Fault diagnosis method and system for grinding machine tool

Abstract

The invention provides a fault diagnosis method and system for a grinding machine tool, and relates to the technical field of fault diagnosis, and the method comprises the steps: collecting a historical fault dominant feature library and a fault traceability database of a grinding machine tool subsystem through a data sharing platform connected with the grinding machine tool, and constructing a fault early warning model and an early warning sensing module; fault propagation model construction and probability identification are carried out on the subsystems based on the traceability library; monitoring is carried out through the sensing module, and monitoring data is input into the early warning model; the early warning signal is input into the identification propagation model for analysis, and troubleshooting control is carried out according to the propagation probability from large to small, so that the technical problems that the fault early warning and troubleshooting control are not timely and accurate enough, the troubleshooting cost is relatively high and the working process of equipment is influenced in the fault diagnosis process of the grinding machine tool are solved; the technical effects that the fault early warning accuracy of the grinding machine tool is improved, the troubleshooting cost is reduced, and the reliability and maintenance efficiency of the machine tool are improved are achieved.

Description

Technical Field

[0001] The invention relates to the technical field of fault diagnosis, and in particular to a fault diagnosis method and system for a grinding machine tool. Background Art

[0002] Grinding machine tools are machine tools used for grinding workpiece surfaces. Due to their inseparable relationship with precision machining and assembly, parameter control and fault diagnosis of grinding machine tools have become important research topics. In addition, because grinding machine tools are complex equipment that integrates mechanics, hydraulics, and electrical functions, the occurrence of their failures is often reflected by the comprehensive combination of these three, making fault diagnosis complicated. With the improvement of the performance of modern CNC systems, their own failure rate has decreased, but the number of failures caused by non-system reasons has increased, such as problems with detection switches, hydraulic components, pneumatic components, electrical actuators, mechanical devices, etc., which have affected the control accuracy of grinding machine tools, reduced fault diagnosis efficiency, reduced equipment reliability, and product quality cannot be guaranteed, bringing new challenges to fault diagnosis.

[0003] In summary, in the fault diagnosis process of grinding machine tools, the existing technology has technical problems such as fault warning and troubleshooting control are not timely and accurate, the troubleshooting cost is high, and the working progress of the equipment is affected. Summary of the invention

[0004] The present application provides a fault diagnosis method and system for a grinding machine tool, which is used to solve the technical problems that the fault warning and troubleshooting control are not timely and accurate enough, the troubleshooting cost is high, and the working progress of the equipment is affected during the fault diagnosis process of the grinding machine tool.

[0005] In view of the above problems, the present application provides a fault diagnosis method and system for a grinding machine tool.

[0006] In a first aspect, the present application provides a fault diagnosis method for a grinding machine tool, the method being applied to a fault diagnosis system for a grinding machine tool, the method comprising: A data sharing platform connected to the grinding machine tool is used to collect a first historical fault explicit feature library and a first fault tracing database of a first subsystem of the grinding machine tool; the first historical fault explicit feature library is used as sample data to construct a first fault warning model and a warning sensor module of the first subsystem, and the modules are deployed to the first subsystem; a fault propagation model of the first subsystem is constructed based on the first fault tracing database to generate a first fault propagation model; a fault propagation probability is identified for the first fault propagation model based on the first fault tracing database to generate a first identification propagation model; the grinding machine tool is monitored in real time through the warning sensor module, and the monitoring data is input into the first fault warning model; when the first fault warning model outputs a fault warning signal, the fault warning signal is input into the first identification propagation model for fault propagation analysis, and a fault risk source sequence is output in descending order of fault propagation probability; and fault troubleshooting control is performed based on the fault risk source sequence.

[0007] In a second aspect, the present application provides a fault diagnosis system for a grinding machine tool, the system comprising: A data acquisition module is used to connect to the data sharing platform of the grinding machine tool to collect the first historical fault explicit feature library and the first fault tracing database of the first subsystem of the grinding machine tool; a first subsystem construction module is used to use the first historical fault explicit feature library as sample data to construct the first fault warning model and the warning sensor module of the first subsystem, and deploy them to the first subsystem; a first fault propagation model acquisition module is used to construct a fault propagation model for the first subsystem based on the first fault tracing database to generate a first fault propagation model; a probability identification module is used to identify the fault propagation probability of the first fault propagation model based on the first fault tracing database to generate a first identification propagation model; a real-time monitoring module is used to monitor the grinding machine tool in real time through the warning sensor module and input the monitoring data into the first fault warning model; a fault risk source sequence output module is used to input the fault warning signal into the first identification propagation model for fault propagation analysis when the first fault warning model outputs a fault warning signal, and output a fault risk source sequence in the order of fault propagation probability from large to small; a fault troubleshooting control module is used to perform fault troubleshooting control based on the fault risk source sequence.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: The present application provides a fault diagnosis method for a grinding machine tool. The method collects a first historical fault explicit feature library and a first fault tracing database of a first subsystem of the grinding machine tool by connecting to a data sharing platform of the grinding machine tool; uses the first historical fault explicit feature library as sample data to construct a first fault warning model and a warning sensor module of the first subsystem, and deploys them to the first subsystem; constructs a fault propagation model of the first subsystem based on the first fault tracing database to generate a first fault propagation model; identifies the fault propagation probability of the first fault propagation model based on the first fault tracing database to generate a first identification propagation model; and uses the warning sensor module to detect the fault. The grinding machine tool is monitored in real time, and the monitoring data is input into the first fault warning model; when the first fault warning model outputs a fault warning signal, the fault warning signal is input into the first identification propagation model for fault propagation analysis, and a fault risk source sequence is output in descending order of fault propagation probability; based on the fault risk source sequence, fault troubleshooting and control are performed, which solves the technical problems that the fault warning and troubleshooting control are not timely and accurate enough, the troubleshooting cost is high, and the working progress of the equipment is affected during the fault diagnosis process of the grinding machine tool, and achieves the technical effect of improving the fault warning accuracy of the grinding machine tool, reducing the fault troubleshooting cost, and improving the reliability and maintenance efficiency of the machine tool. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic flow chart of a fault diagnosis method for a grinding machine tool is provided for the present application.

