Guide vane wear prediction method and system based on deep learning

Abstract

The invention discloses a guide vane wear prediction method and system based on deep learning, and belongs to the technical field of mechanical fault prediction and maintenance, and the method comprises the steps: collecting real-time sensor data of guide vanes, and integrating historical wear data of similar guide vanes; identifying similar features of similar guide vanes, extracting key wear modes through the similar features, and establishing corresponding prediction models for different guide vanes; and based on the extracted features and the prediction model, analyzing the wear condition of the guide vane, and providing a wear prediction result. Through comprehensive monitoring and data analysis, the accuracy of guide vane wear prediction is improved, and the reliability of equipment operation is ensured. Through a real-time alarm system, dynamic management of different wear degrees is realized, maintenance work is effectively guided, fault risks are reduced, the service life of equipment is prolonged, and the operation and maintenance efficiency is improved. According to the invention, the abrasion problem can be prevented in advance, and the normal operation and safety of the water turbine are guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical fault prediction and maintenance, and in particular to a guide vane wear prediction method and system based on deep learning. Background Art

[0002] In today's industrial field, guide vanes are key components of fluid machinery such as turbines, and their wear has always been an important factor affecting the operating efficiency and life of equipment. With the continuous development of technology, the prediction methods for guide vane wear are also constantly evolving. Traditional wear prediction methods mainly rely on empirical formulas and physical models. These methods can predict the wear of guide vanes to a certain extent, but often have problems such as low prediction accuracy and poor adaptability.

[0003] In recent years, deep learning technology has achieved remarkable results in many fields, and its advantages in data processing and analysis have provided new ideas for the prediction of guide vane wear. However, the existing deep learning-based guide vane wear prediction technology still has the following shortcomings: First, the existing technology often ignores the in-depth discussion of different wear types such as uneven wear, corrosive wear, fatigue wear, and collision wear, resulting in inaccurate prediction results; second, the existing technology lacks effective processing methods in the feature recognition and key wear pattern extraction of similar guide vanes, which limits the application scope of the prediction model; finally, the existing technology has not yet formed a complete system for real-time monitoring and early warning of wear prediction results, which makes it difficult to meet the needs of guide vane wear prediction in actual production.

[0004] In view of the shortcomings of the prior art, the present invention proposes a guide vane wear prediction method based on deep learning, which has significant advantages in improving the accuracy of guide vane wear prediction, broadening the application scope of the prediction model, and optimizing the real-time monitoring and early warning system. Summary of the invention

[0005] In view of the above existing problems, the present invention uses real-time sensor data and historical wear data to identify the wear pattern of the guide vanes through a deep learning algorithm, and establishes a prediction model for guide vanes with different characteristics to accurately predict their wear conditions, solving the technical problem that it is difficult to accurately identify and predict the wear of the guide vanes in traditional methods. Its purpose is to discover potential wear problems in advance, provide scientific maintenance plans for equipment such as turbines, reduce failure risks and maintenance costs, and at the same time extend the service life of the equipment and improve operating efficiency and safety.

[0006] In order to solve the above technical problems, a guide vane wear prediction method based on deep learning is proposed, including:

[0007] Collect real-time sensor data of guide vanes and integrate historical wear data of similar guide vanes; identify similar features of similar guide vanes, extract key wear patterns through similar features, and establish corresponding prediction models for different guide vanes; analyze guide vane wear based on the extracted features and prediction models, and provide wear prediction results.

[0008] As a preferred solution of the guide vane wear prediction method based on deep learning described in the present invention, the real-time sensor data includes guide vane displacement, vibration frequency, temperature change, pressure data, flow rate data, operating time and fault log;

[0009] The historical wear data of similar integrated guide vanes is used as a reference for real-time data preprocessing, all real-time Min-Max scaling is standardized, data exceeding the specified range is eliminated, and weights are assigned according to the actual similarity between the real-time data of the guide vanes in use and the historical wear data of the corresponding guide vanes, ensuring that the sum of the similarity weights is greater than 0.5.

