Railway locomotive intelligent health assessment method and system

Abstract

The invention discloses a railway locomotive intelligent health assessment method comprising the following steps: constructing a locomotive health assessment index system, and setting health assessment result empirical values based on hierarchical monitoring; endowing a health assessment index system with the initial weight value, training to form a health assessment calculation model, and inputting data of a to-be-assessed object of the locomotive based on the health assessment calculation model to obtain a health assessment score of the assessed object; clustering the health assessment scores into health assessment grades through clustering operation, and comparing the health assessment grades with empirical values to generate an error vector group; and based on the initial weight value and the error vector group of the health assessment index system, a better weight combination is obtained through optimization calculation and continuous iteration, health assessment calculation steps are input, and an optimal health assessment calculation model is optimized. According to the method, the weight of each index in the health assessment index system is optimally adjusted by introducing the optimal weight adjustment model, so that an automatic dynamic optimization model is formed.

Description

Technical Field

[0001] The present application relates to the field of railway locomotive health assessment methods, and in particular to a railway locomotive health assessment method based on optimized weights and a system thereof. Background Art

[0002] Fault prediction and health management PHM technology refers to the use of sensor information, expert knowledge and maintenance support information, with the help of various intelligent algorithms and reasoning models, to achieve monitoring, prediction, judgment and management of the operating status of target equipment, to achieve high-reliability equipment operation monitoring and control and high-accuracy fault detection and processing, and ultimately to achieve intelligent task planning and intelligent maintenance based on the historical, current and future status of equipment, to replace traditional event-based post-maintenance or time-based regular maintenance. This technology has been widely used in military, aviation, aerospace, shipbuilding and other fields.

[0003] As an important means in the transportation field to promote the transformation of railway equipment operation and maintenance from traditional operation and maintenance mode to digital and intelligent operation and maintenance mode, locomotive PHM research is currently an important technical support for promoting the reform of locomotive maintenance process and system, reducing the operation and maintenance cost of locomotives throughout their life cycle, and building a locomotive equipment operation and maintenance system.

[0004] Health assessment is one of the important contents in PHM research. It is to build an evaluation index system for the health status of the object by comprehensively utilizing a series of data related to its health status, such as basic information, monitoring and detection information, and accident and fault information generated by the object during operation and maintenance. By assigning grades or scores to different indicators, the health status of the object is finally evaluated according to the comprehensive level or total score.

[0005] However, the current research status of locomotive PHM health assessment methods is as follows:

[0006] During the evaluation process, the setting of levels and scores generally requires the participation of industry experts, who will give corresponding setting opinions based on their business experience.

[0007] At the same time, different indicators have different degrees of influence on the health assessment results due to their different contents and natures. For example, basic locomotive data such as vehicle model and production time often have a lower impact on its health status, while operation and maintenance data such as failures, accidents, and mileage often have a higher impact. Therefore, for the constructed health assessment indicator system, it is often necessary to assign corresponding weight values ​​to each indicator to characterize its degree of influence on the health assessment results.

[0008] At present, when providing relevant opinions, industry experts often set the level or score of each indicator based on their business experience, and assign weights to the corresponding indicators based on their experience and subjective opinions. The full participation of industry experts helps to quickly build a health assessment indicator system and form a corresponding health assessment model; however, due to the natural subjectivity of industry experts, the established health assessment model generally has an artificial subjective "color", and as the degree of participation deepens, this subjectivity continues to strengthen, which in turn weakens the endogenous driving force of the data and its objective role to a certain extent, resulting in reduced flexibility and objectivity of the model.

[0009] In order to solve the above-mentioned technical problems, it is urgent to propose a method for constructing a railway locomotive health assessment model based on optimal weight adjustment. In view of the inevitable link of "industry expert participation", by introducing the optimal weight adjustment module, the weights of each indicator in the health assessment index system are optimally adjusted to reduce the degree of human subjective influence, form a dynamic optimization model, and thus improve the flexibility, accuracy and objectivity of model analysis. Summary of the invention

[0010] In order to solve the above-mentioned defects of the prior art, the embodiment of the present application provides a railway locomotive health assessment method based on optimized weights to solve the technical problem that human subjective factors affect the accuracy of the assessment results.

[0011] In a first aspect, an embodiment of the present application provides a railway locomotive intelligent health assessment method, the method comprising:

[0012] Initialization step: construct the locomotive health assessment indicator system and set the health assessment result experience value based on hierarchical monitoring;

[0013] Health assessment calculation steps: assigning initial weight values ​​to the health assessment indicator system, training to form a health assessment calculation model, inputting data of the locomotive to be assessed based on the health assessment calculation model, and obtaining the health assessment score of the assessment object;

[0014] Clustering operation steps: cluster the health assessment scores into health assessment grades through clustering operation, and compare them with the empirical values ​​to generate an error vector group;

[0015] Optimization weight adjustment step: Based on the initial weight value and error vector group of the health assessment indicator system, a better weight combination is obtained through optimization calculation and continuous iteration, and the health assessment calculation step is input to optimize the optimal health assessment calculation model.

[0016] In the embodiment of the present invention, the above health assessment calculation step includes:

[0017] Steps for constructing the health assessment indicator system: construct a health assessment indicator system, which is a three-level indicator system;

[0018] Steps for constructing a health assessment model: construct a health assessment model based on the three-level indicator system and the weight values ​​corresponding to each level of indicators;

[0019] The locomotive data to be evaluated is input into the health assessment model, and the locomotive health assessment score is output through calculation.

[0020] In the embodiment of the present invention, the clustering operation step includes:

[0021] Assessment and grading steps: clustering the health assessment scores of the locomotives using a clustering algorithm to obtain the corresponding health assessment clustering and grading results;

[0022] Error vector group generation step: Based on the clustering grading results, calculate the discrimination error of the health assessment score, and combine it with the corresponding level to form an error vector group, and calculate the error term corresponding to any level, and input the error vector group into the optimization weight adjustment step. The error vector group includes: multiple clustering grades obtained in the clustering operation and locomotive error terms with grading deviations.

[0023] In an embodiment of the present invention, the above-mentioned optimization weight adjustment step further includes: using an acceleration factor to accelerate the adjustment of the weight value of the health assessment index system, assigning an acceleration value to the dimension of the corresponding acceleration factor to perform acceleration adjustment for optimizing the weight adjustment, and the acceleration factor is a set of multidimensional vectors, in which each dimension of the acceleration factor corresponds to the corresponding three-level indicator.