[0010] Figure 2 A schematic diagram of the structure of a fault diagnosis system for a grinding machine tool is provided for this application.

[0011] Explanation of the accompanying drawings: data acquisition module 11, first subsystem construction module 12, first fault propagation model acquisition module 13, probability identification module 14, real-time monitoring module 15, fault risk source sequence output module 16, fault troubleshooting control module 17. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0013] Embodiment 1 A grinding machine is a machine tool that uses abrasive tools (such as grinding wheels, abrasive belts, etc.) to grind the surface of a workpiece. According to different processing methods and uses, grinding machine tools can be divided into many types, such as surface grinders, cylindrical grinders, internal cylindrical grinders, centerless grinders, tool grinders, etc. Grinding machine tools are mainly composed of a bed, a workbench, a grinding wheel rack, a feeding mechanism, a grinding fluid system, an electrical control system, and other parts. Among them, the bed is the basis of the machine tool, the workbench is used to install the workpiece, the grinding wheel rack is used to install the grinding wheel, and the feeding mechanism is used to achieve relative movement between the workpiece and the grinding wheel. Grinding machine tools are widely used in machinery manufacturing, automobile manufacturing, aerospace, mold manufacturing and other fields. They have the characteristics of high precision, high surface quality, wide adaptability and high efficiency. Therefore, in order to ensure the normal operation and processing quality of the machine tool, timely fault diagnosis and analysis have become a long-lasting topic.

[0014] like Figure 1 As shown, the present application provides a fault diagnosis method for a grinding machine tool, the method is applied to a fault diagnosis system for a grinding machine tool, the method comprising: Step S100: connecting to a data sharing platform of a grinding machine tool, and collecting a first historical fault explicit feature library and a first fault tracing database of a first subsystem of the grinding machine tool.

[0015] Specifically, the data sharing platform of the grinding machine tool is a software tool or service designed for grinding machine tool data management and collaboration. By establishing a stable and reliable data connection channel, the grinding machine tool fault diagnosis system is connected to the grinding machine tool data sharing platform, laying the foundation for subsequent data collection and analysis. After connecting to the data sharing platform, the relevant characteristic data of the first subsystem of the grinding machine tool is collected. Since the machine tool is composed of multiple systems, such as mechanical system, cooling system, control system, electrical system, hydraulic system, etc., the first subsystem here refers to any system in the multiple system composition.

[0016] Next, the historical fault explicit feature library and fault tracing database in the corresponding subsystem are obtained to obtain the first historical fault explicit feature library and the first fault tracing database. The explicit feature library refers to a database that records the characteristics and phenomena that can be directly observed when the machine tool fails. These characteristics may include changes in physical quantities such as abnormal sound, vibration, temperature, pressure, or changes in the running state and operating parameters of the machine tool. By connecting to the data sharing platform, these explicit features can be extracted from the historical data of the machine tool and built into a library for subsequent analysis and modeling. The fault tracing database is a database that records the root causes and related data that cause the machine tool to fail. These data may include design defects, manufacturing process problems, use environment conditions, operating errors and other factors of the machine tool. Through the analysis of these data, the occurrence mechanism and propagation path of the fault can be deeply understood, providing strong support for subsequent fault warning and troubleshooting. Similarly, by connecting to the data sharing platform, these traceability data can be extracted from the historical data of the machine tool and built into a library.

[0017] By connecting to the data sharing platform to obtain the historical fault explicit feature library and fault tracing database, we can not only quickly locate the cause of the fault, shorten the troubleshooting time, and improve maintenance efficiency, but also optimize the production process and equipment parameter settings through data sharing and collaboration, thereby improving the overall work efficiency of the equipment and the quality of subsequent products.

[0018] Step S200: Using the first historical fault explicit feature library as sample data, construct a first fault warning model and a warning sensor module for the first subsystem, and deploy them to the first subsystem.

