Special equipment operation fault detection method and device

Abstract

The invention discloses a special equipment operation fault detection method and device, and relates to the related field of data processing, and the method comprises the steps: carrying out the dynamic sensing of special equipment, obtaining an operation dynamic data set of the special equipment, and synchronizing the operation dynamic data set to a long-short-term memory network, determining a plurality of fault influence factors, performing fault joint detection in combination with the equipment use scene parameter set, generating a first fault detection result, activating a fault probabilistic reasoning model, and performing fault reasoning in combination with the operation dynamic data set, and generating a second fault detection result and performing automatic comprehensive analysis and detection on the operation fault of the special equipment in combination with the first fault detection result. The technical problems of low detection efficiency and high maintenance cost caused by difficulty in processing large-scale data sets and real-time monitoring requirements in traditional fault detection are solved, an intelligent and automatic fault detection method is introduced, the accuracy and timeliness of fault detection are improved, and the technical effect of reducing risks and losses caused by equipment operation faults is achieved.

Description

Technical Field

[0001] This application relates to the field of data processing, and in particular, to a method and device for detecting operation faults of special equipment. Background Art

[0002] Special equipment, such as elevators, plays a crucial role in industrial production and daily life. Its safety and operation stability directly affect production efficiency and personal safety. However, there are still many deficiencies in the existing special equipment operation fault detection technologies. Especially in complex and changeable working environments, the limitations of traditional detection technologies are also manifested in the lack of intelligent analysis ability for equipment operation data. It cannot make real-time fault prediction and analysis based on the operation status, historical data, and environmental parameters of the equipment, and can only perform reactive maintenance after a fault occurs.

[0003] Therefore, fault detection in the prior art is difficult to handle large-scale data sets and real-time monitoring requirements, resulting in technical problems such as low detection efficiency and high maintenance costs. Summary of the Invention

[0004] This application provides a method and device for detecting operation faults of special equipment, solves the technical problems that traditional fault detection is difficult to handle large-scale data sets and real-time monitoring requirements, resulting in low detection efficiency and high maintenance costs, realizes the introduction of intelligent and automated fault detection methods, improves the accuracy and timeliness of fault detection, and achieves the technical effect of reducing the risks and losses brought by equipment operation faults.

[0005] This application provides a method for detecting operation faults of special equipment. The method is applied to a device for detecting operation faults of special equipment and includes: dynamically sensing a special equipment according to the operation time series of the special equipment to obtain an operation dynamic data set of the special equipment; constructing a long short-term memory network, synchronizing the operation dynamic data set to the long short-term memory network, and determining a plurality of fault impact factors; introducing a device usage scenario parameter set, and performing joint fault detection by combining the plurality of fault impact factors with the device usage scenario parameter set to generate a first fault detection result, and activating a fault probability inference model according to the first fault detection result; performing fault inference by combining the operation dynamic data set through the fault probability inference model to generate a second fault detection result; and performing automated comprehensive analysis and detection of the operation faults of the special equipment based on the first fault detection result and the second fault detection result.

[0006] In a possible implementation, synchronize the operation dynamic data set to the long short-term memory network, determine multiple fault impact factors, and perform the following processing: slice the operation dynamic data set according to the operation cycle of the special equipment to determine multiple time windows, where the multiple time windows contain multiple operation data; perform state prediction on the operation dynamic data set based on the multiple time windows to generate multiple operation prediction values; compare the multiple operation prediction values with the multiple operation data, capture fluctuations according to the data comparison results, and obtain multiple fluctuation data streams; traverse the multiple fluctuation data streams for anomaly detection to identify multiple fault signals; analyze the impact characteristics of the multiple fault signals to determine the multiple fault impact factors.

[0007] In a possible implementation, analyze the impact characteristics of the multiple fault signals to determine the multiple fault impact factors, and perform the following processing: extract features from the multiple fault signals based on the operation time series of the special equipment to determine multiple operation features; perform impact analysis according to the multiple operation features to obtain an operation fault impact coefficient, where the operation fault impact coefficient has fault correlation; perform clustering analysis on the multiple fault signals according to the operation fault impact coefficient, and combine the clustering analysis results with the fault correlation for identification to determine the multiple fault impact factors.

[0008] In a possible implementation, introduce a device usage scenario parameter set, perform joint fault detection by combining the multiple fault impact factors with the device usage scenario parameter set to generate a first fault detection result, and perform the following processing: construct the device usage scenario parameter set based on the operation external environment and operation operation environment of the special equipment; extract the first device usage scenario parameter based on the device usage scenario parameter set, allocate weights to the multiple fault impact factors according to the first device usage scenario parameter to generate a first weight sequence, and perform iteration in this way until traversing the device usage scenario parameter set stops, obtaining the first weight sequence, the second weight sequence... the Nth weight sequence, where N is an integer greater than 1 and less than or equal to M, and M is the total number of the device usage scenario parameter set; perform joint detection on the multiple fault impact factors in sequence according to the first weight sequence, the second weight sequence... the Nth weight sequence, and combine with the device usage scenario parameter set to generate the first fault detection report, the second fault detection report... the Nth fault detection report; analyze by traversing the first fault detection report, the second fault detection report... the Nth fault detection report according to the target usage scenario parameter of the special equipment to generate the first fault detection result.