[0024] In the embodiment of the present invention, the error vector group Δ may be expressed as:

[0025] Among them, K1, K2, ..., K K Represents the K obtained in the clustering operation K Clustering levels, is the error term of the locomotive with graded deviation;

[0026] Among them, the error term The calculation method is: in, Represents the error term The corresponding cluster group K i The amount of data contained in is the cluster group K i The cluster centroid of Indicates that a certain one should be classified as group K i However, the data values ​​that were not classified into this group in the previous clustering calculation, n jIndicates the total number of data with this deviation.

[0027] In the embodiment of the present invention, the error vector group and the original weight value of the health assessment index system before optimization are input into the optimization weight adjustment step, and further optimization operation is performed to iterate and obtain a new weight combination until the corresponding error value is close to 0 or less than the set threshold. Complete the optimization and adjustment of the model;

[0028] Among them, the minimum error threshold It can be expressed as:

[0029] in, Respectively represent the minimum value set for each level of deviation, then It can be expressed as:

[0030] In the embodiment of the present invention, the health assessment model of the locomotive is:

[0031] u new =100-(p 1_new *(n 1_old +l1)+p 2_new *(n 2_old +l2)+...+p N_new *(n N_old +l N )) Among them, p 1_new 、p 2_new ,…,p N_new Indicates the new indicator weight value, n 1_old 、n 2_old ,…,n N_old Indicates the old indicator scores, l1, l2, ..., l N Indicates the score adjustment amount, the score adjustment amount l1, l2, ..., l N Increase or decrease the existing indicator scores based on the actual production business situation.

[0032] In a second aspect, the embodiment of the present application provides a railway locomotive intelligent health assessment system, which adopts the railway locomotive intelligent health assessment method as described above, and the system includes:

[0033] Initialization module: used to build a locomotive health assessment indicator system and set the experience value of health assessment results based on hierarchical monitoring;

[0034] Health assessment calculation module: used to assign initial weight values ​​to the health assessment indicator system, train to form a health assessment calculation model, input the data of the locomotive to be assessed based on the health assessment calculation model, and obtain the health assessment score of the assessment object;

[0035] Clustering operation module: used to cluster the health assessment scores into health assessment grades through clustering operation, and generate error vector groups by comparing with the empirical values;

[0036] Optimization weight adjustment module: It is used to obtain a better weight combination through continuous iteration based on the initial weight value and error vector group of the health assessment indicator system, input the health assessment calculation steps, and optimize the optimal health assessment calculation model.

[0037] In a third aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, the steps of the above-mentioned railway locomotive intelligent health assessment method are implemented.

[0038] In a fourth aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the railway locomotive intelligent health assessment method as described above when executing the program.

[0039] Compared with the related prior art, it has the following outstanding beneficial effects:

[0040] 1. The present invention is aimed at the actual application needs of locomotive health assessment in the process of locomotive health management and PHM application. At the same time, the weights of various health assessment indicators are different, and there is a lot of manual subjective participation in locomotive health management, and the weights of health assessment indicators are difficult to set effectively. A railway locomotive health assessment model construction method based on optimal weight adjustment is proposed;

[0041] 2. The health assessment model of the present invention aims at the inevitable link of "industry expert participation" by introducing an optimal weight adjustment module to optimize the weights of each indicator in the health assessment index system to reduce the degree of human subjective influence. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0043] Figure 1 A schematic diagram of a railway locomotive health assessment method according to the present invention;

[0044] Figure 2 It is a flowchart of a health assessment method according to a specific embodiment of the present invention;

[0045] Figure 3 This is a schematic diagram of the PSO algorithm flow in a specific embodiment of the present invention;

[0046] Figure 4 A schematic diagram of a railway locomotive health assessment system according to the present invention;

[0047] Figure 5 This is a schematic diagram of the training process of the health assessment model according to a specific embodiment of the present invention;

[0048] Figure 6 Schematic diagram of computer hardware of the present invention. DETAILED DESCRIPTION

[0049] It should be noted that the processor described in the present invention is the control center of the electronic device, which can be a processor or a general term for multiple processing elements. For example, it can be one or more central processing units (CPUs), or application specific integrated circuits (ASICs), or one or more integrated circuits configured to implement the embodiments of the present invention, such as one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (FPGAs).

[0050] Optionally, the processor can perform various functions of the electronic device by running or executing a software program stored in the memory, and calling data stored in the memory.

[0051] In a specific implementation, as an embodiment, the processor may include one or more CPUs. Each of these processors may be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here may refer to one or more devices, circuits, and / or processing cores for processing data (such as computer program instructions). Electronic devices may include: servers, desktop computers, laptops, smart phones, tablet computers, embedded computers, etc., wherein the embedded computers include vehicles and robots, etc.

[0052] The memory is used to store the software program for executing the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0053] It should be noted that the structure of the electronic device shown in the drawings of the present invention does not constitute a limitation thereto, and the actual knowledge structure recognition device may include more or fewer components than shown in the drawings, or a combination of certain components, or a different arrangement of components.

[0054] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0055] It should also be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0056] In the present invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0057] It should also be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0058] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0059] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0060] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0061] 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 can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for enabling 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 various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0062] In order to make the above features and effects of the present invention more clearly and understandably described, embodiments are given below and described in detail with reference to the accompanying drawings. This specification discloses one or more embodiments that include the features of the present invention. The disclosed embodiments are only for illustration. The scope of protection of the present invention is not limited to the disclosed embodiments, and the present invention is defined by the attached claims.

[0063] The following is a system embodiment corresponding to the above method embodiment. This embodiment can be implemented in conjunction with the above embodiment. The relevant technical details mentioned in the above embodiment are still valid in this embodiment. In order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the above embodiment.

[0064] The present invention aims to propose a railway locomotive health assessment method based on clustering and optimal weight adjustment, which minimizes the depth of human participation under reasonable consideration of production reality, and through mathematical calculation and data-driven methods, the weight value of each indicator in the health assessment index system is generated as much as possible through data calculation, thereby enhancing the objectivity and flexibility of the model.