[0019] Optionally, key feature data is extracted from the first historical fault explicit feature library, and the extracted feature data should be able to accurately reflect the state of the first subsystem when it fails. Select a suitable machine learning or deep learning algorithm, such as support vector machine (SVM), random forest, neural network, etc., to build a first fault warning model. Use the extracted historical fault data as a training set to train and adjust the model so that it can accurately identify the fault mode of the first subsystem. At the same time, design a suitable warning sensor module according to the characteristics of the first subsystem and the requirements of the fault warning model. The warning sensor module should be able to monitor the operating status of the first subsystem in real time and collect key data for the model to analyze. The module design should consider factors such as accuracy, reliability, cost, and convenience of installation and maintenance. Evaluate the performance and generalization ability of the model through cross-validation and other technologies to ensure the effectiveness and reliability of the model. Integrate the constructed first fault warning model and warning sensor module into the first subsystem, configure the necessary hardware and software environment in the subsystem, and ensure that the model and module can operate normally. Test the system to verify the accuracy and reliability of the warning model to ensure that the system can detect and warn potential faults in a timely manner.

[0020] By constructing the first fault warning model and warning sensor module based on the first historical fault explicit feature library and successfully deploying them to the corresponding subsystem, real-time monitoring and early warning of faults can be achieved, the reliability and operation efficiency of equipment can be improved, and maintenance costs and production risks can be reduced.

[0021] Step S300: constructing a fault propagation model for the first subsystem based on the first fault tracing database to generate a first fault propagation model.

[0022] Exemplarily, fault data related to the first subsystem is extracted from the first fault tracing database. These data should include the time, location, type, cause of the fault, and related operation and maintenance records. Analyze the extracted fault data, identify the propagation path and mode of the fault within the subsystem, and conduct a comprehensive analysis of the system status, operation records, maintenance records, etc. before and after the fault occurs. Determine the key nodes and paths of fault propagation, which may be a component, interface or operation process in the subsystem, so as to obtain the final analysis results. Next, according to the results of the fault propagation analysis, select a suitable model construction method, such as graph theory model, fault tree analysis (FTA), failure mode and effects analysis (FMEA), etc.

[0023] The first fault propagation model is constructed using the selected method, which can clearly represent the propagation path, impact range and possible consequences of the fault within the subsystem. In addition, factors such as the probability of the fault, the degree of impact and the propagation speed need to be considered when constructing the model in order to more accurately predict the impact and consequences of the fault.

[0024] By building the first fault propagation model, we can better understand and predict the propagation path and impact range of faults within the subsystem, providing strong support for system fault warning, diagnosis and maintenance. At the same time, this also helps to improve the reliability and maintenance efficiency of the system and reduce the impact of faults on production operations.

[0025] Step S400: Based on the first fault tracing database, the first fault propagation model is labeled with a fault propagation probability to generate a first labeled propagation model.

[0026] Furthermore, in order to quantify the possibility of fault propagation within the subsystem, the probability of fault propagation is identified. Fault data related to the first fault propagation model is extracted from the first fault tracing database, and the frequency of fault occurrence, historical records, failure rates of related components and other data are analyzed to understand the statistical characteristics and laws of fault propagation. According to the existing first fault propagation model, the paths along which the fault may propagate within the subsystem are determined. These paths represent the potential ways in which the fault propagates from one component to another. For each fault propagation path, the probability of the fault propagating from the source component to the target component is calculated based on the historical data in the fault tracing database. Specifically, this can be achieved by counting the number of times the target component fails when the source component fails and dividing it by the total number of source component failures. If the historical data is not sufficient to directly calculate the probability, statistical models (such as Bayesian networks, Markov models, etc.) can be used to estimate these probabilities. These models can use known failure rates and conditional probabilities to infer unknown fault propagation probabilities. Finally, the calculated fault propagation probability is identified on the first fault propagation model. The identification method can be achieved by adding probability values ​​to the connections between each component in the model. Next, new or additional fault data is used to verify the first identification propagation model that identifies the fault propagation probability. After verification and optimization, the final first identification propagation model is generated. This model not only includes the propagation path of the fault within the subsystem, but also includes the fault propagation probability on each path, thereby more comprehensively reflecting the fault characteristics of the subsystem.

[0027] Step S500: monitoring the grinding machine tool in real time through the early warning sensor module, and inputting the monitoring data into the first fault early warning model.

[0028] Optionally, an early warning sensor module is installed at the key parts of the grinding machine (such as the motor, bearing, transmission system, etc.), and the configured sensor module is used to collect key data related to the operating status of the machine, such as temperature, vibration, sound, current, etc. During the real-time monitoring process, the early warning sensor module starts to work and collects the operating data of the grinding machine in real time. The sensor module transmits this data to the data collection system or central processing unit at a certain frequency (such as every second, every minute, etc.). Then, the collected raw data is cleaned, filtered and standardized to eliminate noise and outliers. The processed data is formatted into a form suitable for input into the first fault warning model. Finally, the pre-processed monitoring data is input into the first fault warning model, and the input data is analyzed and potential faults are predicted.

[0029] Through the above steps, preventive maintenance of grinding machine tools can be achieved, potential failure risks can be identified in advance, and corresponding measures can be taken to reduce the probability and impact of failures, thereby improving production efficiency and equipment utilization.

