Methods, systems and media for diagnosing operational faults in electromechanical equipment

Abstract

This invention relates to the field of equipment fault diagnosis technology, specifically to a method, system, and medium for diagnosing operational faults in electromechanical equipment. The method includes: acquiring several monitoring data points of a pressure-stabilized water pump using a monitoring device; collecting fault types and historical maintenance records of the pressure-stabilized water pump; establishing a fault tree; obtaining fault feature vectors by analyzing the similarity of monitoring data before and after maintenance for different fault types; conducting disturbance tests on the pressure-stabilized water pump and analyzing the changes in monitoring data before and after maintenance to obtain fault characterization parameters and test vectors; adjusting the test vectors using the fault characterization parameters to obtain fault feature enhancement vectors; obtaining similarity weights through the fault feature vectors and fault feature enhancement vectors, and weighting the occurrence probabilities based on the frequency of each fault type in historical maintenance records, thereby combining the fault tree to diagnose faults in the pressure-stabilized water pump. This invention improves the efficiency of fault diagnosis for pressure-stabilized water pumps.

Description

Technical Field

[0001] This invention relates to the field of equipment fault diagnosis technology, and specifically to methods, systems and media for diagnosing operational faults in electromechanical equipment. Background Technology

[0002] The stable operation of fire protection equipment is directly related to fire prevention efficiency and personnel safety. For fault detection, a multi-layered technical system has been developed, ranging from traditional manual detection to intelligent prediction. Its significance extends far beyond simple maintenance, becoming a core support for modern smart fire protection. Common fire protection equipment includes various types of water pumps. Among them, pressure-stabilizing pumps are used to maintain the water pressure of the fire protection network under normal conditions, meeting the water demand before the main pump starts in the initial stage of a fire alarm. Their operational status needs to be monitored continuously.

[0003] Existing analyses of pressure-stabilized water pump failures mostly rely on building decision trees for fault diagnosis based on abnormal sensor data. Traditional decision tree analysis lacks specificity and efficiency. In other words, the diagnostic process lacks proactive pre-judgment schemes for fault types. When low-frequency fault types occur, the investigation of the true fault cause is too late, requiring a lot of time to obtain the true fault cause, resulting in low diagnostic efficiency. Summary of the Invention

[0004] This invention provides a method, system, and medium for diagnosing operational faults in electromechanical equipment to solve existing problems.

[0005] The present invention provides a method, system, and medium for diagnosing operational faults in electromechanical equipment, which adopts the following technical solution:

[0006] One embodiment of the present invention provides a method for diagnosing operational faults in electromechanical equipment, the method comprising the following steps:

[0007] Several monitoring data of the pressure-stabilizing water pump are obtained using a monitoring device, the fault types and historical maintenance records of the pressure-stabilizing water pump are collected, and a fault tree is established based on the monitoring data and fault types. The historical maintenance records contain several maintenance events.

[0008] By utilizing the similarity of monitoring data before and after maintenance for different fault types in all maintenance events in historical maintenance records, fault feature vectors for each fault type are obtained; disturbance tests are conducted on pressure-stabilizing water pumps, and the changes in monitoring data before and after the disturbance tests are analyzed to obtain fault characterization parameters and test vectors of the monitoring device; the test vectors are adjusted using the fault characterization parameters to obtain fault feature enhancement vectors; and the similarity weights of fault types are obtained through the fault feature vectors and fault feature enhancement vectors of the fault types.

[0009] Based on the frequency of occurrence of each fault type in historical maintenance records, and combined with similarity weights, the weighted probability of occurrence of each fault type is obtained.

[0010] Fault diagnosis of pressure-stabilized water pumps is performed by using the weighted probability of occurrence of fault types and combining it with fault trees.

[0011] Furthermore, the method for obtaining fault feature vectors for each fault type by utilizing the similarity of monitoring data before and after maintenance for all maintenance events in historical maintenance records for different fault types includes:

[0012] For any fault type in any maintenance event, the degree of difference of the monitoring data is obtained based on the DTW distance between any monitoring data collected before and after the maintenance event for the fault type. An array of the degree of difference of all monitoring data for the corresponding fault type is obtained as the feature vector of the maintenance event. For any fault type, the average feature vector of all maintenance events corresponding to the fault type is obtained as the fault feature vector of the fault type.