[0010] As a preferred solution of the guide vane wear prediction method based on deep learning described in the present invention, the identifying similar features of similar guide vanes includes specifying similar requirements for identifying similar guide vanes, processing and learning the geometry, material and operating environment characteristics of the guide vanes through a deep learning algorithm, and specifying the similar requirements as:

[0011] Requirement 1: Geometric parameter characteristics of the guide vane. When the difference in the geometric shape of the guide vane, including length, width, thickness, and angle, is within 10% and the geometric parameter similarity reaches 0.85, it is confirmed as a similar type;

[0012] Requirement 2: The material property characteristics of the guide vanes. When the material characteristic similarity exceeds 0.9 and the material composition of similar guide vanes including hardness, wear resistance, and density does not differ by more than 5%, they are confirmed as similar types;

[0013] Requirement 3: The operating environment characteristics of the guide vanes. When the temperature and humidity changes of the guide vanes do not exceed 15% and the environmental similarity index exceeds 0.8, it is confirmed as a similar type;

[0014] When the characteristic requirements between the guide vanes meet any two requirements from requirement one to requirement three, they are confirmed as similar guide vanes.

[0015] As a preferred solution of the guide vane wear prediction method based on deep learning described in the present invention, wherein: the similar features include, after similar guide vanes are determined, identifying similar features of similar guide vanes according to similar requirements and guide vane properties, and extracting key wear patterns by calculating the distance and similarity between similar features;

[0016] The similar features include extracting guide vane features that meet similar requirements, defining a distance metric, and using an improved Manhattan distance D new (f i ,f j ), convert the distance into similarity S similarity :

[0017]

[0018] Among them, X is a constant, representing the constant of the maximum feature difference;

[0019] After obtaining the similarity between the guide vanes, the guide vanes are mapped to extract the key wear patterns.

[0020] As a preferred solution of the guide vane wear prediction method based on deep learning described in the present invention, the key wear mode includes: performing wear mapping on the guide vane, identifying the key wear mode R by wear mapping, and judging that the key wear mode includes uneven wear mode, corrosive wear mode, fatigue wear mode, and collision wear mode:

[0021] When ΔR>θ1 and repeated wear peaks are observed within the set range, indicating that local wear is aggravated, it is determined to be an uneven wear mode;

[0022] When the R value shows a long-term uniform decline and the decline exceeds θ2, the data relative chemical property analysis is performed. When the wear rate is not within the standard deviation of the uniform value of the change, it is determined to be a corrosive wear mode;

[0023] When the R value shows a change of more than θ3 in x consecutive measurements and the peak value of the change corresponds to the fluctuation of the load data, a secondary judgment is made at this time. When the internal stress analysis results show that the formation rate of microcracks increases by more than θ4 in the time series, it is judged as fatigue wear;

[0024] When any of the R value readings increases suddenly and the corresponding vibration sensor displays an abnormal peak value, it is determined to be a collision wear mode;

[0025] Among them, θ1, θ2, θ3, and θ4 are different thresholds set for different turbine environments and operating conditions, and ΔR is a time-based change value.

[0026] As a preferred solution of the guide vane wear prediction method based on deep learning described in the present invention, the establishment of the corresponding prediction model includes constructing a wear prediction function, and the output is the expected wear degree, which represents the expected future wear condition:

[0027]

[0028] Among them, δ is the sensitivity of the adjustment model, T represents the real-time ambient temperature, and T 0 is the reference standard temperature;

[0029] Compare the predicted values ​​with historical wear data to determine the upward trend of wear and analyze the causes of wear.

[0030] As a preferred solution of the guide vane wear prediction method based on deep learning described in the present invention, wherein: the analysis of the guide vane wear condition includes that when the future predicted wear value exceeds the wear rate of 10%, the wear is serious, it is marked as an emergency alarm, and an inspection is carried out immediately;

[0031] When the predicted wear value is between 5% and 10%, the wear degree is moderate and marked as a moderate alarm. Set weekly monitoring and calculate the change rate ΔF of the predicted wear value in real time. When ΔF>1% / week, the moderate alarm is upgraded to an emergency alarm. When ΔF<1% / week, the original alarm mode remains unchanged.

[0032] When the predicted wear value is less than 5%, it is considered that the wear degree is within the normal loss range and does not affect the normal operation of the turbine. It is allowed to be ignored and the current state is maintained for real-time detection.

[0033] Another object of the present invention is to provide a guide vane wear prediction system based on deep learning. The present invention solves the timeliness and accuracy problems in traditional wear monitoring methods. It predicts the guide vane wear in advance, reduces unplanned downtime, optimizes maintenance resources and costs, improves data utilization, automatically identifies similar guide vanes and establishes personalized prediction models, realizes real-time monitoring and alarms, and has adaptability and scalability, thereby significantly improving the operating efficiency and safety of industrial equipment, ensuring that the equipment is maintained in time before wear causes failure, avoiding sudden failures, and improving the overall operation and maintenance management level.