[0030] Step S600: When the first fault warning model outputs a fault warning signal, the fault warning signal is input into the first identification propagation model for fault propagation analysis, and a fault risk source sequence is output in descending order of fault propagation probability.

[0031] Step S700: performing fault troubleshooting control based on the fault risk source sequence.

[0032] Specifically, in the actual fault warning process, when the first fault warning model detects a potential fault risk, it will output a fault warning signal, which usually contains information about the location, type, and possible impact of the potential fault. The key information (such as the fault location) in the fault warning signal is passed as input to the first identification propagation model, and the model will perform fault propagation analysis based on this information and the previously identified fault propagation probability. The first identification propagation model simulates the propagation path of the fault in the system according to the input fault warning signal. The model will consider all possible propagation paths and calculate the fault risk of each component or part according to the fault propagation probability on each path. Further, all components or parts are sorted according to the size of the fault risk based on the calculated fault risk. The sorting is based on the fault propagation probability, which is arranged from high to low, thereby forming a fault risk source sequence. Finally, the first identification propagation model outputs the sorted fault risk source sequence as the output result, which can help maintenance personnel quickly locate the components or parts that are most likely to be the source of the fault, so as to give priority to inspection and maintenance.

[0033] When performing troubleshooting control, carefully review the sequence of fault risk sources, understand which components, parts or systems have a higher risk of failure, and prioritize troubleshooting of higher-ranked fault risk sources to ensure that the most critical potential problems are resolved first. Develop a troubleshooting plan and check potential fault risk sources one by one according to the troubleshooting plan.

[0034] Through the above steps, the potential fault risk sources in the grinding machine can be identified more accurately, and corresponding measures can be taken for prevention and maintenance, thereby improving the reliability and operating efficiency of the equipment. At the same time, by performing troubleshooting control based on the fault risk source sequence, the components with a higher probability can be repaired first, and potential fault problems can be identified and solved more efficiently, reducing the impact of faults on production operations and improving the stability of the system.

[0035] Furthermore, based on the first fault tracing database, a fault propagation model is constructed for the first subsystem to generate a first fault propagation model. Step S300 of the present application further includes: Step S310: Perform fault tracing connection based on the first fault tracing database to generate multiple fault tracing node chains, wherein the starting node of the multiple fault tracing node chains is the first subsystem.

[0036] Step S320: Perform multi-level connections based on the multiple fault tracing node chains to generate a multi-level fault propagation direction diagram.

[0037] Step S330: performing matrix transformation on the multi-level fault propagation direction diagram to generate the first fault propagation model.

[0038] Optionally, consult the first fault tracing database, which contains the history of faults, causes, impacts, and correlation information between them. Starting from the fault record of the first subsystem, look for related fault sources, which can be upstream systems, components, equipment, or other factors. Connect these fault sources to the first subsystem to form a fault tracing node chain. Among them, the first subsystem is the starting node and the fault source is the subsequent node on the chain. Repeat this process until all relevant fault sources are connected to form multiple fault tracing node chains.

[0039] Then, the fault tracing node chains are analyzed to identify the common nodes and branches between them, and the chains with common nodes are merged or cross-connected to build a more complex fault propagation network. In this network, nodes at different levels represent different levels of fault sources or impact points, and the connections between them represent the path and direction of fault propagation. Finally, a multi-level fault propagation direction diagram is formed, which intuitively shows all the paths and nodes that the fault may propagate to starting from the first subsystem.

[0040] In order to facilitate the quantitative analysis and calculation of faults, the multi-level fault propagation directional diagram is converted into a mathematical matrix form. In the matrix, rows and columns represent different nodes (such as systems, components, equipment, etc.), and the matrix elements represent the probability or impact of fault propagation between nodes. These probabilities or impacts can be determined based on historical data, expert experience or statistical analysis. The final generated matrix is ​​the first fault propagation model, which can be used as a basis for predicting the possibility of faults, evaluating the impact of faults on the system, and formulating fault prevention and response measures.

[0041] Through the above steps, a fault propagation model for the first subsystem can be constructed, which can better understand the propagation law and impact range of faults in the system, so as to more effectively prevent, monitor and manage faults.

[0042] Furthermore, based on the multiple fault tracing node chains being connected at multiple levels to generate a multi-level fault propagation direction diagram, step S320 of the present application further includes: Step S321: performing node propagation depth alignment on the multiple fault tracing node chains.

[0043] Step S322: Connect and integrate the multiple fault tracing node chains based on the node propagation depth alignment result to generate the multi-level fault propagation direction map.

[0044] Exemplarily, since there may be multiple second-level nodes starting from the first subsystem, for example, there may be multiple reasons that directly cause the failure of the first subsystem, it is necessary to align the propagation depth of the nodes. Specifically, first select a reference node, which is the starting node of the fault propagation, that is, the first subsystem. Next, the propagation depth of the nodes on each fault tracing node chain is calculated starting from the reference node. The propagation depth represents the number of nodes passed through on the fault propagation path from the reference node to the current node (excluding the reference node itself). Align the nodes corresponding to the same propagation depth on all fault tracing node chains. If some chains have no nodes at a specific depth, an empty node or placeholder can be inserted to maintain consistency. After the node propagation depth is aligned, it is possible to clearly identify which nodes are common nodes on different fault tracing node chains, and use lines or arrows to connect common nodes with the same propagation depth to represent the fault propagation relationship between them. Arrange nodes of different propagation depths in layers to form a multi-level fault propagation structure. High-level nodes represent the fault source further upstream, and low-level nodes represent the impact or symptoms downstream. In the multi-level fault propagation direction map, additional information such as fault probability, propagation time, fault type, etc. can be added to more comprehensively describe the characteristics and impact of fault propagation. After the above steps, a multi-level fault propagation direction map will be generated. This map can intuitively show the process of fault propagation from the first subsystem to other parts of the system through different paths and nodes, providing strong support for fault management and system reliability analysis.