[0013] Furthermore, the disturbance test on the pressure-stabilizing water pump, and the analysis of the changes in monitoring data before and after the disturbance test, to obtain the fault characterization parameters and test vectors of the monitoring device, include the following specific methods:

[0014] A preset threshold range is set for the monitoring data collected by each monitoring device of the pressure-stabilizing water pump. When the monitoring data corresponding to the monitoring device exceeds the threshold range, the monitoring data is regarded as abnormal monitoring data, and a disturbance test item is set to test the pressure-stabilizing water pump. The range of data points in the monitoring data is taken as the fluctuation amplitude. The fluctuation amplitude of the disturbance monitoring data obtained by any monitoring device and the monitoring data before the test are obtained respectively. According to the difference in the fluctuation amplitude of the disturbance monitoring data and the monitoring data before the test, the fault characterization parameter of the monitoring device is obtained. The difference in fluctuation amplitude is negatively correlated with the fault characterization parameter.

[0015] Using disturbance monitoring data and pre-test monitoring data obtained from all monitoring devices with abnormal monitoring data, and combined with the method for obtaining feature performance vectors, a feature performance vector is obtained, denoted as the test vector. Each element in the test vector corresponds to a monitoring device.

[0016] Furthermore, the specific method for adjusting the test vector using fault characterization parameters to obtain the fault feature enhancement vector includes:

[0017] By multiplying the corresponding elements of the monitoring device in the test vector using the fault characterization parameters of any monitoring device, the adjusted values ​​of the corresponding elements of the monitoring device in the test vector are obtained, thereby obtaining the fault feature enhancement vector.

[0018] Furthermore, the specific method for obtaining the similarity weight of the fault type through the fault feature vector and the fault feature enhancement vector of the fault type includes:

[0019] Obtain the fault type corresponding to all leaf nodes in the fault tree for the anomaly monitoring data, and denote it as the first fault type. Use the cosine similarity between the fault feature vector and the fault feature enhancement vector of the first fault type as the similarity weight of the first fault type.

[0020] Furthermore, the method for obtaining the weighted probability of occurrence of each fault type based on its frequency of occurrence in historical maintenance records and combined with similarity weights includes the following specific methods:

[0021] The total frequency of all fault types occurring when any monitoring data in the historical maintenance records shows anomalies is recorded as the first parameter. Any fault type occurring when any monitoring data shows anomalies is recorded as the target fault type. The frequency of the target fault type is recorded as the second parameter of the target fault type. The ratio of the second parameter to the first parameter is recorded as the occurrence probability of the target fault type. The occurrence probabilities of all fault types occurring when all monitoring data shows anomalies are superimposed and linearly normalized to obtain the new occurrence probability of all fault types in the fault tree. The new occurrence probability of the fault type is weighted and adjusted using similarity weights to obtain the weighted occurrence probability of the fault type.

[0022] Furthermore, the specific method for diagnosing faults in a pressure-stabilizing water pump using a weighted probability distribution of fault types combined with a fault tree is as follows:

[0023] The fault trees are sorted in descending order based on the weighted probability of each fault type, and the sorting result is used as the order for troubleshooting.

[0024] Furthermore, the specific methods for establishing a fault tree based on monitoring data and fault types include:

[0025] Based on the data types collected by all monitoring devices, a fault tree analysis method is used to construct a fault tree for the pressure-stabilizing water pump. The tree is constructed from top to bottom, with the top node representing each monitoring data point and the leaf nodes of each monitoring data point corresponding to several fault types.

[0026] A fault diagnosis system for electromechanical equipment includes: at least one memory for storing a program; and at least one processor for loading the program to execute any of the fault diagnosis methods for electromechanical equipment described above.

[0027] A storage medium storing processor-executable instructions, which, when executed by a processor, are used to implement any of the methods for diagnosing operational faults in electromechanical equipment.