[0034] As a preferred solution of the guide vane wear prediction system based on deep learning described in the present invention, it is characterized by comprising a data processing module, a guide vane judgment module, a prediction module and a wear analysis module;

[0035] The data processing module collects and integrates real-time sensor data and historical wear data of similar guide vanes, performs standardization processing on the real-time data, assigns weights to the integrated data to ensure that the similarity weight is greater than 0.5, and transmits the assigned data to the guide vane judgment module;

[0036] The guide vane judgment module receives the standardized data from the data processing module, identifies and confirms similar guide vanes, extracts similar features of similar guide vanes based on geometric parameters, material properties and operating environment characteristics, calculates the distance and similarity between the features, maps and extracts key wear patterns, and passes the results to the prediction module;

[0037] The prediction module receives the key wear pattern from the guide vane judgment module, constructs a wear prediction function, performs wear prediction based on the deep learning model, outputs a predicted value of future wear degree, compares the predicted value with historical data, analyzes the wear trend, and passes the prediction result and wear trend analysis to the wear analysis module;

[0038] The wear analysis module receives the wear prediction result of the prediction module, analyzes the predicted wear condition, determines the severity of the wear, sets the alarm level according to the severity of the wear, monitors the change rate of the wear prediction value, and adjusts the alarm level.

[0039] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of a guide vane wear prediction method based on deep learning are implemented.

[0040] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of a guide vane wear prediction method based on deep learning are implemented.

[0041] Beneficial effects of the present invention: The present invention realizes comprehensive monitoring of the wear status of the guide vanes and full utilization of historical data, provides rich basic data support for subsequent wear prediction, and improves the accuracy and reliability of the prediction. By standardizing and weighting real-time data, the consistency and comparability of the data are guaranteed, the influence of data noise is reduced, and the generalization ability of the prediction model is improved. Identifying the characteristics of similar guide vanes and stipulating similar requirements, accurately classifying the properties of the guide vanes, and helping to screen out guide vanes with similar wear patterns, providing a basis for establishing a targeted prediction model. Extracting the key wear patterns of similar guide vanes and deeply exploring the wear characteristics, it is helpful to identify different wear types and provide more targeted training samples for the prediction model. By mapping the wear of the guide vanes and identifying the key wear patterns, the wear type is accurately determined, potential problems are discovered in advance, and support is provided for equipment maintenance and troubleshooting. Constructing a wear prediction function and outputting the expected wear degree, the prediction of future wear conditions is realized, which helps to formulate maintenance plans, reduce the risk of failures, and extend the service life of equipment. Finally, by analyzing the wear conditions and setting different alarm levels, real-time monitoring and early warning are achieved, guiding maintenance personnel to take corresponding measures to ensure the safe and stable operation of the equipment, and adjusting the alarm level according to the change rate of the wear prediction value, thus achieving dynamic management and improving operation and maintenance efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work, among which:

[0043] Figure 1 An overall flow chart of a guide vane wear prediction method based on deep learning is provided for one embodiment of the present invention.

[0044] Figure 2 A system solution module diagram of a guide vane wear prediction system based on deep learning provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an embodiment that is mutually exclusive with other embodiments, either individually or selectively.

[0048] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0049] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions 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 direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0050] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0051] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a guide vane wear prediction method based on deep learning, comprising:

[0052] S1: Collect real-time sensor data of guide vanes and integrate historical wear data of similar guide vanes.

[0053] Specifically, real-time sensor data includes guide vane displacement, vibration frequency, temperature changes, pressure data, flow rate data, operating time, and fault logs;

[0054] The historical wear data of similar guide vanes are integrated as a reference for real-time data preprocessing, and all real-time data are standardized by Min-Max scaling. Data that exceeds the specified range are eliminated, and weights are assigned according to the actual similarity between the real-time data of the guide vanes in use and the historical wear data of the corresponding guide vanes to ensure that the sum of the similarity weights is greater than 0.5.

[0055] Real-time monitoring of the following:

[0056] When the real-time guide vane displacement changes exceed the set threshold, an alarm is automatically issued and additional analysis data is requested from similar guide vane historical data;

[0057] When the vibration frequency exceeds the set frequency by ±10%, the wear pattern of the guide vanes under the same conditions in the historical data is called for comparison and maintenance recommendations are generated.

[0058] When both temperature and pressure are abnormal at the same time, the basic deep learning model training is initiated, and the wear condition is re-evaluated using the processed historical data set.

