filtering technology to learn from historical cases and obtain the best recommended maintenance measures includes the following steps: S51. Collect failure cases of hydraulic stations; S52, taking the fault cause as the user and the maintenance measure as the project, creating a fault cause and maintenance measure feedback matrix; S53, defining a trust degree for each combination of a maintenance measure and a fault cause in the fault cause and maintenance measure feedback matrix, and quantifying the trust degree into a numerical value; S54, using a similarity measurement method to calculate the similarity between different fault causes, and selecting a group of neighbor fault causes for each fault cause according to the similarity and the trust value; S55: using the maintenance measure feedback of the neighbor's fault cause and the corresponding trust value, output the maintenance measures applicable to the potential fault cause, and generate a recommended list of maintenance measures for the potential fault cause.

2. A hydraulic station status monitoring method based on cloud data according to claim 1, characterized in that: The method of analyzing the mode and trend of the hydraulic station operation status in the cloud using data mining technology includes the following steps: Calculate the mean of the hydraulic station's operating data and reflect the average operating status of the hydraulic station; Calculate the variance of the hydraulic station's operating data and evaluate the stability of the hydraulic station's operating status; Use time series analysis to smooth the data and obtain the long-term trend of hydraulic station operation; Monitor how operating data changes over time by drawing control charts.

3. The method for monitoring the hydraulic station status based on cloud data according to claim 2 is characterized in that: The construction of the state prediction model and prediction of potential failures of the hydraulic station include the following steps: Features are extracted from the operating data using feature selection technology, and merged with the long-term trend feature data of the hydraulic station operation, and fault records are used as labels to obtain a data set; Select a state prediction model, and train and evaluate the state prediction model using a data set; The trained state prediction model is applied to real-time data to predict potential failures of the hydraulic station.

4. The method for monitoring the hydraulic station status based on cloud data according to claim 1, characterized in that: The step of generating association rules from frequent item sets and obtaining an association rule set comprises the following steps: S441. For frequent item sets, calculate the confidence and lift of the association rules; S442, using the minimum confidence threshold and the minimum lift threshold to filter the association rules, and retain the rules that meet the conditions; When filtering association rules, all association rules are traversed, and for each association rule, if the confidence of the association rule is less than the minimum confidence threshold, the association rule is discarded; If the confidence of the association rule meets the requirements, the lift is checked again. If the lift is less than the minimum lift threshold, the association rule is discarded.

5. The method for monitoring the hydraulic station status based on cloud data according to claim 1, characterized in that: When the trust metric is converted into a numerical value, the calculation formula is: ； In the formula, Indicates at time t Next, the cause of the failure f And maintenance measures m The trust value between Sim f，i Indicates the cause of the fault f and operating characteristics i The similarity between Indicates at time t Next, the operating characteristics i And maintenance measures m The trust value between N Denotes the total number of features considered.

6. A hydraulic station status monitoring system based on cloud data, used to implement the hydraulic station status monitoring method based on cloud data according to any one of claims 1 to 5, characterized in that: The hydraulic station condition monitoring system based on cloud data includes an acquisition encryption module, a prediction module, an early warning module, a diagnosis module and a recommendation module; Wherein, the collection and encryption module is connected to the prediction module, the prediction module is connected to the early warning module and the diagnosis module, and the diagnosis module is connected to the recommendation module; The collection and encryption module is used to collect the operating data of the hydraulic station, encrypt the operating data, and upload the encrypted operating data to the cloud server through the Internet of Things technology; The prediction module is used to analyze the mode and trend of the hydraulic station operation status in the cloud using data mining technology, and to build a status prediction model and predict potential failures of the hydraulic station, while integrating the status prediction model into the data communication process; The early warning module is used to maintain the hydraulic station by combining the real-time monitoring data and the prediction results of the state prediction model, and to set the early warning threshold, and automatically trigger the maintenance process when the early warning threshold is reached; The diagnostic module is used to diagnose the predicted potential faults based on the association rule algorithm to find out the potential causes of the faults; The recommendation module is used to learn from historical cases using collaborative filtering technology to obtain recommended maintenance measures.

Images