[0028] The beneficial effects of the technical solution of this invention are as follows: Fault feature vectors are extracted from historical maintenance records, and fault feature enhancement vectors are obtained by combining them with disturbance testing. Similarity weights are used to match the differences between the two. Dynamic disturbance testing amplifies the weak features of early faults (such as limited fluctuations in sensor data). Combined with historical feature vectors, early faults that are difficult to capture using traditional threshold methods can be identified. The reliability of sensor data (fault characterization parameters) is verified through disturbance testing, effectively eliminating environmental noise interference and avoiding misdiagnosis. Furthermore, based on fault tree analysis, the calculation range of similarity weights is limited, focusing only on fault types related to the current abnormal monitoring data, combined with the historical frequency of fault types (statistical analysis). By combining the probability of occurrence (statistical probability) with real-time similarity weights (dynamic matching degree), a weighted probability of occurrence is calculated, which enables the fault tree to provide physical correlation constraints and avoid interference from irrelevant fault types. The weighted probability takes into account both historical patterns and real-time characteristics, improving diagnostic reliability and providing stronger analytical capabilities for multiple coexisting or related faults (such as pressure fluctuations caused by abnormal motor power). Furthermore, by quantifying the priority of each fault type through the weighted probability of occurrence, the elements in the feature vector are matched one-to-one with the monitoring device, which can locate key fault sources, thereby prioritizing the investigation of high-weight fault types, reducing blind maintenance time, and effectively improving the efficiency of fault diagnosis for pressure-stabilizing water pumps. Attached Figure Description

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1 This is a flowchart of the steps in the method for diagnosing operational faults of electromechanical equipment according to the present invention;

[0031] Figure 2 This is a partial schematic diagram of the corresponding subtree in the fault tree when there is an abnormality in the pump bearing temperature, according to an embodiment of the present invention. Detailed Implementation

[0032] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation methods, structures, features, and effects of the method, system, and medium for diagnosing operational faults in electromechanical equipment proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0034] The following description, in conjunction with the accompanying drawings, details the specific solutions for the operational fault diagnosis method, system, and medium for electromechanical equipment provided by this invention.

[0035] Please see Figure 1 The diagram illustrates a flowchart of a method for diagnosing operational faults in electromechanical equipment according to an embodiment of the present invention. The method includes the following steps:

[0036] Step S001: Use the monitoring device to obtain some monitoring data of the pressure-stabilizing water pump, collect the fault types and historical maintenance records of the pressure-stabilizing water pump, and establish a fault tree based on the monitoring data and fault types.

[0037] Specifically, in order to implement the fault diagnosis method for electromechanical equipment proposed in this embodiment, it is first necessary to collect multi-dimensional monitoring data through several monitoring devices. The specific process is as follows:

[0038] First, monitoring devices were installed at several key locations of the fire-fighting pressure-stabilizing pump to obtain corresponding monitoring data.

[0039] As an optional embodiment, the specific process for acquiring monitoring data is as follows: a high-precision pressure monitoring device and a flow meter are installed at the pump outlet; a motor protector is installed to monitor current and power; a vibration acceleration sensor is installed on the motor bearing base; a thermistor temperature sensor is installed on the motor bearing, stator winding, and pump bearing; and a level gauge is installed on the pressure tank output by the pressure-stabilizing pump. All data collected by the monitoring devices are collectively referred to as monitoring data.

[0040] Then, based on the data types collected by all monitoring devices, a fault tree for the pressure-stabilizing water pump is constructed using the Fault Tree Analysis (FTA) method. The tree is constructed from top to bottom, with the top node representing each monitoring data point and the leaf nodes of each monitoring data point corresponding to several fault types.

[0041] It should be noted that, as Figure 2The diagram shows a partial subtree in the fault tree when there is an abnormality in the pump bearing temperature. In this subtree, the node "abnormal bearing temperature" corresponds to the monitoring of the bearing temperature. When an abnormality occurs, its leaf node in the subtree corresponds to the possible fault types, including: insufficient lubricating oil, lubricating oil contamination, bearing damage, excessive axial preload, misalignment, cooling system aging, excessively high ambient temperature, and ventilation blockage.