[0045] Furthermore, based on the first fault tracing database, the first fault propagation model is marked with a fault propagation probability to generate a first marked propagation model. Step S400 of the present application further includes: Step S410: obtaining first-level tracing nodes, second-level tracing nodes, and even N-th-level tracing nodes having a front-and-back fault induction relationship in the first fault propagation model, wherein the first-level tracing node is the first subsystem.

[0046] Step S420: Based on the first fault tracing database, the fault induction probability is calculated for the first-level tracing nodes, the second-level tracing nodes, and up to the N-th level tracing nodes to generate the first-level node fault propagation probability, the second-level node fault propagation probability, and up to the N-th level node fault propagation probability.

[0047] Step S430: Identify the first-level node fault propagation probability, the second-level node fault propagation probability, and even the N-th-level node fault propagation probability into the first fault propagation model to generate the first identified propagation model.

[0048] Further, the traceability nodes of each level in the first fault propagation model are obtained. These nodes are arranged in the order of fault propagation, starting from the first level (i.e., the subsystem or component where the fault originates) to the Nth level (i.e., the end node of the fault propagation chain). In this process, the first-level traceability node is clearly defined as the first subsystem, which is the starting point of fault propagation. Then, the fault induction probability of each level of traceability nodes is calculated based on the first fault tracing database. The fault induction probability describes the possibility that when the current node fails, its downstream node will also fail. In the calculation process, it is necessary to refer to the historical data, expert knowledge, and possible fault propagation mechanisms in the fault tracing database. These data can accurately estimate the fault induction probability of each node under specific conditions. Finally, a list of node fault propagation probabilities from the first level to the Nth level will be generated, including the first-level node fault propagation probability, the second-level node fault propagation probability, and the Nth-level node fault propagation probability.

[0049] After obtaining the fault propagation probability of nodes at each level, these probability values ​​are marked on the first fault propagation model. The marking method can be to directly mark the probability value next to the node in the model, or to indicate the probability through visual elements such as color and size. Through this step, the original fault propagation model is converted into a model with fault propagation probability marking, that is, the first identification propagation model. Finally, the fault propagation model with the fault propagation probability marked is sorted and optimized to ensure the accuracy and readability of the information. The first identification propagation model can not only clearly show the propagation path of the fault within the system, but also quantitatively describe the influence of each node in the fault propagation process, providing strong support for subsequent fault prevention, diagnosis and response.

[0050] Further, based on the first fault tracing database, the fault induction probability is calculated for the first-level tracing node, the second-level tracing node, and the N-th level tracing node to generate the first-level node fault propagation probability, the second-level node fault propagation probability, and the N-th level node fault propagation probability. Step S420 of the present application also includes: Step S421: Initialize the first-level node fault propagation probability of the first-level traceability node to 1.

[0051] Step S422: extract the first fault record of the first-level tracing node in the first fault tracing database, calculate the proportion coefficient of the actual fault cause of the second-level tracing node in the first fault record, and generate the fault propagation probability of the second-level node.

[0052] Step S423: extract the second fault record of the second-level tracing node in the first fault tracing database, calculate the proportion coefficient of the third-level tracing node in the second fault record whose actual fault cause is the fault propagation probability of the second-level node, and weight it with the fault propagation probability of the second-level node to generate the fault propagation probability of the third-level node.

[0053] Step S424: and so on, continue to obtain the N-th level node fault propagation probability.

[0054] Specifically, the fault propagation probability of the first-level traceability node (i.e., the first subsystem or the starting fault node) is initialized to 1. This is because as the starting point of fault propagation, when the first-level node fails, its fault propagation probability is 100%, that is, it will definitely happen. Next, the first fault records related to the first-level traceability node are extracted from the first fault traceability database, and these fault records are analyzed to count the number of times the actual cause of the fault is the second-level traceability node, and calculate its proportion coefficient. This proportion coefficient reflects the probability that the second-level node is the actual cause when the first-level node fails. The calculated proportion coefficient is used as the fault propagation probability of the second-level node.

[0055] Furthermore, the second fault records related to the second-level traceability nodes are extracted from the first fault tracing database, and these fault records are analyzed to count the number of times the actual cause of the fault is the third-level traceability node, and calculate its proportion coefficient. Since the occurrence of the third-level node fault is affected by the second-level node, it is necessary to perform a weighted calculation on this proportion coefficient and the second-level node fault propagation probability. The weighting method can be selected according to the actual situation, such as simple multiplication or a more complex probability model. The result obtained after the weighted calculation is the third-level node fault propagation probability.