[0009] In a possible implementation, after successively combining the multiple fault impact factors according to the first weight sequence, the second weight sequence,..., the Nth weight sequence and the device usage scenario parameter set for joint detection, the following processing is performed: Fault tracing is performed based on the multiple fault impact factors to determine multiple fault sources, and there is a corresponding relationship between the multiple fault sources and the multiple fault impact factors; The multiple fault impact factors are associated with the operation dynamic data set and the device usage scenario parameter set respectively according to the multiple fault sources, and joint detection rules are set according to the association network; It is judged whether there is a weight sequence that meets the target usage scenario parameters in the first weight sequence, the second weight sequence,..., the Nth weight sequence. If not, the ith weight sequence is extracted according to the target usage scenario parameters, and the weight of the ith weight sequence is dynamically adjusted to generate the jth weight sequence; Based on the joint detection rules, the multiple fault impact factors are jointly detected according to the jth weight sequence and the target usage scenario parameters to generate the jth fault detection report.

[0010] In a possible implementation, the fault probability inference model is activated according to the first fault detection result, and the following processing is performed: The fault probability inference model is constructed by using a Bayesian network, and the first fault detection result is used as the input node; Node state assignment is performed by traversing the fault probability inference model according to the first fault detection result to obtain a state assignment result, and the state assignment result includes multiple assigned nodes; An expected operation threshold is set, and it is judged whether all the multiple assigned nodes reach the expected operation threshold; If any one of the multiple assigned nodes does not reach the expected operation threshold, a start instruction is generated, and the fault probability inference model is activated through the start instruction.

[0011] In a possible implementation, after automatically comprehensively analyzing and detecting the operation faults of special equipment based on the first fault detection result and the second fault detection result, the following processing is performed: The first fault detection result and the second fault detection result are data-fused to generate a data fusion result; Automatic fault analysis is performed according to the data fusion result to generate a fault analysis log; An alarm signal is triggered according to the fault analysis log, and the alarm signal is fed back to a remote terminal to formulate a maintenance plan for special equipment for fault response.

[0012] The present application also provides a special equipment operation fault detection device, including: a dynamic sensing module, which is used to dynamically sense special equipment according to the operation time series of the special equipment to obtain an operation dynamic data set of the special equipment; a data synchronization module, which is used to construct a long short-term memory network, synchronize the operation dynamic data set to the long short-term memory network, and determine multiple fault influence factors; a joint detection module, which is used to introduce a device usage scenario parameter set, and perform joint fault detection by combining the multiple fault influence factors with the device usage scenario parameter set to generate a first fault detection result, and activate a fault probability inference model according to the first fault detection result; a fault inference module, which is used to perform fault inference by combining the fault probability inference model with the operation dynamic data set to generate a second fault detection result; a comprehensive analysis module, which is used to perform automatic comprehensive analysis and detection of the operation faults of the special equipment based on the first fault detection result and the second fault detection result.

[0013] One or more technical solutions provided in the present application have at least the following technical effects or advantages: A special equipment operation fault detection method and device provided by the present application relate to the technical field of data processing, solve the technical problems that traditional fault detection is difficult to process large-scale data sets and real-time monitoring requirements, resulting in low detection efficiency and high maintenance costs, and realize the introduction of an intelligent and automatic fault detection method, improve the accuracy and timeliness of fault detection, and achieve the technical effect of reducing the risks and losses brought by equipment operation faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application will be briefly introduced below. Flowcharts are used in the present application to illustrate the operations performed by the devices according to the embodiments of the present application. It should be understood that the operations in the front or below do not necessarily need to be executed precisely in sequence. On the contrary, according to needs, they can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several operations can be removed from these processes.

[0015] Figure 1 It is a schematic flowchart of a special equipment operation fault detection method provided by an embodiment of the present application; Figure 2 It is a schematic structural diagram of a special equipment operation fault detection device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] The above description is only an overview of the technical solution of the present application. In order to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the specific embodiments of the present application are specifically given below.

[0017] In order to make the purpose, technical solution and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations of the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0018] In the following description, "some embodiments" are involved, which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict. The terms "first / second" involved are only used to distinguish similar objects and do not represent a specific order for the objects. The terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application.

[0019] The embodiments of the present application provide a method for detecting operation failures of special equipment. The method is applied to a device for detecting operation failures of special equipment, as Figure 1 shown, the method includes: Step A100, dynamically sense special equipment according to the operating time series of the special equipment to obtain an operating dynamic data set of the special equipment; first, a reasonable sampling frequency can be set according to the operating characteristics of the equipment. For example, for equipment with high-frequency vibration or rapid changes, the sampling frequency can be set to hundreds of times per second; for slower parameters (such as temperature), a lower sampling frequency (such as once per second) can be set, and the operating status of the equipment is collected in real time through each sensor, and the equipment parameters at each time point are recorded. And the format of data collection is a time series, and there is a corresponding record of equipment parameters at each time step, forming data points. Through the data synchronization mechanism, the data of all sensors are synchronously collected according to the same time step length to generate a multi-dimensional dynamic data set, and the multi-dimensional dynamic data set is used to characterize the complete operating status of the special equipment at each time step. Further, an operating dynamic data set of the equipment is constructed through a multi-dimensional time series. Each row in the operating dynamic data set represents the collected value of each parameter of the equipment at a certain time step, providing strong data support for the analysis of the equipment operating status, fault detection, and predictive maintenance.

[0020] Execute step A200, construct a long short-term memory network, synchronize the operating dynamic data set to the long short-term memory network, and determine multiple fault impact factors; in a possible implementation manner, step A200 further includes step A210, slice the operating dynamic data set according to the operating cycle of the special equipment to determine multiple time windows, and the multiple time windows contain multiple operating data; first, determine the operating cycle (such as every day, every hour, etc.) according to the actual operating situation of the equipment, and then slice the operating dynamic data set according to the operating cycle, which means slicing the operating dynamic data into multiple time windows with a fixed length according to the time cycle, that is, multiple time windows, and each window contains the operating data within a period of time. The multiple time windows can be used to capture the changes in the equipment operating status, so the multiple time windows contain multiple operating data.