[0065] Step 1: Integrate the available data points related to locomotive operation and maintenance to build a locomotive health assessment indicator system;

[0066] Step 2: Select an appropriate amount of operation and maintenance production data related to locomotive health assessment indicators as sample input data to participate in the setting and optimization of model parameters;

[0067] Step 3: Industry experts set the health scores (generally a deduction system) associated with the relevant data of each indicator of the constructed health assessment indicator system according to management regulations, production practices and experience, and introduce "initial weights" to the weight values ​​of each indicator of the indicator system, thereby completing the setting of the score and weight of the indicator system;

[0068] Step 4: Input the sample set data into the "health assessment model" to obtain the corresponding output value after initialization, that is, the health assessment score corresponding to the locomotive;

[0069] Step 5: Input the obtained health assessment scores into the "Clustering Operation (Health Grading)" module, use the clustering algorithm to perform clustering operations on the health scores, and divide the health score data into corresponding groups according to the set number of clusters, and then divide them into corresponding levels, so as to obtain score ranges of different levels;

[0070] In step 6, industry experts adjust the health classification intervals if necessary based on the actual production and operation and maintenance conditions of the locomotives, or screen out locomotives whose cluster classification results do not match the actual operation and maintenance conditions, and then trace back and calculate the discrimination errors of the health scores of locomotives with classification differences, and combine them with the corresponding levels to form an error vector group;

[0071] Step 7: Input the original weight values ​​of each indicator in the health assessment indicator system and the "error vector group" into the "optimal weight adjustment step" and obtain a new weight combination through continuous iteration through optimization operations. Then update the new weight combination and assign it to the "basic calculation step of health assessment" so that the health grade corresponding to the model output value is continuously adjusted to the actual production until the difference in the "error vector group" is 0 or the difference enters the ideal range;

[0072] In step 8, the final “health assessment model” containing the optimized index weight values ​​is put into formal application, new locomotive data is input, and the corresponding health assessment results of the locomotive are obtained.

[0073] In addition, during the model calculation and the optimization adjustment of indicator weights, industry experts can identify the severity of the impact of actual production on different input data, and then use the "acceleration factor" to drive the "basic calculation module of health assessment" to appropriately increase the deduction value of the corresponding indicator item, or drive the "optimization weight adjustment module" to increase the weight value of the corresponding indicator at a certain rate during the calculation process, thereby accelerating the model calculation results to approach the actual production situation.

[0074] In addition, in the formal application of the model, when the "production-based hierarchical monitoring (adjustment mechanism)" issues a new adjustment change instruction, the model will start a new parameter adjustment, and the "optimal weight adjustment module" will also recalculate the weight value of each evaluation indicator, so that the "health assessment basic calculation module" can complete the iteration and update the locomotive's health assessment score and grading results.

[0075] Embodiment 1

[0076] like Figure 1 and Figure 2 As shown, the present application embodiment provides a railway locomotive intelligent health assessment method, comprising:

[0077] Initialization step S10: constructing a locomotive health assessment index system and setting health assessment result experience values ​​based on hierarchical monitoring;

[0078] Health assessment calculation step S20: assigning initial weight values ​​to the health assessment index system, training to form a health assessment calculation model, inputting data of the locomotive to be assessed based on the health assessment calculation model, and obtaining a health assessment score of the assessment object;

[0079] Clustering operation step S30: clustering the health assessment scores into health assessment grades through clustering operation, and comparing them with the empirical values ​​to generate an error vector group;

[0080] Optimization weight adjustment step S40: Based on the initial weight value and error vector group of the health assessment indicator system, a better weight combination is obtained through optimization calculation and continuous iteration, and input into the health assessment calculation step to optimize the optimal health assessment calculation model.

[0081] In the embodiment of the present invention, the health assessment calculation step S20 includes:

[0082] Steps for constructing the health assessment indicator system: construct a health assessment indicator system, which is a three-level indicator system;

[0083] Steps for constructing a health assessment model: construct a health assessment model based on the three-level indicator system and the weight values ​​corresponding to each level of indicators;

[0084] The locomotive data to be evaluated is input into the health assessment model, and the locomotive health assessment score is output through calculation.

[0085] In a specific embodiment of the present invention, the above health assessment calculation step S20 is mainly based on the locomotive operation and maintenance characteristics, using various locomotive operation and maintenance data, and constructing a locomotive health assessment index system, on which a basic calculation structure for the locomotive health condition is formed. The input data of this module is the specific value of each indicator in the index system, and its output is the corresponding health score of the evaluated object.

[0086] The locomotive health assessment index system is generally composed of indicators in multiple dimensions, and these dimensions basically cover the entire locomotive operation and maintenance process, such as basic information, operation information, maintenance information, maintenance information, etc. These dimensions can also be called "first-level indicators". In these dimensions, the content of the relevant dimensions is generally further subdivided according to the actual situation of production operations and management, and the corresponding evaluation indicators are expanded. These specific evaluation indicators can be called "secondary indicators". For example, basic information can include "manufacturer", "assigned unit", "line type", etc.; for example, operation information can include "daily mileage", "accumulated mileage", "equipment alarm", "operation failure", etc.; for example, maintenance information can include "repair process", "repair times", "repair failure", "fault level" and other indicator items.

[0087] Furthermore, how different "secondary indicators" are associated with different health assessment scores (generally a deduction system) also needs to form different specific contents according to the actual acquisition of relevant data, that is, further refined indicator points, which can be called "third-level indicators". For example, fault indicators often include a certain fault point of a certain equipment, while the operating mileage is different kilometer values; these different fault points or kilometer values ​​often reflect the different severity of the problem or the different conditions of the operating quality. These different degrees or conditions will correspond to different scores (generally a deduction system), thereby realizing the calculation and acquisition of the locomotive health score.

[0088] It should be noted that the specific "secondary indicators" and "third-level indicators" included in these "first-level indicators" generally need to be combined with the data support of relevant sensors, production information systems or operation inspections to construct a corresponding three-level evaluation indicator system (among which, the automatic acquisition and collection of production information system data is the most important); at the same time, the setting of relevant dimensions is often related to the data situation, and it is necessary to classify and form corresponding dimensions based on the data available. Therefore, the dimensions and indicator settings of the locomotive health assessment system are not static, and need to be combined with the production and operation of the locomotive to form an assessment system that is more in line with its operation and maintenance characteristics. Table 1 is an example of the construction of a health assessment indicator system.

[0089] Table 1 Example of constructing a health assessment indicator system

[0090]

[0091]

[0092] Table 1 is a health assessment index system that contains different locomotive operation and maintenance information from the perspective of different locomotive operation and maintenance links. Similarly, from the perspective of locomotive assembly relationships, a corresponding health assessment index system can be constructed based on different key components of the locomotive. This often requires the construction of an appropriate health assessment index system based on the actual production operations of the locomotive, data acquisition methods, management needs, etc.

[0093] After the health assessment index system is built, industry experts will set the health scores associated with the relevant production and operation data for each indicator included in the "three-level indicators" in the constructed health assessment index system according to the management regulations, production practices and experience, and the full score can be set to 100 points. When the corresponding data occurs, the corresponding three-level indicator will deduct the corresponding score from the total score.

[0094] At the same time, the "initial weight" will be introduced into the weight value of each indicator of the "three-level indicator" in the indicator system, so as to complete the setting of the score and weight of the indicator system. According to actual production, the initial weight of the fault indicator can be set to 1, and the initial weight of other types of indicators can be set to 0.5.