[0056] For nodes at higher levels (fourth level, fifth level, etc. up to level N), repeat the above process, extract the corresponding fault records, calculate the proportion coefficient, and perform weighted calculation. Each time the calculation is performed, the fault propagation probability of the node at the previous level needs to be used as the weight. Finally, the calculated fault propagation probabilities of nodes at each level are sorted out to generate a list of fault propagation probabilities from the first level to the Nth level. This list will serve as an important parameter of the fault propagation model for subsequent fault analysis and prediction.

[0057] Through the above method, the fault propagation probability of each level of tracing nodes can be accurately calculated based on the first fault tracing database, providing strong support for fault tracing and fault prevention.

[0058] Furthermore, step S430 of the present application also includes: Step S431: Determine whether there are multiple induced relationship nodes among the first-level tracing nodes, the second-level tracing nodes, and even the N-th-level tracing nodes.

[0059] Step S432: If yes, perform maximum value extraction on the node fault propagation probability of the multi-induction relationship nodes, and identify the first fault propagation model with the maximum value extraction result.

[0060] Optionally, in the fault propagation model, sometimes a faulty node may be induced by multiple upstream nodes, which is called a multi-induced relationship node. In order to more accurately represent this relationship, it is necessary to specially handle the fault propagation probability of nodes with multi-induced relationships. Specifically, traverse the first-level traceability nodes to the Nth-level traceability nodes, and check whether each node has multiple upstream nodes inducing its failure. This can usually be achieved by checking the connection relationship in the fault propagation model. If a node has two or more upstream nodes connected to it, and these upstream nodes may cause the node to fail, then the node is regarded as a multi-induced relationship node. For each multi-induced relationship node, it is necessary to calculate the maximum value of the fault propagation probability of all its upstream nodes. This is because in actual situations, when multiple upstream nodes fail at the same time, usually only one upstream node's failure will directly cause the node to fail, and the failures of other upstream nodes may only play an auxiliary role. The maximum value can be extracted by comparing the fault propagation probabilities of all upstream nodes and selecting the largest one as the fault propagation probability of the node, which means that among multiple upstream nodes, the node is most likely to be induced by the node with the largest fault propagation probability. Finally, the calculated maximum value extraction result is identified on the corresponding multi-induced relationship node in the first fault propagation model.

[0061] Through the above steps, the first fault propagation model can be updated to a model that more accurately reflects the fault propagation probability of multi-induced relationship nodes, which helps to more accurately describe the propagation of faults in the system, especially in the presence of multiple potential causes. By extracting the maximum value as the fault propagation probability of multi-induced relationship nodes, we can better understand which upstream nodes play a key role in fault propagation, thereby formulating more effective fault prevention and response measures to ensure the accuracy of fault diagnosis.

[0062] Furthermore, the fault warning signal is input into the first identification propagation model for fault propagation analysis, and a fault risk source sequence is output in descending order of fault propagation probability. Step S600 of the present application further includes: Step S610: performing fault impact analysis on the multi-level nodes in the first identification propagation model to generate multi-level node fault impact indicators.

[0063] Step S620: combining the multi-level node fault impact index and the fault propagation probability, and outputting the fault risk source sequence.

[0064] Specifically, the received fault warning signal is input into the first identification propagation model. This signal usually indicates that an abnormality or potential fault has occurred at a certain starting point (such as the first-level traceability node). The fault propagation probability in the first identification propagation model is used to analyze the paths and nodes through which the fault propagates from the starting point to other parts of the system. In this process, special attention is paid to those paths and nodes with higher fault propagation probabilities, because these paths and nodes are more likely to cause system-level failures or impacts. Fault impact analysis is performed on the multi-level nodes in the first identification propagation model to evaluate the impact of each node on the overall function, performance or security of the system when a failure occurs. Fault impact analysis can consider multiple aspects, such as the importance of the node, the failure frequency of the node, and the chain reaction of the failure to other parts of the system. Based on the analysis results, a fault impact index is generated for each node to quantify the degree of impact on the system when the node fails.

[0065] Furthermore, the fault impact index of each node is combined with its fault propagation probability in the fault propagation model to obtain a comprehensive fault risk assessment value, which takes into account both the importance of the node in fault propagation and the impact of the node failure on the system as a whole. All nodes are sorted according to the comprehensive fault risk assessment value to generate a fault risk source sequence. The nodes in the sequence are arranged from high to low according to the fault risk, that is, the nodes that are most likely to cause system-level failures or the greatest impact are ranked first. This fault risk source sequence helps to quickly locate key risk points and take corresponding preventive and response measures.

[0066] Through the technical solutions of the above-mentioned embodiments, a fault diagnosis method for a grinding machine tool provided in the present application solves the technical problems that the fault warning and troubleshooting control are not timely and accurate enough, the troubleshooting cost is high, and the working progress of the equipment is affected during the fault diagnosis process of the grinding machine tool. The technical effect of improving the fault warning accuracy of the grinding machine tool, reducing the fault troubleshooting cost, and improving the reliability and maintenance efficiency of the machine tool is achieved.