[0021] Execute step A220, perform state prediction on the operating dynamic data set based on the multiple time windows to generate multiple operating prediction values; use the long short-term memory network to perform state prediction on the operating data of each time window. It means combining the operating data of the equipment within each time window with the operating dynamic data set and inputting them into the long short-term memory network, and predicting the future operating status of the special equipment by using the long short-term memory network, and outputting the prediction value of the next time step, and predicting the dynamic data of each time window. For example, predicting multiple operating prediction values such as the speed prediction value, vibration prediction value, and temperature prediction value of the next time step.

[0022] Execute step A230 to compare the multiple operation prediction values with the multiple operation data, perform fluctuation capture according to the data comparison result, and obtain multiple fluctuation data streams; execute step A240 to traverse the multiple fluctuation data streams for anomaly detection and identify multiple fault signals. First, compare the multiple operation prediction values of each time window with the multiple operation data of the special equipment, calculate the difference between them. If the difference between the multiple operation prediction values and the multiple operation data exceeds the set threshold, it is regarded that the operation data of the special equipment has fluctuated. The captured fluctuations are used to characterize the anomalies or changes in the equipment operation state. Then, according to the difference between the multiple operation prediction values and the multiple operation data, generate multiple fluctuation data streams, where each fluctuation data stream is used to reflect the abnormal fluctuations of the equipment within a certain period of time. Further, traverse each fluctuation data stream to detect abnormal data points, and at the same time record the time points with larger fluctuation amplitudes and mark them as potential faults. When the anomaly degree of the multiple fluctuation data streams exceeds the preset threshold, a fault signal with the corresponding anomaly is generated.

[0023] Execute step A250 to perform impact feature analysis on the multiple fault signals and determine the multiple fault impact factors. In a possible implementation manner, step A250 further includes step A251, perform feature extraction on the multiple fault signals based on the operation time series of the special equipment to determine multiple operation features; execute step A252, perform impact analysis according to the multiple operation features to obtain an operation fault impact coefficient, and the operation fault impact coefficient has fault correlation; execute step A253, perform clustering analysis on the multiple fault signals according to the operation fault impact coefficient, and combine the clustering analysis result with the fault correlation for identification to determine the multiple fault impact factors.

[0024] Performing feature extraction on multiple fault signals based on the operation time series of special equipment can be carried out by analyzing and extracting from multiple dimensions, which can include time dimension features, amplitude features, frequency features, fluctuation features, etc. The time dimension features refer to extracting the occurrence time, duration, and signal frequency of multiple fault signals, etc. The amplitude feature refers to determining by analyzing the intensity of multiple fault signals. The frequency feature refers to calculating the frequency and periodicity of multiple fault signals, and at the same time analyzing whether multiple fault signals occur repeatedly in a specific operation cycle. The fluctuation feature refers to extracting the fluctuation amplitude of multiple fault signals as a basis to judge the change rate and amplitude of equipment operation parameters, so as to determine the multiple operation features of special equipment. The multiple operation features can include abnormal speed, that is, the speed of the equipment suddenly slows down or speeds up during operation, which may be a signal of motor failure, too high temperature, that is, the abnormal rise of elevator component temperature may be a manifestation of mechanical component wear or poor lubrication, and abnormal vibration, that is, the increase in vibration frequency is usually related to mechanical failure or looseness.

[0025] Furthermore, by analyzing the operating characteristics of each fault signal and calculating its impact on the equipment operation, it means analyzing the correlation between each fault signal and the equipment performance parameters (such as production efficiency, energy consumption, downtime rate, etc.). A strong correlation means that the fault signal has a greater impact on the equipment performance. By calculating the fault correlation coefficient of each fault signal to evaluate the impact of the signal, the operating fault impact coefficient is obtained. The operating fault impact coefficient is used to reflect the impact degree of a certain fault signal on the equipment health state. At the same time, the operating fault impact coefficient has fault correlation, which means analyzing the correlation of multiple fault signals in the time dimension. Exemplarily, if some fault signals occur simultaneously within the same time period, it may indicate an internal connection between these signals. For example, abnormal temperature and abnormal vibration may occur simultaneously, meaning that increased mechanical friction causes the temperature to rise.

[0026] Finally, cluster analysis is performed on multiple fault signals according to the obtained operating fault impact coefficient above. It means grouping the fault signals according to the similarity of the feature vectors of multiple fault signals. The feature vectors can be time, amplitude, frequency, etc. Fault signals with multiple similar features are classified into the same group, generating multiple fault signal categories. Then, according to the fault correlation of the fault signals in each cluster, the average impact coefficient of all fault signals in the cluster is evaluated, and an identification label is assigned to each clustering result. Thus, according to the identification label, the key fault impact factors in each cluster are determined, providing data support and a scientific basis for the fault diagnosis of special equipment, and helping to improve the safety and reliability of equipment operation.

[0027] Execute step A300, introduce the equipment usage scenario parameter set, perform joint fault detection through the multiple fault impact factors in combination with the equipment usage scenario parameter set, generate the first fault detection result, and activate the fault probability inference model according to the first fault detection result; In a possible implementation manner, step A300 further includes step A310, constructing the equipment usage scenario parameter set based on the external operating environment and the operating operation environment of the special equipment; First, collect the external environment parameters related to the equipment operation, including environmental temperature, humidity, air pressure, vibration, etc. The external environment is used to characterize the external operating environment of the special equipment. Then, collect the operating parameters of the equipment itself, such as load conditions, equipment operation modes (high load, low load, no load), speed, acceleration, etc. The operating operation environment is used to characterize the control and sensing parameters of the special equipment during operation. Finally, combine the external environment parameters and the operating operation environment parameters to form a complete equipment usage scenario parameter set, where each equipment usage scenario parameter corresponds to the working state of the equipment in a certain specific environment.