[0042] It should be noted that fault tree analysis is a top-down deductive failure analysis method that uses Boolean logic to combine low-order events to analyze unwanted states in the system. Since fault tree analysis is an existing method, the embodiments of this invention will not be described in detail.

[0043] Finally, the historical maintenance records of the electromechanical equipment are obtained. These records contain several maintenance events and fault types. For each maintenance event, the fault type it addresses is determined. The records before and after each maintenance event are then retrieved. Various monitoring data within the time range, including This is the preset first duration parameter.

[0044] It should be noted that the first duration parameter is preset based on experience. The duration is 1 hour, but can be adjusted according to actual circumstances. This embodiment of the invention does not impose specific limitations.

[0045] Thus, the monitoring data, fault tree, and maintenance events contained in historical maintenance records are obtained through the above methods.

[0046] Step S002: Utilize the similarity of monitoring data before and after maintenance for different fault types in all maintenance events in historical maintenance records to obtain fault feature vectors for each fault type; conduct disturbance tests on the pressure-stabilizing water pump, analyze the changes in monitoring data before and after the disturbance test, and obtain the fault characterization parameters and test vectors of the monitoring device respectively; adjust the test vectors using the fault characterization parameters to obtain fault feature enhancement vectors; obtain the similarity weights of fault types through the fault feature vectors and fault feature enhancement vectors of the fault types.

[0047] It should be noted that fire-fighting pressure-stabilizing water pumps have accumulated a large amount of historical fault maintenance data after a long period of operation. Since the pressure-stabilizing water pump and pressure tank operate in tandem as a system, different faults exhibit distinct data characteristics. By acquiring monitoring data before maintenance, a historical characteristic matrix for different faults can be established. Furthermore, by analyzing the consistency between the current fault characteristics of the pressure-stabilizing water pump and those of historical faults, the current fault status of the pump can be pre-determined. However, for acquiring real-time fault characteristics during the current operation of the pressure-stabilizing water pump, the current fault status is less obvious than historical faults, often being in the early stages, and its characteristics are not readily apparent. If the same method used to extract historical fault characteristics is applied, the fault type identification error will be significant. However, considering that when the data is abnormal, the motor disturbance test can be used to determine whether the data can represent the actual fault situation, since the fire-fighting pressure stabilization system drives water flow through the motor, and other monitoring parameters are closely related to the motor's operating status, when the motor is in normal condition, the disturbance test will cause the monitoring data to fluctuate significantly; while when a certain monitoring parameter is in a fault state, the data fluctuation caused by the disturbance test will be limited, thus allowing for a quick prediction of the contribution of the monitoring data to fault type identification.

[0048] Therefore, firstly, the parameter changes of the fault are determined based on history to construct the historical characteristics of the fault; secondly, the enhanced characteristics of the currently operating equipment are extracted through disturbance testing to obtain the weight of the pre-judged fault type; finally, the optimization probability is obtained by using the occurrence probability of each fault type in the fault tree.

[0049] Specifically, in step S201, based on the similarity between the same monitoring data of the electromechanical equipment before and after any maintenance event, the feature performance vector of the maintenance event is obtained, and based on the feature performance vectors in all maintenance events for the same fault type, the fault feature vector of the fault type is obtained.

[0050] It should be noted that obtaining various monitoring data from logs requires a reference object to assess the anomaly characteristics of different monitoring data. Typically, for a period of time after system installation and debugging, since all equipment in the system is brand new, the possibility of system failure is low, and it can usually be considered a normal situation. Therefore, the initial operating data after system debugging can be used as the basic reference object.

[0051] As a preferred embodiment, the method for obtaining the feature representation vector of the maintenance event includes:

[0052] First, for any fault type in any maintenance event, the degree of difference of the monitoring data is obtained based on the DTW distance between any monitoring data collected before and after the maintenance event for the fault type. The DTW distance is positively correlated with the degree of difference.

[0053] Then, an array of the degree of difference of all monitoring data under the corresponding fault type for the maintenance event is obtained, which is used as the feature representation vector of the maintenance event.