[0067] Embodiment 2 Based on the same inventive concept as the fault diagnosis method for a grinding machine tool in the aforementioned embodiment, Figure 2 As shown, the present application provides a fault diagnosis system for a grinding machine tool, the system comprising: The data acquisition module 11 is used to connect to the data sharing platform of the grinding machine tool and acquire a first historical fault explicit feature library and a first fault tracing database of a first subsystem of the grinding machine tool.

[0068] The first subsystem construction module 12 is used to construct a first fault warning model and a warning sensor module of the first subsystem using the first historical fault explicit feature library as sample data, and deploy them to the first subsystem.

[0069] The first fault propagation model obtaining module 13 is used to construct a fault propagation model for the first subsystem based on the first fault tracing database to generate a first fault propagation model.

[0070] The probability identification module 14 is used to perform fault propagation probability identification on the first fault propagation model based on the first fault tracing database to generate a first identified propagation model.

[0071] The real-time monitoring module 15 is used to monitor the grinding machine tool in real time through the early warning sensor module and input the monitoring data into the first fault early warning model.

[0072] The fault risk source sequence output module 16 is used to input the fault warning signal into the first identification propagation model for fault propagation analysis when the first fault warning model outputs a fault warning signal, and output a fault risk source sequence in the order of fault propagation probability from large to small.

[0073] The fault detection control module 17 is used to perform fault detection control based on the fault risk source sequence.

[0074] Furthermore, the first fault propagation model obtaining module 13 is further configured to perform the following steps: A fault tracing connection is performed based on the first fault tracing database to generate a plurality of fault tracing node chains, wherein the starting node of the plurality of fault tracing node chains is the first subsystem.

[0075] Based on the multiple fault tracing node chains, multi-level connections are made to generate a multi-level fault propagation direction diagram.

[0076] Performing matrix transformation on the multi-level fault propagation direction diagram to generate the first fault propagation model.

[0077] Furthermore, the first fault propagation model obtaining module 13 is further configured to perform the following steps: Perform node propagation depth alignment on the multiple fault tracing node chains.

[0078] The multiple fault tracing node chains are connected and integrated based on the node propagation depth alignment result to generate the multi-level fault propagation direction map.

[0079] Furthermore, the probability identification module 14 is further configured to perform the following steps: Obtain first-level tracing nodes, second-level tracing nodes, and even N-level tracing nodes having a front-and-back fault induction relationship in the first fault propagation model, wherein the first-level tracing node is the first subsystem.

[0080] Based on the first fault tracing database, the fault induction probability is calculated for the first-level tracing nodes, the second-level tracing nodes, and up to the N-th level tracing nodes to generate the first-level node fault propagation probability, the second-level node fault propagation probability, and up to the N-th level node fault propagation probability.

[0081] The first-level node fault propagation probability, the second-level node fault propagation probability, and even the N-th-level node fault propagation probability are identified in the first fault propagation model to generate the first identified propagation model.

[0082] Furthermore, the probability identification module 14 is further configured to perform the following steps: Initialize the first-level node fault propagation probability of the first-level traceability node to 1.

[0083] The first fault record of the first-level tracing node is extracted from the first fault tracing database, and the proportion coefficient of the actual fault cause of the second-level tracing node in the first fault record is calculated to generate the second-level node fault propagation probability.

[0084] The second fault record of the second-level tracing node is extracted from the first fault tracing database, and the proportion coefficient of the actual fault cause of the third-level tracing node in the second fault record is calculated, and the coefficient is weighted with the fault propagation probability of the second-level node to generate the fault propagation probability of the third-level node.

[0085] And so on, continue to obtain the fault propagation probability of nodes at the Nth level.

[0086] Furthermore, the probability identification module 14 is further configured to perform the following steps: Determine whether there are multiple induced relationship nodes among the first-level tracing nodes, the second-level tracing nodes, and even the N-th-level tracing nodes.

[0087] If so, a maximum value extraction is performed on the node fault propagation probability of the multi-induction relationship nodes, and the first fault propagation model is identified with the maximum value extraction result.

[0088] Furthermore, the fault risk source sequence output module 16 is further configured to perform the following steps: Perform a fault impact analysis on the multi-level nodes in the first identification propagation model to generate a multi-level node fault impact index.

[0089] The fault risk source sequence is output by combining the multi-level node fault impact index and the fault propagation probability.

[0090] Through the above-mentioned detailed description of a fault diagnosis method for a grinding machine tool in this specification, those skilled in the art can clearly understand a fault diagnosis system for a grinding machine tool in this embodiment. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0091] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A fault diagnosis method for a grinding machine tool, characterized in that: include: Connecting to the data sharing platform of the grinding machine tool, collecting a first historical fault explicit feature library and a first fault tracing database of a first subsystem of the grinding machine tool; Using the first historical fault explicit feature library as sample data, construct a first fault warning model and a warning sensor module of the first subsystem, and deploy them to the first subsystem; Constructing a fault propagation model for the first subsystem based on the first fault tracing database to generate a first fault propagation model; Based on the first fault tracing database, the first fault propagation model is marked with a fault propagation probability to generate a first marked propagation model; Performing real-time monitoring of the grinding machine tool through the early warning sensor module, and inputting the monitoring data into the first fault early warning model; When the first fault warning model outputs a fault warning signal, the fault warning signal is input into the first identification propagation model to perform fault propagation analysis, and a fault risk source sequence is output in descending order of fault propagation probability; Based on the fault risk source sequence, fault troubleshooting control is performed.