[0028] Execute step A320, extract the first device usage scenario parameter based on the device usage scenario parameter set, allocate weights to the multiple fault impact factors according to the first device usage scenario parameter, generate the first weight sequence, and iterate accordingly until traversing the device usage scenario parameter set stops, obtaining the first weight sequence, the second weight sequence... the Nth weight sequence, where N is an integer greater than 1 and less than or equal to M, and M is the total number of the device usage scenario parameter set; randomly extract according to the device usage scenario parameter set, record the extracted parameter as the first device usage scenario parameter, and then allocate the weight of each fault impact factor according to the first device usage scenario parameter. Exemplarily, in the scenario of high temperature and high load, the weight of temperature anomaly may be higher, while in low load, the weight of speed fluctuation may be higher. Therefore, by analyzing and dynamically adjusting the weight of each impact factor, in the high temperature and high load scenario in the above example, the weight of temperature anomaly can be set: 0.6, the weight of vibration anomaly: 0.3, the weight of speed fluctuation: 0.1. Sort the weights of each fault impact factor to generate the first weight sequence, and the first weight sequence is used to represent the most critical fault impact factors in this scenario.

[0029] Further, sequentially extract different scenario parameters from the device usage scenario parameter set, analyze each scenario, and re-allocate the weights of the fault impact factors according to the specific environmental conditions of each scenario. Each iteration generates a new weight sequence, and the first weight sequence, the second weight sequence are obtained in turn until the Nth weight sequence, where N is an integer greater than 1 and less than or equal to M, and M is the total number of the device usage scenario parameter set. Iterate sequentially until all scenario parameters are traversed, so as to obtain the first weight sequence, the second weight sequence... the Nth weight sequence.

[0030] Execute step A330, sequentially combine the multiple fault impact factors according to the first weight sequence, the second weight sequence... the Nth weight sequence, and combine them with the device usage scenario parameter set for joint detection to generate the first fault detection report, the second fault detection report... the Nth fault detection report; perform fault detection by combining the fault impact factors according to the priorities of the first weight sequence, the second weight sequence... the Nth weight sequence, and each weight sequence corresponds to one joint detection. The fault detection process will screen faults according to the weight priorities, and in one possible implementation manner during the joint detection process, step A330 further includes step A331, perform fault tracing based on the multiple fault impact factors to determine multiple fault sources, and there is a corresponding relationship between the multiple fault sources and the multiple fault impact factors; execute step A332, associate the multiple fault impact factors with the operation dynamic data set and the device usage scenario parameter set respectively according to the multiple fault sources, and set joint detection rules according to the association network; execute step A333, determine whether there is a weight sequence that meets the target usage scenario parameters in the first weight sequence, the second weight sequence... the Nth weight sequence, if not, extract the ith weight sequence according to the target usage scenario parameters, and dynamically adjust the weights of the ith weight sequence to generate the jth weight sequence; execute step A334, based on the joint detection rules, combine the multiple fault impact factors with the target usage scenario parameters according to the jth weight sequence for joint detection to generate the jth fault detection report.

[0031] Performing fault tracing based on multiple fault impact factors means conducting correlation analysis on each impact factor, determining the specific fault sources of each impact factor through historical data analysis and reasoning, and then determining multiple fault sources. The fact that there is a corresponding relationship between the multiple fault sources and the multiple fault impact factors means that each fault impact factor has a corresponding fault source. Then, associating the multiple fault impact factors with the operation dynamic data set and the device usage scenario parameter set respectively according to the multiple fault sources means that time series analysis techniques can be used to find the corresponding relationship between the fault signals and the operation data, ensure data synchronization, so as to associate the fault impact factor (such as abnormal temperature) with the operation dynamic data set of the device, further analyze the operation performance of the device under different usage scenarios, match the operation state of the device with the occurrence frequency of the fault impact factor, find its association rule, and on this basis, associate the fault source with the device usage scenario parameter set, and finally form an association network among the operation dynamic data, the device usage scenario parameter set and the fault impact factor.

[0032] Furthermore, joint detection rules are set based on the association network. The joint detection rules may include the priorities of multiple fault impact factors, weight allocation, and the thresholds of associated data. Exemplarily, if the temperature anomaly has a high degree of association with high load and high temperature environment, then the temperature anomaly is preferentially detected in this scenario, so as to set the threshold and joint detection rules. When the sensor data or scenario parameters exceed the threshold, fault detection is triggered. At the same time, it is sequentially determined whether there is a weight sequence that conforms to the target usage scenario parameters in the first weight sequence, the second weight sequence... the Nth weight sequence, which means that according to the target usage scenario parameters (such as the device running in a high load mode for a long time), the generated first weight sequence, second weight sequence... the Nth weight sequence are traversed to determine whether there is a weight sequence that matches the target usage scenario. If a certain weight sequence conforms to the target usage scenario, then directly use this weight sequence for subsequent detection; if there is no weight sequence that conforms to the target usage scenario, then extract the ith weight sequence that is closest to the target usage scenario from the generated weight sequences as the basis for adjustment, and dynamically adjust the weights of the fault impact factors according to the parameters of the target usage scenario, and record the adjusted weight sequence as the jth weight sequence to ensure its adaptation to the target usage scenario.