[0095] Assume that there are N three-level indicators, and their corresponding deduction values ​​are n1, n2, …, n N Among them, the number of fault indicators is I, and the deduction value of these indicators is n. I Indicates; the number of other types of indicators is J, and the deduction value of these indicators is n J Indicates that I+J=N. Correspondingly, the weight values ​​of each third-level indicator are p1, p2, …, p N , and in the model initialization stage, the initial weight value p of the fault class indicator I =1, the initial weight value of other types of indicators p J = 0.5. Then for the “health assessment basic calculation module”, the calculation method of the health assessment score u of a certain locomotive is:

[0096] u=100-(p1*n1+p2*n2+…+p N *n N ) (1-1)

[0097] Among them, the health assessment score obtained in the initialization stage can be summarized as u = 100-(p I *n I +p J *n J ).

[0098] Through the above calculation method, the operation and maintenance production data of the locomotive can be converted into the health assessment score of the locomotive through calculation.

[0099] It should be noted that the calculation method of the locomotive health assessment score and the setting of the initial weight can be diverse and need to be set according to actual production and management needs. The present invention is only described here using the percentage deduction value as an example.

[0100] In the embodiment of the present invention, the clustering operation step S30 includes:

[0101] Assessment and grading steps: clustering the health assessment scores of the locomotives using a clustering algorithm to obtain the corresponding health assessment clustering and grading results;

[0102] Error vector group generation step: Based on the clustering grading results, calculate the discrimination error of the health assessment score, and combine it with the corresponding level to form an error vector group, and calculate the error term corresponding to any level, and input the error vector group into the optimization weight adjustment step. The error vector group includes: multiple clustering grades obtained in the clustering operation and locomotive error terms with grading deviations.

[0103] In a specific embodiment of the present invention, the above clustering operation step S30 will take the K-means++ clustering algorithm as an example to illustrate the calculation of the corresponding health assessment grade by clustering the health assessment score of the locomotive. The K-means++ clustering algorithm is a set of algorithms formed by upgrading and improving the K-means clustering algorithm. The K-means clustering algorithm has the advantages of fast, efficient and highly applicable calculations, but its clustering effect is easily affected by the initial cluster centroid setting and falls into the local optimum; the K-means++ clustering algorithm is an improvement and correction to this problem.

[0104] The K-means algorithm and the K-means++ algorithm are both distance-based clustering algorithms. They use distance as the similarity evaluation index and divide the data into K pre-set categories based on minimizing the error function. Each category consists of objects that are close to each other. The algorithm takes obtaining compact and independent clusters as the ultimate goal. The basic process of the K-means algorithm is as follows:

[0105] Step 1: For a dataset U with n samples, n}, randomly select K samples as the initial cluster centroid, denoted as μ i , where K≤n, i≤n, and the clustering category is represented by C=C1,C2,…,C k}.

[0106] Step 2: Calculate each sample x in the dataset separately i The distance from the K cluster centroids is calculated, and the sample is divided into the class corresponding to the cluster center with the smallest distance.

[0107] Step 3: After traversing all sample points, recalculate the cluster centroid of each category according to formula (2-1) as the new center point of the category in the next clustering process, where i≤K.

[0108]

[0109] Step 4: Repeat steps 2 and 3 until the cluster centroids of each category no longer change.

[0110] For sample data in Euclidean space, the square error and e SSEAs the objective function of clustering and as an indicator to measure the effects of different clustering, it is expressed as the distance from the sample point x to the cluster category C. i The cluster centroid μ i The sum of the squares of the distances between SSE The value of is calculated according to formula (2-2), when e SSE The smaller the value, the more similar the samples within the cluster category are. The optimal clustering result should make e SSE Reached minimum value.

[0111]

[0112] In order to solve the problem that the K-means algorithm is easily trapped in the local optimum due to the influence of the initial cluster centroid selection, which leads to the reduction of calculation accuracy, the K-means++ algorithm makes the following corrections:

[0113] In order to improve the clustering effect of clustering operation, the distance between the initial centroids should be as far as possible when the K-means++ algorithm selects the initial centroids. The basic idea of ​​centroid optimization selection is:

[0114] Step 1: From the data set U = {x1, x2, ..., x n} randomly select a sample x i As the first cluster centroid μ1.

[0115] Step 2: Calculate the shortest distance D(x) between each sample and the currently selected cluster centroid, and use it as the weight value of the sample to be selected as the next cluster centroid. i D(x i ) value is larger, indicating that the sample x i The greater the probability of being selected as the next cluster center; choose D(x i )The largest sample point becomes the next cluster centroid.

[0116] Step 3: Repeat step 2 until K cluster centroids are selected.

[0117] After selecting K initial centroids, proceed to steps 2 and 3 of the K-means algorithm.

[0118] Maximizing the distance between the initial centroids can significantly improve the error of the final result. Although it takes more time to select the initial centroids, it can increase the convergence speed of the clustering process and the stability of the algorithm, thereby improving the clustering effect.

[0119] Through the above clustering operation, when the "health assessment model" outputs a batch of locomotive health assessment scores, these values ​​can be clustered to obtain the locomotive health assessment classification. At the same time, different levels will also form corresponding division intervals according to the corresponding score data. After completing this operation process, if the operation parameters of the overall model are not changed, the subsequent locomotive health assessment scores will be automatically divided and classified into the corresponding health assessment levels according to this division interval.

[0120] In addition, the setting of the number of clusters K in the clustering operation needs to be proposed according to the actual needs of production operations and operation and maintenance management. Generally, it can be set to 4 levels, that is, K=4, to represent the locomotive health status of "excellent, good, medium, and poor". In daily locomotive operation and maintenance production and management, different health levels often correspond to different operation and maintenance operations. For example, a locomotive with an "excellent" health grade means that no additional inspection and maintenance is required and it can continue to operate; a locomotive with a "medium" health grade means that an in-depth inspection is required to solve existing problems and faults.

[0121] It should be noted that there are many types of clustering algorithms. Other algorithms such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise), GMM (Gaussian Mixture Model), OPTICS (Ordering Points To Identify the Clustering Structure) are also commonly used clustering algorithms. This invention only introduces K-means and K-means++ clustering algorithms as examples.

[0122] The "error vector group" introduced in the present invention integrates the differences between the "clustering operation (health grading)", "production-based grading monitoring (adjustment mechanism)" and the model output scores to form a vector group containing score errors and their corresponding grades. Its function is to drive the "optimization weight adjustment module" to perform optimization operations.