[0059] When the flow rate data fluctuates irregularly, data with similar flow rates are searched in the historical data for comparative analysis, and the real-time update prediction model is activated.

[0060] S2: Identify similar features of similar guide vanes, extract key wear patterns through similar features, and establish corresponding prediction models for different guide vanes.

[0061] Furthermore, similar requirements for identifying similar guide vanes are specified. The geometry, material and operating environment characteristics of the guide vanes are processed and learned through a deep learning algorithm. The similar requirements are specified as follows:

[0062] Requirement 1: Geometric parameter characteristics of the guide vane. When the difference in the geometric shape of the guide vane, including length, width, thickness, and angle, is within 10% and the geometric parameter similarity reaches 0.85, it is confirmed as a similar type;

[0063] Requirement 2: The material property characteristics of the guide vanes. When the material characteristic similarity exceeds 0.9 and the material composition of similar guide vanes including hardness, wear resistance, and density does not differ by more than 5%, they are confirmed as similar types;

[0064] Requirement 3: The operating environment characteristics of the guide vanes. When the temperature and humidity changes of the guide vanes do not exceed 15% and the environmental similarity index exceeds 0.8, it is confirmed as a similar type;

[0065] When the characteristic requirements between the guide vanes meet any two requirements from requirement one to requirement three, they are confirmed as similar guide vanes.

[0066] It should be noted that after similar guide vanes are determined, similar features of similar guide vanes are identified according to similar requirements and guide vane properties, and key wear patterns are extracted by calculating the distance and similarity between similar features;

[0067] The similar features include extracting the guide vane features d that meet similar requirements, and the feature vectors f of the guide vanes in different states are:

[0068]

[0069] Among them, β represents the influence of the first part, n is the number of guide vane feature points, d i The i-th guide vane feature point, is the mean of the guide vane feature data set, γ adjusts the second part to reduce the influence of extreme values, i and j are variable indexes, The first part represents the sum of squares of the differences between all data points and the mean, Calculate the sum of all data points for the second part;

[0070] Define the distance metric and use the improved Manhattan distance D new :

[0071]

[0072] Among them, (f i ,f j ) represents the i-th and j-th eigenvectors, k and l are the variable indices for summation, p represents the number of features contained in the eigenvector, |f i,k -f j,k | represents the difference between the i-th and j-th feature vectors on the k-th feature;

[0073] Convert distance into similarity S similarity :

[0074]

[0075] Among them, X is a constant, representing the constant of the maximum feature difference;

[0076] After obtaining the similarity between the guide vanes, the guide vanes are mapped to extract the key wear patterns.

[0077] Furthermore, the guide vane wear is mapped:

[0078] M(A)=∑ B∈C(A) S similarity (f i ,f j )·(C B -C A )

[0079] Where M(A) represents the wear pattern mapping of guide vane A, C(A) is the set of all guide vanes similar to guide vane A, and C B and C A are the initial wear state levels of guide vanes B and A set according to normal wear of the guide vanes;

[0080] By identifying the key wear mode R = argmax(M(A)) through wear mapping, the key wear modes include uneven wear mode, which is the situation where local wear on the guide vane surface is aggravated; corrosive wear mode, which is uniform wear caused by the chemical properties of the material; fatigue wear mode, which is the wear of microcracks inside the material due to repeated loads; and collision wear mode, which is instantaneous wear caused by external forces:

[0081] When ΔR>θ1 and repeated wear peaks are observed within the set range, indicating that local wear is aggravated, it is determined to be an uneven wear mode;

[0082] When the R value shows a long-term uniform decline and the decline exceeds θ2, the data relative chemical property analysis is performed. When the wear rate is not within the standard deviation of the uniform value of the change, it is determined to be a corrosive wear mode;

[0083] When the R value shows a change of more than θ3 in x consecutive measurements and the peak value of the change corresponds to the fluctuation of the load data, a secondary judgment is made at this time. When the internal stress analysis results show that the formation rate of microcracks increases by more than θ4 in the time series, it is judged as fatigue wear;

[0084] When any of the R value readings increases suddenly and the corresponding vibration sensor displays an abnormal peak value, it is determined to be a collision wear mode;

[0085] Among them, θ1, θ2, θ3, and θ4 are different thresholds set for different turbine environments and operating conditions, and ΔR is a time-based change value.

[0086] S3: Based on the extracted features and prediction model, the guide vane wear is analyzed and the wear prediction results are provided.