[0033] Finally, the fault impact factors are sorted according to the jth weight sequence for priority, and joint detection is performed in combination with the target usage scenario parameters. At the same time, during the detection process, it will be sequentially detected whether the impact factors with higher weights exceed the set thresholds. After each impact factor exceeds the threshold, the system will record the fault signal and evaluate the severity of the fault. And during the joint detection process, the operating state and scenario parameters of the special equipment can also dynamically adjust the joint detection rules to ensure real-time response, so as to generate the jth fault detection report according to the detection results. The jth fault detection report contains information such as the detection situation of each fault impact factor, the situation of exceeding the threshold, the fault source, and the detection time, etc., to support the efficient maintenance and optimized operation of the equipment.

[0034] Execute step A340, analyze by traversing the first fault detection report, the second fault detection report... the Nth fault detection report according to the target usage scenario parameters of the special equipment, and generate the first fault detection result.

[0035] According to the actual operation objectives of the device (such as long-term high load, short-term low load, etc.), determine the target usage scenario parameters. According to the target usage scenario parameters, traverse the first fault detection report, the second fault detection report... the Nth fault detection report, etc., analyze the fault signals and severity levels in each report, and synthesize the detection results of all reports. Analyze according to the weight priority and scenario parameters, and finally generate the fault detection result of the device. If certain fault signals continuously appear and have a high weight in multiple reports, the system will mark this fault as a high-risk fault. Exemplarily, in multiple detections, the temperature anomaly frequently appears and has a high weight. The system will judge that this fault poses a greater threat to the device, and finally output the first fault detection result. The first fault detection result includes the fault type, fault severity level, recommended maintenance measures, etc., and the first fault detection result has a corresponding relationship with the target usage scenario parameters, ensuring the accuracy, flexibility, and timeliness of the device fault detection.

[0036] In a possible implementation manner, step A300 further includes step A350 of constructing the fault probability inference model by using a Bayesian network, and taking the first fault detection result as an input node; execute step A360 of traversing the fault probability inference model according to the first fault detection result for node state assignment to obtain a state assignment result, where the state assignment result includes multiple assigned nodes; execute step A370 of setting an expected operation threshold and judging whether all the multiple assigned nodes reach the expected operation threshold; execute step A380. If any one of the multiple assigned nodes does not reach the expected operation threshold, generate a start instruction, and activate the fault probability inference model through the start instruction.

[0037] First, a Bayesian network is used to define multiple nodes and edges of the fault probability inference model. Each node represents the operating state of a device or a fault impact factor, and the edges between nodes represent the causal relationships between fault impact factors. Subsequently, taking the first fault detection result as an input node means mapping the fault impact factors (such as abnormal temperature, abnormal vibration, etc.) in the first fault detection result to the relevant nodes in the fault probability inference model, and assigning a state to the input nodes of the fault probability inference model according to the fault information in the first fault detection result. Further, traversing the fault probability inference model for node state assignment based on the first fault detection result means, through the forward propagation algorithm, starting from the input node (such as abnormal temperature), reasoning along the directed edges in the fault probability inference model, and calculating the probability state of each node in turn. For each node, based on the state of its parent node and the conditional probability table, calculate the probability that the current node is "normal" or "abnormal", and perform state assignment according to the inferred probability state of each node until all nodes in the fault probability inference model are traversed and all node states are assigned, and record the inferred node states to generate a state assignment result. The state assignment result includes the current state ("normal" or "abnormal") of each node and its corresponding probability.

[0038] Furthermore, a desired operating threshold is set for each node. For example, the desired threshold for the temperature node can be set to 95%, that is, when the probability of abnormal temperature is lower than 95%, the state is "normal"; when it is higher than 95%, the state is "abnormal". Then, compare the assignment result of each node with its set desired operating threshold, and traverse all assigned nodes to determine whether they reach their respective desired operating thresholds. If the states of all nodes reach the desired operating thresholds, the system determines that the device is operating normally. If there is any node that does not reach its desired operating threshold (i.e., abnormal state), a start instruction is generated, and the start instruction will reactivate the fault probability inference model. The fault probability inference model will use the running dynamic data and the fault detection result to re-infer the possible fault causes and the fault development trend. During the inference process, it will check the states of other relevant nodes again and update them to ensure that the fault state of the device is fully captured, ensure timely response to potential faults, and take preventive measures.

[0039] Execute step A400 to perform fault inference by combining the running dynamic data set through the fault probability inference model to generate a second fault detection result; Based on the real-time data in the operation dynamic dataset, the state of the input nodes of the fault probability inference model (such as abnormal temperature, abnormal vibration, etc.) is updated. Then, through the forward propagation algorithm, according to the state changes of the input nodes, probability inference is carried out along the directed edges between the fault impact factors. And the state of each node is derived from the states of its parent nodes and the conditional probability table. At the same time, probability calculations are performed on all the fault nodes in the network (such as equipment fault nodes, subsystem fault nodes). And the state of each node is updated in real time according to the dynamic data, and its fault probability is adjusted through the conditional probability table. Finally, based on the current operation dynamic data, the fault probability inference model can not only infer the current fault state, but also predict the potential fault trend according to the time series changes of the fault impact factors.

[0040] Finally, the state probabilities of each fault node inferred are formed into the second fault detection result. The second fault detection result may include the probabilities of multiple fault types, the time period when the fault occurs, the correlation analysis between the operation dynamic data and the fault state, etc. Also, according to the inferred fault probabilities, the system will sort the fault types and give priority to handling the faults with higher fault probabilities and greater impacts, providing accurate fault detection information for equipment management personnel.