[0123] When industry experts need to adjust the health classification interval according to the actual production and operation and maintenance of the locomotive, or screen out locomotives whose cluster classification results do not match the actual operation and maintenance situation, the model can trace back and calculate the discrimination error of the health score of the locomotive with different classifications, and combine it with the corresponding level to form an error vector group. The error vector group Δ can be expressed as:

[0124]

[0125] Among them, K1, K2, ..., K K represents the K clustering levels obtained in the clustering operation, is the locomotive error term with graded deviation, which is the average deviation distance between its evaluation score and the centroid of the cluster to which it belongs. The calculation method is:

[0126]

[0127] in, Represents the error term The corresponding cluster group K i The amount of data contained in, 1≤K i ≤K K , is the cluster group K i The cluster centroid of Indicates that a certain one should be classified as group K i However, the data values ​​that were not classified into this group in the previous clustering calculation, n j Indicates the total number of data with this deviation.

[0128] According to the above calculation method, the corresponding errors for different cluster groups can be obtained and together constitute an error vector group. It will be used as the driving factor for the "optimal weight adjustment module" to carry out operations, and together with the original weight values ​​of each three-level indicator in the health assessment index system, it will be input into the "optimal weight adjustment module" and through further optimization operations, it will continuously iterate to obtain a new weight combination until the corresponding error value is close to 0 or less than the set threshold. This completes the optimization adjustment of the model.

[0129] Among them, the minimum error threshold It can be expressed as

[0130]

[0131] in, Respectively represent the minimum value set for each level of deviation.

[0132] but It can be expressed as

[0133]

[0134] It should be noted that there are many ways to measure the error value. The present invention only introduces the calculation method of formulas (3-1), (3-2), (3-3) and (3-4) as an example.

[0135] In an embodiment of the present invention, the above-mentioned optimization weight adjustment step S40 further includes: using an acceleration factor to accelerate the adjustment of the weight value of the health assessment index system, assigning an acceleration value to the dimension of the corresponding acceleration factor to perform acceleration adjustment for optimizing the weight adjustment, and the acceleration factor is a set of multidimensional vectors, in which each dimension of the acceleration factor corresponds to the corresponding three-level indicator.

[0136] In the specific embodiment of the present invention, the optimization weight adjustment step will take the particle swarm optimization (PSO) algorithm as an example to illustrate the optimization calculation of the index weight. The PSO algorithm has the advantages of simple structure, fewer parameters, fast convergence speed, high robustness, and strong optimization ability, and is a widely used optimization algorithm.

[0137] The PSO algorithm is a swarm intelligence evolutionary algorithm that originates from the study of bird flock predation behavior. The PSO algorithm simulates a flock of birds by generating a group of particles in the solution space. The particles have two characteristics: speed and position. The position of the particles in the solution space is changed by the speed of movement. The best position searched by each particle is called the individual extreme value P. best , the best individual extreme value among all particles in the population is recorded as the global optimal solution G of the entire particle swarm best In each iteration, all particles are best and G best To dynamically adjust its speed and position, it keeps moving towards the optimal solution. Figure 3 As shown, Figure 3 Shown is the flow chart of the PSO algorithm;

[0138] The calculation process of the PSO algorithm is mainly divided into the following steps:

[0139] Step 1: Initialize the particle population.

[0140] In an N-dimensional search space, a population X consisting of M particles is randomly generated. i ,…,X M ). The i-th particle can be represented as X i =(x i1 ,x i2 ,…,x iN ), which represents the position of the ith particle, i.e., a potential solution to the optimization problem. i By bringing the objective function of the optimization problem into play, we can calculate the corresponding target value, which is used to measure X i The speed of the i-th particle is V i =v i1 ,v i2 ,…,v iN), the individual optimal solution of the i-th particle is P best_i =(p best_i1 ,p best_i2 ,…,p best_iN ), the global optimal solution of the particle population is G best_i =(g best_i1 ,g best_i2 ,…,g best_iN ).

[0141] Step 2: Calculate the velocity of the particle at each iteration.

[0142] Assume the total number of iterations of the particle swarm is G k . Then at the beginning of the g+1th iteration, the i-th particle X i (The iteration speed of g can be calculated by formula (4-1).

[0143]

[0144] Among them, C1 and C2 are called acceleration constants, which are used to measure P best (g) and G best (g) To X i The influence weight of (g) is generally C1=C2∈[0,4], and is usually 2. ω is called the inertia factor, and its value is non-negative. It can be dynamically obtained by using the weight linear reduction strategy as shown in formula (4-2):

[0145]

[0146] Among them, ω ini is the initial inertia weight, generally 0.9; ω end It is the inertia weight when the iteration reaches the maximum number of iterations, which is generally 0.4.

[0147] In addition, V i The initial value V i (0) Generally, the value is randomly selected between [-2, 2].

[0148] Step 3: Generate new particles.

[0149] After the g+1th iteration, the new position X of the i-th particle can be calculated by formula (4-3): i (g+1):

[0150] X i (g+1)=X i (g)+V i (g+1) (4-3)

[0151] According to the above method, the update of all particle positions is completed in one iteration process to obtain a new particle swarm distribution.

[0152] Step 4: Find the individual optimal solution and the global optimal solution.

[0153] X i (g+1) and P best_i (g) Substitute the objective function and compare the advantages and disadvantages of the two to obtain the individual optimal solution P of the i-th particle after the g+1th iteration. best_i (g+1). By comparing the individual optimal solutions of all examples, the global optimal solution G of the particle population after the g+1th iteration is obtained. best (g+1).

[0154] Step 5: Iteration termination judgment.

[0155] Determine whether the particle population meets the iteration termination condition, that is, whether the error value in the error vector group is close to 0 or less than the set value. If the condition is met, output G best , assign new weight values ​​to the "three-level indicators" in the health assessment indicator system; if the termination condition is not met, return to step 2 to start a new iteration process.

[0156] Through the above optimization process, it can be seen that when the error value contained in the error vector group is higher than the set threshold, the "optimization weight adjustment module" is triggered to start the optimization operation until the error value in the error vector group is close to 0 or less than the set threshold. The "optimization weight adjustment module" stops the operation and outputs a new indicator weight to the "health assessment basic calculation module" to complete the update iteration of the model. After completing this operation process, if other model optimization adjustment mechanisms are not triggered, the subsequent new locomotive health assessment operation will be performed with this weight setting.

[0157] It should be noted that there are many types of optimization algorithms, such as genetic algorithm (GA), ant colony optimization (ACO), differential evolution algorithm (DE), imperialist competitive algorithm (ICA), etc., which are also common optimization algorithms. The present invention only takes the PSO algorithm as an example for introduction.