2. A fault diagnosis method for a grinding machine tool as claimed in claim 1, characterized in that: Constructing a fault propagation model for the first subsystem based on the first fault tracing database to generate a first fault propagation model includes: Performing fault tracing connection based on the first fault tracing database to generate multiple fault tracing node chains, wherein the starting node of the multiple fault tracing node chains is the first subsystem; Based on the multiple fault tracing node chains, multi-level connections are made to generate a multi-level fault propagation direction diagram; Performing matrix transformation on the multi-level fault propagation direction diagram to generate the first fault propagation model.

3. A fault diagnosis method for a grinding machine tool as claimed in claim 2, characterized in that: Based on the multiple fault tracing node chains, multi-level connections are made to generate a multi-level fault propagation direction diagram, including: Performing node propagation depth alignment on the multiple fault tracing node chains; The multiple fault tracing node chains are connected and integrated based on the node propagation depth alignment result to generate the multi-level fault propagation direction map.

4. A fault diagnosis method for a grinding machine tool as claimed in claim 1, characterized in that: Performing fault propagation probability identification on the first fault propagation model based on the first fault tracing database to generate a first identification propagation model includes: Obtaining a first-level tracing node, a second-level tracing node, and an N-th-level tracing node having a front-and-back fault induction relationship in the first fault propagation model, wherein the first-level tracing node is the first subsystem; Based on the first fault tracing database, the fault induction probability of the first-level tracing nodes, the second-level tracing nodes, and even the N-th-level tracing nodes is calculated to generate the first-level node fault propagation probability, the second-level node fault propagation probability, and even the N-level node fault propagation probability; The first-level node fault propagation probability, the second-level node fault propagation probability, and even the N-th-level node fault propagation probability are identified in the first fault propagation model to generate the first identified propagation model.

5. A fault diagnosis method for a grinding machine tool as claimed in claim 4, characterized in that: Based on the first fault tracing database, the fault induction probability of the first-level tracing nodes, the second-level tracing nodes, and even the N-level tracing nodes is calculated to generate the first-level node fault propagation probability, the second-level node fault propagation probability, and even the N-level node fault propagation probability, including: Initialize the first-level node fault propagation probability of the first-level traceability node to 1; Extract the first fault record of the first-level tracing node in the first fault tracing database, calculate the proportion coefficient of the actual cause of the fault in the first fault record being the second-level tracing node, and generate the second-level node fault propagation probability; Extract the second fault record of the second-level tracing node in the first fault tracing database, calculate the proportion coefficient of the third-level tracing node whose actual fault cause in the second fault record is a fault tracing node, and weight it with the second-level node fault propagation probability to generate the third-level node fault propagation probability; And so on, continue to obtain the fault propagation probability of nodes at the Nth level.

6. A fault diagnosis method for a grinding machine tool as claimed in claim 5, characterized in that: Also includes: Determine whether there are multiple induced relationship nodes among the first-level traceability nodes, the second-level traceability nodes, and even the N-th-level traceability nodes; If so, a maximum value extraction is performed on the node fault propagation probability of the multi-induction relationship nodes, and the first fault propagation model is identified with the maximum value extraction result.

7. A fault diagnosis method for a grinding machine tool as claimed in claim 1, characterized in that: The fault warning signal is input into the first identification propagation model to perform fault propagation analysis, and a fault risk source sequence is output in descending order of fault propagation probability, further comprising: Performing fault impact analysis on the multi-level nodes in the first identification propagation model to generate a multi-level node fault impact index; The fault risk source sequence is output by combining the multi-level node fault impact index and the fault propagation probability.

8. A fault diagnosis system for a grinding machine tool, characterized in that: For implementing the method according to any one of claims 1 to 7, the system comprises: A data acquisition module, used for connecting to a data sharing platform of a grinding machine tool, and acquiring a first historical fault explicit feature library and a first fault tracing database of a first subsystem of the grinding machine tool; A first subsystem construction module, used to construct a first fault warning model and a warning sensor module of the first subsystem using the first historical fault explicit feature library as sample data, and deploy them to the first subsystem; A first fault propagation model obtaining module, configured to construct a fault propagation model for the first subsystem based on the first fault tracing database to generate a first fault propagation model; a probability identification module, configured to perform fault propagation probability identification on the first fault propagation model based on the first fault tracing database, and generate a first identification propagation model; A real-time monitoring module, used for real-time monitoring of the grinding machine through the early warning sensor module, and inputting monitoring data into the first fault early warning model; a fault risk source sequence output module, configured to, when the first fault warning model outputs a fault warning signal, input the fault warning signal into the first identification propagation model for fault propagation analysis, and output a fault risk source sequence in descending order of fault propagation probability; The fault detection control module is used to perform fault detection control based on the fault risk source sequence.

Images