[0041] Next, step A500 is executed to perform an automated comprehensive analysis and detection of the operation faults of the special equipment based on the first fault detection result and the second fault detection result. In a possible implementation manner, step A500 further includes step A510 of fusing the first fault detection result and the second fault detection result to generate a data fusion result; executing step A520 of performing an automated fault analysis according to the data fusion result to generate a fault analysis log; and executing step A530 of triggering an alarm signal according to the fault analysis log and feeding it back to the remote terminal through the alarm signal to formulate a maintenance plan for the special equipment for fault response.

[0042] Before fusing the first fault detection result and the second fault detection result, the time of the first fault detection result and the second fault detection result is synchronized to ensure that the data at different time points can be accurately corresponded. Then, the first fault detection result and the second fault detection result are fused. Exemplarily, if the probability of a fault impact factor in the first fault detection result is 70%, and the same impact factor in the second fault detection result is 85%, then the probability of the fused fault impact factor can take the weighted average or a higher probability value, thereby generating a data fusion result. The data fusion result may include the state, fault probability, time of each fault impact factor, and the overall fault state of the equipment.

[0043] Furthermore, extract the characteristics of each fault impact factor (such as amplitude, frequency, duration, etc.) from the fused data, and analyze them in combination with the operating environment parameters of the device (such as temperature, load) to identify the key fault characteristics. Then, use machine learning algorithms or rule engines to analyze the cause of the device fault based on the extracted fault characteristics. By combining historical fault data, it is possible to predict the possible evolution trend of the fault, determine whether the device may have further faults or downtime, determine the fault analysis result, and record and store it according to the fault analysis result to generate a fault analysis log. Further, set the alarm trigger conditions and issue alarms according to the fault severity, fault probability, and fault trend. The set alarm trigger conditions can include multiple levels, namely low-level alarms, with a fault probability between 50% and 70%, at which time special equipment needs to be monitored; medium-level alarms, with a fault probability between 70% and 90%, at which time special equipment may require short-term maintenance; high-level alarms, with a fault probability exceeding 90%, at which time immediate handling is required. Once the alarm conditions are met, an alarm signal is generated. The alarm signal can include the fault type, the status and probability of relevant fault impact factors, the fault trend analysis result, preliminary response measures, etc., and the alarm signal is transmitted to the remote monitoring terminal through the network to ensure that managers or maintenance teams can receive real-time information on device faults in a timely manner. On the remote terminal, the alarm signal will be converted into visual fault information, including the fault occurrence time, fault type, fault impact factor status, and fault probability, etc. Finally, based on the fault information and maintenance suggestions provided in the fault analysis log, a maintenance plan is formulated for the special equipment. At the same time, according to the maintenance plan, the management team can respond to the fault of the device through remote or on-site operations. After the fault is repaired, the device operating status will be fed back to the system again, and the system will automatically detect whether the device fault has been resolved and record the maintenance process and results to optimize the subsequent detection model and maintenance process, realizing an automated closed-loop control from fault detection, analysis to maintenance response, and greatly improving the efficiency and accuracy of device management.

[0044] The embodiments of this application solve the technical problems that traditional fault detection is difficult to handle large-scale data sets and real-time monitoring requirements, resulting in low detection efficiency and high maintenance costs, and realize the introduction of intelligent and automated fault detection methods, improve the accuracy and timeliness of fault detection, and achieve the technical effect of reducing the risks and losses brought by device operation faults.

[0045] In the above text, with reference to Figure 1 A method for detecting operating faults of special equipment according to an embodiment of this application is described in detail. Next, with reference to Figure 2 A device for detecting operating faults of special equipment according to an embodiment of this application will be described.

[0046] A special equipment operation fault detection device according to an embodiment of the present application is used to solve the technical problems that traditional fault detection is difficult to process large-scale data sets and real-time monitoring requirements, resulting in low detection efficiency and high maintenance costs, and realizes the introduction of intelligent and automated fault detection methods, improves the accuracy and timeliness of fault detection, and achieves the technical effect of reducing the risks and losses brought by equipment operation faults. A special equipment operation fault detection device includes: a dynamic sensing module 10, a data synchronization module 20, a joint detection module 30, a fault inference module 40, and a comprehensive analysis module 50.

[0047] The dynamic sensing module 10 is configured to dynamically sense the special equipment according to the operation time series of the special equipment to obtain an operation dynamic data set of the special equipment; The data synchronization module 20 is configured to construct a long short-term memory network, synchronize the operation dynamic data set to the long short-term memory network, and determine a plurality of fault impact factors; The joint detection module 30 is configured to introduce a device usage scenario parameter set, perform fault joint detection by combining the plurality of fault impact factors with the device usage scenario parameter set to generate a first fault detection result, and activate a fault probability inference model according to the first fault detection result; The fault inference module 40 is configured to perform fault inference by combining the fault probability inference model with the operation dynamic data set to generate a second fault detection result; The comprehensive analysis module 50 is configured to perform automated comprehensive analysis and detection of the operation faults of the special equipment based on the first fault detection result and the second fault detection result.

[0048] Next, the specific configuration of the data synchronization module 20 will be described in detail. As described above, the operation dynamic data set is synchronized to the long short-term memory network to determine a plurality of fault impact factors. The data synchronization module 20 may further include: slicing the operation dynamic data set according to the operation cycle of the special equipment to determine a plurality of time windows, where the plurality of time windows include a plurality of operation data; performing state prediction on the operation dynamic data set based on the plurality of time windows to generate a plurality of operation prediction values; comparing the plurality of operation prediction values with the plurality of operation data, performing fluctuation capture according to the data comparison result to obtain a plurality of fluctuation data streams; traversing the plurality of fluctuation data streams for anomaly detection to identify a plurality of fault signals; performing influence feature analysis on the plurality of fault signals to determine the plurality of fault impact factors.