[0158] In the specific embodiment of the present invention, in order to make the model operation more flexible and efficient, and to prevent the model from falling into the local optimum during the operation process, a driving quantity "acceleration factor" is designed. The "acceleration factor" is based on production needs, and industry experts observe and control the input data at the same time through "production-based hierarchical monitoring (adjustment mechanism)" according to the actual production operation. The part of the indicator data that has a greater impact on the production operation is used to form a set of weighted vectors for the corresponding indicators to improve the parameter optimization speed of the "optimization weight adjustment module" and the "health assessment basic calculation module", and guide the global optimal solution to be obtained as soon as possible.

[0159] When accelerating the "optimization weight adjustment module", or in order to prevent the optimization algorithm from falling into a local optimum during the operation, the acceleration factor is a set of N-dimensional vectors T, where each dimension corresponds to a corresponding third-level indicator. When, through data and actual production observations, it is necessary to accelerate the optimization adjustment process of the weight values ​​of certain third-level indicators during the model optimization adjustment process, an acceleration value τ is assigned to the corresponding dimension, and the acceleration values ​​of the remaining dimensions are 0. When the optimization algorithm performs optimal calculations, particle X i The new position X i (g+1) can be expressed as:

[0160] X i (g+1)=X i (g)+V i (g+1)+T (5-1)

[0161] Among them, the acceleration value τ contained in the vector T can be calculated by the following formula:

[0162]

[0163] Among them, G max (g) represents the maximum weight value in the g-th step optimization calculation; G min (g) represents the minimum weight value, and N represents the number of third-level indicators.

[0164] Similarly, when accelerating adjustments are made to the "basic calculation module for health assessment", that is, when the assessment scores (generally a deduction system) associated with certain third-level indicators in the health assessment index system need to be adjusted according to actual production conditions, the acceleration factor can be used to inject an adjustment vector into the health score calculation process in the "basic calculation module for health assessment" (that is, the corresponding scores of the corresponding third-level indicators), thereby changing the calculation results of the health assessment scores, and further triggering the adjustment of the clustering and grading results and the "optimization weight adjustment module" to start a new round of optimization operations.

[0165] Among them, the health assessment score u of a locomotive calculated by the "health assessment model" is new The calculation method is

[0166]

[0167] Among them, p 1_new 、p 2_new ,…,p N_new Indicates the new indicator weight value, n 1_old 、n 2_old ,…,n N_old Indicates the old indicator scores, l1, l2, ..., l N represents the score adjustment amount, and the score adjustment amount l1, l2, ..., l N The existing indicator scores can be increased or decreased according to the actual production business situation, and can be increased or decreased within the numerical range of -0.3,0.3].

[0168] It should be noted that there are many ways to determine the value of the adjustment variable related to the "acceleration factor" introduced in the present invention. The present invention only introduces the calculation method of formulas (5-2) and (5-3) as an example.

[0169] Embodiment 2

[0170] like Figure 4 As shown, the embodiment of the present application provides a railway locomotive intelligent health assessment system, which adopts the railway locomotive intelligent health assessment method as described above, including:

[0171] Initialization module 10: used to construct a health assessment index system for locomotives and set an experience value of health assessment results based on hierarchical monitoring;

[0172] Health assessment calculation module 20: used to assign initial weight values ​​to the health assessment index system, train to form a health assessment calculation model, input data of the locomotive to be assessed based on the health assessment calculation model, and obtain the health assessment score of the assessment object;

[0173] Clustering operation module 30: used to cluster the health assessment scores into health assessment grades through clustering operation, and generate error vector groups by comparing with the experience values;

[0174] The optimal weight adjustment module 40 is used to obtain a better weight combination through continuous iteration based on the initial weight value and error vector group of the health assessment indicator system through optimal calculation, input the health assessment calculation steps, and optimize the optimal health assessment calculation model.

[0175] The overall architecture of a railway locomotive health assessment system based on clustering and optimal weight adjustment proposed in a specific embodiment of the present invention is as follows: Figure 2As shown in the figure: The architecture is mainly composed of modules such as "basic calculation module for health assessment", "clustering operation (health classification)", "optimal weight adjustment module", and "production-based classification monitoring (regulation mechanism)", so as to convert the input data into the health score and classification of the assessment object through calculation. At the same time, parameters such as "initial weight", "error vector group", and "acceleration factor" are introduced into the overall architecture to form the optimization driving factors of the model operation results.

[0176] The basic unit of the system is the "health assessment basic calculation module", which is a basic locomotive health assessment index system established according to the characteristics of locomotive operation and maintenance, and a basic calculation structure for locomotive health based on this index system. The input data of this module is the specific value of each index in the index system, and its output is the health score of the object being assessed.

[0177] The health score of the assessment object is processed by the "Clustering Operation (Health Grading)" module, and the health grade of the assessment object can be obtained through clustering operation. "Clustering Operation (Health Grading)" is a module that converts the health assessment score of the locomotive into a health assessment grade. By constructing a reasonable clustering algorithm and setting the number of clusters, clustering operations can be performed on a large number of sample set data to obtain the score ranges corresponding to different health levels, thereby realizing the grading operation. Generally, in production operations, different locomotive health assessment grades will correspond to different subsequent locomotive maintenance and disposal.

[0178] In order to represent the participation of industry experts and adjust the model according to actual production when necessary, the "production-based hierarchical monitoring (adjustment mechanism)" module represents the participation of industry experts. In order to minimize human participation, this module is composed of industry experts who adjust the division intervals of the evaluation grading according to the actual situation of locomotive operation and maintenance, change the originally set grades or scores, and thus improve the model calculation results.

[0179] In this regard, in order to balance human participation and influence, the architecture processes the grading error through the "optimal weight adjustment module" and optimizes the weight value of each indicator in the "health assessment basic calculation module" through optimization operations. The corresponding new weight value is assigned to each evaluation indicator in the form of data calculation, and after each indicator obtains the optimized weight value, the output of the health assessment model is equal to or close to the ideal result, thereby avoiding direct human intervention in the weight value setting of each indicator and improving the flexibility and reliability of model analysis and calculation.

[0180] In addition, the "initial weight", "error vector group" and "acceleration factor" introduced in the overall system architecture, as optimization driving factors of model results, have different functions. "Initial weight" is used to assign initial weight values ​​to each evaluation indicator in the initial startup stage of model operation to facilitate the normal startup of model operation; "error vector group" is a vector group containing score errors and their corresponding grades, which is formed by integrating "clustering operation (health classification)", "production-based hierarchical monitoring (regulation mechanism)" and the difference between model output scores, and is used to drive the "optimization weight adjustment module" to perform optimization operations; "acceleration factor" is a group of weighted vectors formed by industry experts according to actual production operations through "production-based hierarchical monitoring (regulation mechanism)" for indicators that have a greater impact on production operations in the input data, which is used to improve the parameter optimization speed of "optimization weight adjustment module" and "health assessment basic calculation module" and guide the global optimal solution to be obtained as soon as possible.