[0087] Furthermore, a wear prediction function is constructed, and the output is the expected degree of wear, which indicates the expected future wear:

[0088]

[0089] Among them, δ is the sensitivity of the adjustment model, T represents the real-time ambient temperature, and T 0 is the reference standard temperature;

[0090] Compare the predicted values ​​with historical wear data to determine the upward trend of wear and analyze the causes of wear.

[0091] It should be noted that when the future predicted wear value exceeds the wear rate of 10%, the wear is serious and marked as an emergency alarm, and an inspection is carried out immediately;

[0092] When the predicted wear value is between 5% and 10%, the wear degree is moderate and marked as a moderate alarm. Set weekly monitoring and calculate the change rate ΔF of the predicted wear value in real time. When ΔF>1% / week, the moderate alarm is upgraded to an emergency alarm. When ΔF<1% / week, the original alarm mode remains unchanged.

[0093] When the predicted wear value is less than 5%, it is considered that the wear degree is within the normal loss range and does not affect the normal operation of the turbine. It is allowed to be ignored and the current state is maintained for real-time detection.

[0094] Embodiment 2, the second embodiment of the present invention, is different from the first two embodiments in that:

[0095] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0096] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0097] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0098] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0099] Example 3, reference Figure 2 , which is the third embodiment of the present invention, and provides a guide vane wear prediction system based on deep learning, including a data processing module 10, a guide vane judgment module 20, a prediction module 30 and a wear analysis module 40;

[0100] The data processing module 10 collects and integrates the real-time sensor data and the historical wear data of similar guide vanes, performs standardization processing on the real-time data, assigns weights to the integrated data to ensure that the similarity weight is greater than 0.5, and transmits the assigned data to the guide vane judgment module 20;

[0101] The guide vane judgment module 20 receives the standardized data from the data processing module 10, identifies and confirms similar guide vanes, extracts similar features of similar guide vanes based on geometric parameters, material properties and operating environment characteristics, calculates the distance and similarity between the features, maps and extracts key wear patterns, and transmits the results to the prediction module 30;

[0102] The prediction module 30 receives the key wear pattern from the guide vane judgment module 20, constructs a wear prediction function, performs wear prediction based on the deep learning model, outputs a predicted value of the future wear degree, compares the predicted value with the historical data, analyzes the wear trend, and passes the prediction result and the wear trend analysis to the wear analysis module 40;

[0103] The wear analysis module 40 receives the wear prediction results of the prediction module 30, analyzes the predicted wear conditions, determines the severity of the wear, sets an alarm level according to the severity of the wear, monitors the rate of change of the wear prediction value, and adjusts the alarm level.

[0104] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A guide vane wear prediction method based on deep learning, characterized in that: include, Collect real-time sensor data of guide vanes and integrate historical wear data of similar guide vanes; Identify similar features of similar guide vanes, extract key wear patterns through similar features, and establish corresponding prediction models for different guide vanes; Based on the extracted features and prediction model, the guide vane wear is analyzed and the wear prediction results are provided.

2. The guide vane wear prediction method based on deep learning according to claim 1, characterized in that: The real-time sensor data includes guide vane displacement, vibration frequency, temperature change, pressure data, flow rate data, operating time and fault log; The historical wear data of similar integrated guide vanes is used as a reference for real-time data preprocessing, all real-time Min-Max scaling is standardized, data exceeding the specified range is eliminated, and weights are assigned according to the actual similarity between the real-time data of the guide vanes in use and the historical wear data of the corresponding guide vanes, ensuring that the sum of the similarity weights is greater than 0.

5.

3. The guide vane wear prediction method based on deep learning according to claim 2, characterized in that: The identifying similar features of similar guide vanes includes specifying similar requirements for identifying similar guide vanes, processing and learning the geometry, material and operating environment characteristics of the guide vanes through a deep learning algorithm, and specifying the similar requirements as: Requirement 1: Geometric parameter characteristics of the guide vane. When the difference in the geometric shape of the guide vane, including length, width, thickness, and angle, is within 10% and the geometric parameter similarity reaches 0.85, it is confirmed as a similar type; Requirement 2: The material property characteristics of the guide vanes. When the material characteristic similarity exceeds 0.9 and the material composition of similar guide vanes including hardness, wear resistance, and density does not differ by more than 5%, they are confirmed as similar types; Requirement 3: The operating environment characteristics of the guide vanes. When the temperature and humidity changes of the guide vanes do not exceed 15% and the environmental similarity index exceeds 0.8, it is confirmed as a similar type; When the characteristic requirements between the guide vanes meet any two requirements from requirement one to requirement three, they are confirmed as similar guide vanes.