[0049] Next, the specific configuration of the data synchronization module 20 will be described in detail. As described above, by performing impact feature analysis on the multiple fault signals to determine the multiple fault impact factors, the data synchronization module 20 may further include: extracting features from the multiple fault signals based on the operating time series of the special equipment to determine multiple operating features; performing impact analysis according to the multiple operating features to obtain an operating fault impact coefficient, where the operating fault impact coefficient has fault correlation; performing clustering analysis on the multiple fault signals according to the operating fault impact coefficient, and combining the clustering analysis results with the fault correlation for identification to determine the multiple fault impact factors.

[0050] Next, the specific configuration of the joint detection module 30 will be described in detail. As described above, by combining the multiple fault impact factors with the equipment usage scenario parameter set for fault joint detection to generate a first fault detection result, the joint detection module 30 may further include: constructing the equipment usage scenario parameter set based on the external operating environment and the operating operation environment of the special equipment; extracting the first equipment usage scenario parameter based on the equipment usage scenario parameter set, and assigning weights to the multiple fault impact factors according to the first equipment usage scenario parameter to generate a first weight sequence, and performing iteration in this way until traversing the equipment usage scenario parameter set and stopping to obtain the first weight sequence, the second weight sequence... the Nth weight sequence, where N is an integer greater than 1 and less than or equal to M, and M is the total number of the equipment usage scenario parameter set; successively combining the multiple fault impact factors according to the first weight sequence, the second weight sequence... the Nth weight sequence with the equipment usage scenario parameter set for joint detection to generate the first fault detection report, the second fault detection report... the Nth fault detection report; traversing the first fault detection report, the second fault detection report... the Nth fault detection report for analysis according to the target usage scenario parameter of the special equipment to generate the first fault detection result.

[0051] Next, the specific configuration of the joint detection module 30 will be described in detail. As described above, after the multiple fault impact factors are jointly detected in sequence according to the first weight sequence, the second weight sequence... the Nth weight sequence, in combination with the device usage scenario parameter set, the joint detection module 30 may further include: performing fault traceability based on the multiple fault impact factors to determine multiple fault sources, where there is a corresponding relationship between the multiple fault sources and the multiple fault impact factors; associating the multiple fault impact factors with the operation dynamic data set and the device usage scenario parameter set respectively according to the multiple fault sources, and setting joint detection rules according to the association network; determining whether there is a weight sequence that meets the target usage scenario parameters in the first weight sequence, the second weight sequence... the Nth weight sequence, and if not, extracting the ith weight sequence according to the target usage scenario parameters and dynamically adjusting the weights of the ith weight sequence to generate the jth weight sequence; based on the joint detection rules, jointly detecting the multiple fault impact factors according to the jth weight sequence in combination with the target usage scenario parameters to generate the jth fault detection report.

[0052] Next, the specific configuration of the joint detection module 30 will be described in detail. As described above, activating the fault probability inference model according to the first fault detection result, the joint detection module 30 may further include: constructing the fault probability inference model using a Bayesian network, with the first fault detection result as the input node; traversing the fault probability inference model according to the first fault detection result to assign node states and obtaining a state assignment result, where the state assignment result includes multiple assigned nodes; setting an expected operation threshold and determining whether all the multiple assigned nodes reach the expected operation threshold; if any one of the multiple assigned nodes does not reach the expected operation threshold, generating a start command to activate the fault probability inference model through the start command.

[0053] Next, the specific configuration of the comprehensive analysis module 50 will be described in detail. As described above, after performing automated comprehensive analysis and detection of the operation faults of special equipment based on the first fault detection result and the second fault detection result, the comprehensive analysis module 50 may further include: fusing the first fault detection result and the second fault detection result to generate a data fusion result; performing automated fault analysis according to the data fusion result to generate a fault analysis log; triggering an alarm signal according to the fault analysis log, and feeding back the alarm signal to a remote terminal to formulate a maintenance plan for special equipment for fault response.

[0054] A special equipment operation fault detection device provided by an embodiment of the present application can execute a special equipment operation fault detection method provided by any embodiment of the present application, and has functional modules and beneficial effects corresponding to the execution of the method.

[0055] Although the present application makes various references to certain modules in the device according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The included individual units and modules are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of the functional units are only for the convenience of mutual distinction and do not limit the protection scope of the present application.

[0056] The above specific implementation manners do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A method for detecting operation failure of special equipment, characterized in that: The method comprises: Perform dynamic sensing on the special equipment according to the operation time sequence of the special equipment to obtain the operation dynamic data set of the special equipment; Building a long short-term memory network, synchronizing the running dynamic data set to the long short-term memory network, and determining multiple fault influencing factors; Introducing a device usage scenario parameter set, performing joint fault detection by combining the multiple fault influencing factors with the device usage scenario parameter set, generating a first fault detection result, and activating a fault probability reasoning model according to the first fault detection result; Performing fault reasoning by combining the fault probability reasoning model with the operation dynamic data set to generate a second fault detection result; Based on the first fault detection result and the second fault detection result, an automated comprehensive analysis and detection is performed on the operation faults of the special equipment.