[0054] Finally, for any fault type, the average feature vector of all maintenance events corresponding to the fault type is obtained as the fault feature vector of the fault type.

[0055] It should be noted that the DTW distance is calculated using the DTW (Dynamic Time Warping) algorithm, which is an existing similarity calculation algorithm, and therefore will not be described in detail in this embodiment of the invention.

[0056] As an optional embodiment, the specific method for calculating the degree of difference is as follows:

[0057]

[0058] In the formula, Indicates the first The second repair incident was related to the first When the fault type is 1, the 2nd fault type is 3. The degree of difference in monitoring data collected by each monitoring device; This represents the dynamic warping algorithm; Indicates the first The second repair incident was related to the first Before the first fault type, the first The monitoring data collected by each monitoring device; Indicates the first The second repair incident was related to the first After the first fault type, the second The monitoring data collected by each monitoring device; express and DTW distance; This represents the linear normalization function.

[0059] It should be noted that the degree of difference reflects the dynamic changes of the same monitoring parameter in the time series before and after the maintenance event. The DTW algorithm can capture the non-linear alignment characteristics of the time series and quantify the degree of deviation of the data form before and after the fault. The greater the degree of difference, the higher the sensitivity of the monitoring parameter to the fault. Therefore, the degree of difference serves as a core indicator of historical fault characteristics, establishing a distinguishable feature basis for different fault types.

[0060] It should be noted that, since the severity of each fault type varies during each maintenance, the feature vector of the same fault type will also vary. Therefore, in this embodiment of the invention, the average feature vector of multiple maintenance events of the same fault type (i.e., the fault feature vector of the fault type) is used as the feature of the fault type in terms of fault severity.

[0061] Step S202: Perform a disturbance test on the pressure-stabilizing water pump, analyze the changes in monitoring data before and after the disturbance test, obtain the fault characterization parameters and test vector of the monitoring device, adjust the test vector using the fault characterization parameters to obtain the fault feature enhancement vector, and obtain the similarity weight of the fault type through the fault feature vector and the fault feature enhancement vector of the fault type.

[0062] It should be noted that the above steps, based on historical maintenance records and monitoring data, constructed characteristic manifestations corresponding to different fault types. However, during real-time monitoring, anomalies were detected in data from individual sensor types. Since sensors are susceptible to external environmental interference, the current anomalies may be false anomalies caused by the environment. Therefore, if the monitoring data is directly compared with the operating data at the time of installation to obtain the current fault characteristics, it would be assumed that the current monitoring data is under equipment fault conditions. This results in a certain difference between the fault state and the actual performance of the equipment. Therefore, it is necessary to conduct a certain disturbance test on the currently operating equipment to eliminate false anomalies caused by environmental interference. The motor is the core of the pressure-stabilizing water pump, and its power determines the pressure provided by the pump. When the pressure in the pressure tank drops to the specified pressure, the motor of the pressure-stabilizing water pump will start and operate at a specific high power to pump water into the pressure tank to maintain a certain pressure. When the pressure tank reaches the specified pressure, the motor will operate at low power to prevent damage caused by prolonged shutdown and to meet the requirements of instantaneous response. During normal operation, the motor's low-power and high-power operation are typically stable and last for a certain duration. Any data anomalies also occur during these two periods. False anomalies caused by the environment usually change along with the motor's operating conditions. Therefore, when monitored data anomalies occur, changing the motor power will significantly alter multiple data points such as pumped water flow and pressure. However, in cases of genuine faults, due to abnormal sensor monitoring locations, the data fluctuations are relatively limited. Therefore, during disturbance testing involving changes in motor power, a smaller change in a particular sensor-monitored parameter is more likely to indicate a fault.

[0063] Specifically, in step S221, disturbance test conditions are set and disturbance test is performed on the pressure-stabilizing water pump. The difference in fluctuation amplitude of the monitoring data before and after the disturbance test is used to obtain the fault characterization parameters of the monitoring device.