[0181] The present invention introduces in detail a series of calculations such as the construction of a health assessment index system, the calculation of health assessment scores, the K-means++ clustering and grading algorithm, the error vector group calculation, the PSO optimization algorithm, and the acceleration factor calculation, and organically combines these calculation processes and applies them to the entire process of locomotive health assessment calculation, so that the model can optimize the weights of each locomotive health assessment indicator in the process of calculating the locomotive health assessment score and grading, so that the model operation parameters approach the best combination state and obtain a more ideal output result. Based on the calculation processes such as the health assessment score calculation, K-means++ clustering and grading algorithm, the error vector group calculation, and the PSO optimization algorithm introduced in the previous subsection, the main operation process of the "railway locomotive health assessment system based on clustering and optimal weight adjustment" proposed in the present invention is as follows Figure 5 As shown, Figure 5 Flowchart for model training operations;

[0182] Combining the model training process and the calculation of health assessment scores, K-means++ clustering algorithm, PSO optimization algorithm and other calculation processes, the calculation process of the system health assessment model training is mainly divided into the following steps:

[0183] Step 1: Prepare data and build the basic model structure

[0184] Define the data range and use it as a sample data set to train the model. According to the content, dimensions, etc. of the locomotive health assessment related data, construct the corresponding health assessment index system and obtain a series of N three-level indicators corresponding to the data items, thereby forming the basic operation structure of the assessment model. At the same time, the three-level indicators of the constructed index system are given initial weights p1, p2, …, p N, the initial weight value p of the fault index I =1, the initial weight value of other types of indicators p J =0.5. At the same time, according to the actual situation of locomotive production and operation, the evaluation deduction value n1, n2, ..., n corresponding to each third-level indicator is set according to the percentage deduction system. N , thus completing the initial setting of parameters of the basic calculation structure of health assessment.

[0185] Step 2: Use K-means++ algorithm to perform clustering and grading operations on the evaluation scores

[0186] The sample data are input into the health assessment basic calculation module in sequence, and the health assessment scores of the locomotives corresponding to these sample data are obtained. Then these scores are input into the clustering and grading module in sequence, and the K-means++ clustering algorithm is started to perform clustering and grading operations. Among them, the number of clusters N=4 is set, that is, 4 levels are divided. After the clustering operation is completed, according to the evaluation scores and grading results, the division score intervals of different grades can be obtained; at the same time, the cluster centroids of different clustering levels can also be obtained.

[0187] Step 3: Use the hierarchical monitoring mechanism to adjust the clustering and grading results as needed

[0188] Through the hierarchical monitoring and adjustment mechanism, the clustering and grading results obtained in step 2 are checked according to the actual operation and maintenance of locomotives. If the evaluation and grading results of the grading interval or some locomotives are different from the actual operation and maintenance, the clustering and grading, health scores and actual production conditions are comprehensively considered to construct the corresponding error vector group Δ, and at the same time, a certain level K is calculated. i The corresponding error term Then go to step 4 and start the PSO algorithm to perform the index weight optimization calculation. If the clustering and classification results are consistent with the actual production, jump to step 7 to end the model training.

[0189] Step 4: Initialize the PSO algorithm and start optimizing the indicator weights

[0190] According to the number N of the three-level indicators in the health assessment index system, a population X consisting of M particles is randomly generated in the N-dimensional search space. i ,…,X M ), that is, M potential indicator weight solutions. Among them, the i-th particle X i =(x i1 ,x i2 ,…,x iN ) represents a potential solution in the indicator weight solution. i Assign locomotive health assessment indicators and complete model initialization.

[0191] Then, the sample data is input into the locomotive health assessment module again, and M groups of new assessment scores and assessment grades are obtained in turn. Then, through the hierarchical monitoring mechanism, the differences between the M groups of clustering grades and health scores and the actual production situation are comprehensively compared, and M error vector groups Δ1, Δ2, …, Δ M .

[0192] Afterwards, the M error vector groups are compared and the error vector group with the smallest error is selected. m , and the corresponding potential solution X generated by the PSO algorithm m , as the optimal solution G of the index weight obtained by this round of optimization iteration calculation best Provide an evaluation indicator system.

[0193] Furthermore, Δ m The minimum error is set For comparison, if If the weight value representing the health assessment index meets the requirements, jump to step 7 and the model training ends; The model training then enters step 5 and continues to calculate the weight values ​​of the optimal health assessment indicators.

[0194] Step 5: Iterate the PSO algorithm

[0195] Run the PSO algorithm to continue iterative operations, and at the same time, according to the error vector group comparison analysis method described in step 4, find the optimal solution G for the index weights formed by each round of iterative optimization operations. best (g), g is the current iteration number.

[0196] Step 6: Determine whether the iteration operation is terminated

[0197] The global optimal solution G generated in each iteration best (g) Assign locomotive health assessment indicators and calculate and analyze the corresponding error vector group Δ m (g). Then determine Δ m (g) with If the size between Then the iteration is terminated and the process goes to step 7, and G best (g) as the final weight value of the health assessment indicator; if Then return to step 5 and continue iterative calculations until the final global optimal solution is found.

[0198] Step 7: Model construction is complete and normal operation begins

[0199] The final global optimal solution G bestThe health assessment index system is assigned so that each third-level index obtains its corresponding weight value, thereby completing the model training. If the basic calculation structure of the health assessment changes, that is, the assessment index system is adjusted, the model training will be repeated from step 1.

[0200] In addition, during the model training process, if the model cannot obtain an evaluation and grading result that is more consistent with the actual production after a long period of training, it indicates that the model training may fall into a local optimal state or a minimum error state. It may be set too small. In case the model training may fall into the local optimum, the "acceleration factor" Τ for the "optimization weight adjustment module" or "health assessment basic calculation module" can be generated through the hierarchical monitoring mechanism - the value of the potential solution in the weight optimization is adjusted as needed, or the evaluation score corresponding to the three-level indicators, so as to adjust the model operation process.

[0201] Embodiment 3

[0202] An embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of the above-mentioned railway locomotive intelligent health assessment method are implemented.

[0203] Embodiment 4

[0204] An embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned railway locomotive intelligent health assessment method when executing the program.

[0205] In addition, combined Figure 1 The railway locomotive intelligent health assessment method described in the embodiment of the present application can be implemented by a computer device. Figure 6 Schematic diagram of the hardware structure of a computer device according to an embodiment of the present application.

[0206] In some of these embodiments, the computer device may further include a communication interface 83 and a bus 80. Figure 6 As shown, the processor 81, the memory 82, and the communication interface 83 are connected via a bus 80 and communicate with each other.