4. The guide vane wear prediction method based on deep learning according to claim 3, characterized in that: The similar features include: After similar guide vanes are identified, similar features of similar guide vanes are identified according to similar requirements and guide vane properties, and key wear patterns are extracted by calculating the distance and similarity between similar features; The similar features include extracting guide vane features that meet similar requirements, defining a distance metric, and using an improved Manhattan distance D new (f i ,f j ), convert the distance into similarity S similarity : Among them, X is a constant, representing the constant of the maximum feature difference; After obtaining the similarity between the guide vanes, the guide vanes are mapped to extract the key wear patterns.

5. The guide vane wear prediction method based on deep learning according to claim 4, characterized in that: The key wear mode includes performing wear mapping on the guide vane, identifying the key wear mode R through the wear mapping, and judging that the key wear mode includes uneven wear mode, corrosive wear mode, fatigue wear mode, and collision wear mode: When ΔR>θ1 and repeated wear peaks are observed within the set range, indicating that local wear is aggravated, it is determined to be an uneven wear mode; When the R value shows a long-term uniform decline and the decline exceeds θ2, the data relative chemical property analysis is performed. When the wear rate is not within the standard deviation of the uniform value of the change, it is determined to be a corrosive wear mode; When the R value shows a change of more than θ3 in x consecutive measurements and the peak value of the change corresponds to the fluctuation of the load data, a secondary judgment is made at this time. When the internal stress analysis results show that the formation rate of microcracks increases by more than θ4 in the time series, it is judged as fatigue wear; When any of the R value readings increases suddenly and the corresponding vibration sensor displays an abnormal peak value, it is determined to be a collision wear mode; Among them, θ1, θ2, θ3, and θ4 are different thresholds set for different turbine environments and operating conditions, and ΔR is a time-based change value.

6. The guide vane wear prediction method based on deep learning according to claim 5, characterized in that: The establishment of the corresponding prediction model includes constructing a wear prediction function, the output of which is the expected degree of wear, indicating the expected future wear condition: Among them, δ is the sensitivity of the adjustment model, T represents the real-time ambient temperature, and T0 is the reference standard temperature; Compare the predicted values ​​with historical wear data to determine the upward trend of wear and analyze the causes of wear.

7. The guide vane wear prediction method based on deep learning according to claim 6, characterized in that: The guide vane wear condition analysis includes: when the future predicted wear value exceeds a wear rate of 10%, the wear is serious, an emergency alarm is marked, and an inspection is performed immediately; When the predicted wear value is between 5% and 10%, the wear degree is moderate and marked as a moderate alarm. Set weekly monitoring and calculate the change rate ΔF of the predicted wear value in real time. When ΔF>1% / week, the moderate alarm is upgraded to an emergency alarm. When ΔF<1% / week, the original alarm mode remains unchanged. When the predicted wear value is less than 5%, it is considered that the wear degree is within the normal loss range and does not affect the normal operation of the turbine. It is allowed to be ignored and the current state is maintained for real-time detection.

8. A system using a guide vane wear prediction method based on deep learning as claimed in any one of claims 1 to 7, characterized in that: It includes data processing module, guide vane judgment module, prediction module and wear analysis module; The data processing module collects and integrates real-time sensor data and historical wear data of similar guide vanes, performs standardization processing on the real-time data, assigns weights to the integrated data to ensure that the similarity weight is greater than 0.5, and transmits the assigned data to the guide vane judgment module; The guide vane judgment module receives the standardized data from the data processing module, identifies and confirms similar guide vanes, extracts similar features of similar guide vanes based on geometric parameters, material properties and operating environment characteristics, calculates the distance and similarity between the features, maps and extracts key wear patterns, and passes the results to the prediction module; The prediction module receives the key wear pattern from the guide vane judgment module, constructs a wear prediction function, performs wear prediction based on the deep learning model, outputs a predicted value of future wear degree, compares the predicted value with historical data, analyzes the wear trend, and passes the prediction result and wear trend analysis to the wear analysis module; The wear analysis module receives the wear prediction result of the prediction module, analyzes the predicted wear condition, determines the severity of the wear, sets the alarm level according to the severity of the wear, monitors the change rate of the wear prediction value, and adjusts the alarm level.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a guide vane wear prediction method based on deep learning as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a guide vane wear prediction method based on deep learning as described in any one of claims 1 to 7 are implemented.

Images