[0064] First, a preset threshold range is set for the monitoring data collected by each monitoring device of the pressure-stabilizing water pump. When the monitoring data corresponding to the monitoring device exceeds the threshold range, the monitoring data is regarded as abnormal monitoring data, and a test item for interference is set to test the pressure-stabilizing water pump.

[0065] It should be noted that when setting the threshold range for each monitoring data of the pressure-stabilizing water pump in the embodiments of the present invention, the parameter requirements for the operation of the pressure-stabilizing water pump in "GB50974-2014 Technical Specification for Fire Water Supply and Fire Hydrant System" and "GB6245-2006 Fire Pump" can be referred to.

[0066] As an optional embodiment, the specific method for setting the disturbance test item to test the pressure-stabilizing water pump is as follows: the disturbance test item includes running the motor at low power, high power, and short-term shutdown respectively; when there is abnormal monitoring data and the pressure-stabilizing water pump is in normal low-power operation, a short-term shutdown disturbance test is performed; if the pressure-stabilizing water pump is in high-power operation, a low-power operation disturbance test is performed; and the duration of the disturbance test is preset to be no less than The monitoring data of each monitoring device during the disturbance test is recorded and denoted as disturbance monitoring data; simultaneously, monitoring data of the same duration as the disturbance test before the monitoring data anomaly occurs is acquired and denoted as pre-test monitoring data; among which... This is the preset second duration parameter.

[0067] It should be noted that the second duration parameter is preset to 1 minute based on experience, but it can be adjusted according to the actual situation. This embodiment of the invention does not impose any specific limitations.

[0068] Then, the range of data points in the monitoring data is used as the fluctuation amplitude, and the fluctuation amplitude of the disturbance monitoring data and the pre-test monitoring data obtained by any monitoring device are obtained respectively; based on the difference in fluctuation amplitude between the disturbance monitoring data and the pre-test monitoring data, the fault characterization parameter of the monitoring device is obtained, and the difference in fluctuation amplitude is negatively correlated with the fault characterization parameter.

[0069] As an optional embodiment, the specific calculation method for the fault characterization parameters is as follows:

[0070]

[0071] In the formula, Indicates the first Fault characterization parameters of each monitoring device; Indicates the first The fluctuation range of disturbance monitoring data collected by each monitoring device; Indicates the first The fluctuation range of the monitoring data collected by each monitoring device before the test; This represents the absolute value function.

[0072] It should be noted that the fault characterization parameter is used to describe the probability that the monitoring data collected by the corresponding monitoring device can reflect the existence of a fault in the pressure-stabilized water pump, that is, to reflect the ability of the monitoring data to characterize the actual fault. The larger the value of the fault characterization parameter, the greater the probability that the monitoring data collected by the corresponding monitoring device can effectively reflect the existence of a fault in the pressure-stabilized water pump. This provides dynamic weights for the subsequent fault feature enhancement vector, strengthening the contribution of high-reliability sensors to fault diagnosis. After the pressure-stabilized water pump is subjected to a disturbance test, for each monitoring device whose collected monitoring data is normal, the fluctuation amplitude of the collected monitoring data will change significantly. However, for monitoring devices whose collected monitoring data is abnormal, the change in the fluctuation amplitude of the collected monitoring data before and after the disturbance test is relatively small. Therefore, the difference between the fluctuation amplitude of the disturbance monitoring data collected by the monitoring device before and after the disturbance test and the monitoring data before the test is significant. The smaller the value, the better the monitoring device can reflect the fault condition of the pressure-stabilizing water pump; in addition, the monitoring data obtained during the disturbance test includes both abnormal and non-abnormal monitoring data.

[0073] Step S222: Using the changes in monitoring data before and after the disturbance test, obtain the test vector, and use the fault characterization parameters to adjust the corresponding elements of the monitoring device in the test vector to obtain the fault feature enhancement vector.

[0074] As a preferred embodiment, disturbance monitoring data and pre-test monitoring data obtained from all monitoring devices with abnormal monitoring data are used, and combined with the method for obtaining feature performance vectors, a feature performance vector is obtained, denoted as a test vector. Each element in the test vector corresponds to a monitoring device. The fault characterization parameters of any monitoring device are used to multiply the corresponding element of the monitoring device in the test vector to obtain the adjusted value of the corresponding element of the monitoring device in the test vector, thereby obtaining the fault feature enhancement vector.