[0207] Specifically, the processor 81 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0208] The memory 82 may be used to store or cache various data files required for processing and / or communication, as well as possible computer program instructions executed by the processor 81 .

[0209] The processor 81 reads and executes the computer program instructions stored in the memory 82 to implement any one of the railway locomotive health assessment methods based on optimal weight adjustment in the above embodiments.

[0210] In summary, compared with the prior art, the method of the present invention is oriented to the actual application needs of locomotive health assessment in the process of locomotive health management and PHM application. At the same time, in view of the different weights of various health assessment indicators and the current locomotive health management, the weights of health assessment indicators are difficult to set effectively, and a railway locomotive intelligent health assessment method and system are proposed. While reducing the degree of human subjective participation as much as possible, the model constructs a health assessment index system, introduces computing units such as clustering operation classification module and optimal weight adjustment module, and organically integrates each unit module, and then optimizes the weight of each indicator in the health assessment index system through the measurement of the error vector group, so as to improve the intelligent computing level of the evaluation model. The present invention also proposes the concept of acceleration factor, through which the training rate of the model can be adjusted when necessary, and the flexibility of model training can be improved. The present invention introduces in detail the basic architecture of the model and the calculation operation process between each module, and then elaborates on the training and calculation steps of the model based on the evaluation index system operation, K-means++ clustering algorithm, error vector group, PSO optimization algorithm and other algorithm principles and calculation processes. The training steps of the model are clear and simple, easy to operate, and have good practicality. A dynamic optimization model based on data-driven and optimization theory is constructed, which improves the flexibility, accuracy and objectivity of the model analysis as a whole, enriches the technical and theoretical connotation of the locomotive health assessment method, and provides a good methodological support for promoting the development of locomotive PHM technology.

[0211] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0212] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the invention of the present application shall be subject to the attached claims.

Claims

1. A railway locomotive intelligent health assessment method, characterized in that: Methods include: Initialization step: construct the locomotive health assessment indicator system and set the health assessment result experience value based on hierarchical monitoring; Health assessment calculation step: assigning an initial weight value to the health assessment index system, training to form a health assessment calculation model, inputting data of the locomotive to be assessed based on the health assessment calculation model, and obtaining a health assessment score of the assessment object; Clustering operation step: clustering the health assessment scores into health assessment grades through clustering operation, and comparing them with the empirical values ​​to generate an error vector group; Optimization weight adjustment step: Based on the initial weight value of the health assessment index system and the error vector group, a better weight combination is obtained through optimization calculation and continuous iteration, and input into the health assessment calculation step to optimize the optimal health assessment calculation model.

2. The railway locomotive intelligent health assessment method according to claim 1 is characterized in that: The health assessment calculation step comprises: Health assessment indicator system construction step: construct the health assessment indicator system, which is a three-level indicator system; Health assessment model construction step: construct a health assessment model according to the three-level indicator system and the weight values ​​corresponding to the indicators at each level; The locomotive data to be evaluated is input into the health evaluation model, and the health evaluation score of the locomotive is output through calculation.

3. The railway locomotive intelligent health assessment method according to claim 1 is characterized in that: The clustering operation step comprises: Evaluation and grading step: using a clustering algorithm to perform clustering operations on the health evaluation scores of the locomotive to obtain corresponding health evaluation clustering and grading results; Error vector group generation step: Based on the clustering grading result, calculate the discrimination error of the health assessment score, and combine it with the corresponding level to form an error vector group, and calculate the error term corresponding to any level, and input the error vector group into the optimization weight adjustment step. The error vector group includes: multiple clustering grades obtained in the clustering operation and locomotive error terms with grading deviations.

4. The railway locomotive intelligent health assessment method according to claim 2 is characterized in that: The optimization weight adjustment step further includes: using an acceleration factor to accelerate the adjustment of the weight value of the health assessment index system, assigning an acceleration value to the dimension corresponding to the acceleration factor to perform accelerated adjustment for optimizing the weight adjustment, the acceleration factor is a set of multidimensional vectors, wherein each dimension of the acceleration factor corresponds to a corresponding third-level indicator.

5. The railway locomotive intelligent health assessment method according to claim 3 is characterized in that: The error vector group Δ may be expressed as: Among them, K1, K2, ..., K K Represents the K obtained in the clustering operation K Clustering levels, is the error term of the locomotive with graded deviation; Among them, the error term The calculation method is: in, Represents the error term The corresponding cluster group K i The amount of data contained in, 1≤K i ≤K K , is the cluster group K i The cluster centroid of Indicates that a certain one should be classified as group K i However, the data values ​​that were not classified into this group in the previous clustering calculation, n j Indicates the total number of data with this deviation.

6. The railway locomotive intelligent health assessment method according to claim 5 is characterized in that: The error vector group and the original weight value of the health assessment index system before optimization are input into the optimization weight adjustment step, and further optimization operations are performed to iterate and obtain a new weight combination until the corresponding error value is close to 0 or less than the set threshold. Complete the optimization and adjustment of the model; Among them, the minimum error threshold It can be expressed as: in, Respectively represent the minimum value set for each level of deviation, then It can be expressed as:

7. The railway locomotive intelligent health assessment method according to claim 2 is characterized in that: The health assessment model of the locomotive is: u new =100-(p 1_new *(n 1_old +l1)+p 2_new *(n 2_old +l2)+…+p N_new *(n N_old +l N )), where p 1_new 、p 2_new ,…,p N_new Indicates the new indicator weight value, n 1_old 、n 2_old ,…,n N_old Indicates the old indicator scores, l1, l2, ..., l N Indicates the score adjustment amount, the score adjustment amount l1, l2, ..., l N The existing indicator scores can be increased or decreased according to the actual production business conditions.

8. A railway locomotive intelligent health assessment system, using the railway locomotive intelligent health assessment method as claimed in any one of claims 1 to 7, characterized in that: The system includes: Initialization module: used to build a locomotive health assessment indicator system and set the experience value of health assessment results based on hierarchical monitoring; Health assessment calculation module: used to assign initial weight values ​​to the health assessment index system, train to form a health assessment calculation model, input data of the locomotive to be assessed based on the health assessment calculation model, and obtain the health assessment score of the assessment object; Clustering operation module: used for clustering the health assessment scores into health assessment grades through clustering operation, and generating an error vector group by comparing with the experience value; Optimization weight adjustment module: used to obtain a better weight combination through continuous iteration based on the initial weight value of the health assessment indicator system and the error vector group, input the health assessment calculation steps, and optimize the optimal health assessment calculation model.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the railway locomotive intelligent health assessment method described in any one of claims 1-7 are implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the railway locomotive intelligent health assessment method according to any one of claims 1 to 7 are implemented.

Images