2. A special equipment operation fault detection method according to claim 1, characterized in that: The running dynamic data set is synchronized to the long short-term memory network to determine multiple fault influencing factors, the method comprising: According to the operation cycle of the special equipment, the operation dynamic data set is sliced ​​to determine a plurality of time windows, wherein the plurality of time windows contain a plurality of operation data; Performing state prediction on the operation dynamic data set based on the multiple time windows to generate multiple operation prediction values; Comparing the plurality of operation prediction values ​​with the plurality of operation data, and performing fluctuation capture according to the data comparison result to obtain a plurality of fluctuation data streams; Traversing the multiple fluctuating data streams to perform anomaly detection and identify multiple fault signals; Performing impact characteristic analysis on the multiple fault signals to determine the multiple fault impact factors.

3. A special equipment operation fault detection method as claimed in claim 2, characterized in that: Performing an impact feature analysis on the multiple fault signals to determine the multiple fault impact factors, the method comprising: Extracting features of the plurality of fault signals based on the operation time series of the special equipment to determine a plurality of operation features; Performing an impact analysis according to the plurality of operating characteristics to obtain an operating fault impact coefficient, wherein the operating fault impact coefficient has fault correlation; The multiple fault signals are clustered and analyzed according to the operational fault influence coefficients, and the cluster analysis results are identified in combination with the fault correlation to determine the multiple fault influence factors.

4. A special equipment operation fault detection method according to claim 1, characterized in that: Introducing a device usage scenario parameter set, performing joint fault detection by combining the multiple fault influencing factors with the device usage scenario parameter set, and generating a first fault detection result, the method comprising: Based on the external operating environment of the special equipment and the operating environment, a parameter set of the equipment usage scenario is constructed; Extracting a first device usage scenario parameter based on the device usage scenario parameter set, weighting the multiple fault influencing factors according to the first device usage scenario parameter, generating a first weight sequence, iterating until the traversal of the device usage scenario parameter set is completed, and obtaining a first weight sequence, a second weight sequence... an Nth weight sequence, where N is an integer greater than 1 and less than or equal to M, and M is the total number of the device usage scenario parameter sets; The multiple fault influencing factors are sequentially detected according to the first weight sequence, the second weight sequence ... the Nth weight sequence, combined with the device usage scenario parameter set, to generate a first fault detection report, a second fault detection report ... the Nth fault detection report; According to the target usage scenario parameters of the special equipment, the first fault detection report, the second fault detection report...the Nth fault detection report are traversed and analyzed to generate the first fault detection result.

5. A special equipment operation fault detection method as claimed in claim 4, characterized in that: After the multiple fault influencing factors are sequentially detected according to the first weight sequence, the second weight sequence ... the Nth weight sequence in combination with the device usage scenario parameter set, the method includes: Perform fault tracing based on the multiple fault influencing factors to determine multiple fault sources, where the multiple fault sources correspond to the multiple fault influencing factors; According to the multiple fault sources, the multiple fault influencing factors are respectively associated with the operation dynamic data set and the equipment usage scenario parameter set, and a joint detection rule is set according to the associated network; Determine whether there is a weight sequence that meets the target usage scenario parameters among the first weight sequence, the second weight sequence, ... the Nth weight sequence; if not, extract the i-th weight sequence according to the target usage scenario parameters, and dynamically adjust the weight of the i-th weight sequence to generate the j-th weight sequence; Based on the joint detection rule, the multiple fault influencing factors are jointly detected according to the j-th weight sequence in combination with the target usage scenario parameters to generate a j-th fault detection report.

6. A special equipment operation fault detection method according to claim 1, characterized in that: Activating a fault probability reasoning model according to the first fault detection result, the method includes: Using a Bayesian network to construct the fault probability reasoning model, taking the first fault detection result as an input node; According to the first fault detection result, traverse the fault probability reasoning model to perform node state assignment to obtain a state assignment result, wherein the state assignment result includes a plurality of assignment nodes; Setting an expected operation threshold, and determining whether the plurality of value assignment nodes all reach the expected operation threshold; If any one of the multiple value assignment nodes fails to reach the expected operation threshold, a startup instruction is generated, and the fault probability reasoning model is activated through the startup instruction.

7. A special equipment operation fault detection method according to claim 1, characterized in that: After performing automatic comprehensive analysis and detection on the operation fault of the special equipment based on the first fault detection result and the second fault detection result, the method includes: Performing data fusion on the first fault detection result and the second fault detection result to generate a data fusion result; Performing automated fault analysis based on the data fusion results and generating a fault analysis log; An alarm signal is triggered according to the fault analysis log, and the alarm signal is fed back to the remote terminal to formulate a maintenance plan for special equipment to respond to the fault.

8. A special equipment operation fault detection device, characterized in that: The special equipment operation fault detection device is used to implement the special equipment operation fault detection method according to any one of claims 1 to 7, and the device comprises: A dynamic sensing module, wherein the dynamic sensing module is used to dynamically sense the special equipment according to the operation time sequence of the special equipment to obtain an operation dynamic data set of the special equipment; A data synchronization module, the data synchronization module is used to build a long short-term memory network, synchronize the running dynamic data set to the long short-term memory network, and determine multiple fault influencing factors; a joint detection module, the joint detection module being used to introduce a device usage scenario parameter set, perform joint fault detection by combining the multiple fault influencing factors with the device usage scenario parameter set, generate a first fault detection result, and activate a fault probability reasoning model according to the first fault detection result; A fault reasoning module, the fault reasoning module is used to perform fault reasoning by combining the fault probability reasoning model with the operation dynamic data set to generate a second fault detection result; A comprehensive analysis module is used to perform automatic comprehensive analysis and detection on operation faults of special equipment based on the first fault detection result and the second fault detection result.

Images