[0075] As an optional embodiment, the specific calculation method for the enhanced fault characteristics is as follows:

[0076]

[0077] In the formula, Indicates the first The adjusted value of the corresponding element in the test vector for each monitoring device; Indicates the first Fault characterization parameters of each monitoring device; Indicates the first The numerical value of the corresponding element in the test vector for each monitoring device.

[0078] It should be noted that enhancing fault characteristics involves introducing fault characterization parameters to amplify the contribution of fault characteristics from high-reliability sensors and suppress interference from low-reliability sensors; for example, when When the value is large, the monitoring data is more likely to reflect the real fault, and its feature weight increases accordingly. Therefore, by dynamically weighting, the fault indication role of key monitoring devices is highlighted, the distinguishability of feature vectors is improved, and by combining historical features with real-time data from disturbance tests, the sensitivity to early faults is enhanced and the false negative rate is reduced.

[0079] Step S223: Calculate the similarity between the fault feature enhancement vector and the fault feature vector to obtain the similarity weight of the fault type.

[0080] It should be noted that the current enhanced fault features reflect the further deepening of fault data features after the current data is abnormal. By calculating the similarity with historical fault feature vectors, the higher the similarity, the more likely it is to be this type of fault.

[0081] As a preferred embodiment, the specific method for obtaining the similarity weight of the fault is as follows: obtain the fault type corresponding to all leaf nodes of the abnormal monitoring data in the fault tree, denoted as the first fault type, and use the cosine similarity between the fault feature vector and the fault feature enhancement vector of the first fault type as the similarity weight of the first fault type.

[0082] It should be noted that in the fault tree, each monitoring data point corresponds to several leaf nodes, and each leaf node corresponds to a fault type.

[0083] As an optional embodiment, the specific method for calculating the similarity weight is as follows:

[0084]

[0085] In the formula, Indicates the first The similarity weights for the first fault type; Represents the fault feature enhancement vector; Indicates the first A fault feature vector for the first fault type; This represents the cosine similarity function.

[0086] It should be noted that the similarity weight represents the degree of matching between the current fault characteristics and historical fault types. The larger the value (closer to 1), the more likely it is to belong to that fault type. By using cosine similarity, the directional consistency in the high-dimensional feature space can be quantified, ignoring amplitude differences. At the same time, it is convenient to use the similarity weight directly as the basis for prioritizing fault types, assisting maintenance decisions (such as prioritizing the investigation of high-weight faults).

[0087] Thus, the similarity weights are obtained through the above method.

[0088] Step S003: Based on the frequency of occurrence of each fault type in historical maintenance records and combined with similarity weights, obtain the weighted probability of occurrence of each fault type.

[0089] It should be noted that by obtaining the similarity weights of the current pressure-stabilizing water pump under different fault types, and considering the probability of the pressure-stabilizing water pump failing when the monitoring device is abnormal, the similarity weights are combined to optimize the fault probability.

[0090] Specifically, firstly, based on the proportion of the occurrence frequency of different fault types in the total occurrence frequency of all fault types, the probability of occurrence of each fault type is obtained.

[0091] In a preferred embodiment of the present invention, the method for obtaining the occurrence probability includes: statistically analyzing the total frequency of all fault types that occur when any monitoring data in historical maintenance records is abnormal, and recording it as a first parameter; recording any fault type that occurs when any monitoring data is abnormal as a target fault type; recording the frequency of the target fault type as a second parameter of the target fault type; and recording the ratio of the second parameter to the first parameter as the occurrence probability of the target fault type.

[0092] As an optional embodiment, the specific method for calculating the probability of occurrence is as follows:

[0093]

[0094] In the formula, Indicates the first When an anomaly occurs in the monitoring data, the first one in the fault tree... The probability of occurrence of each fault type corresponding to a leaf node; Indicates the first The first parameter when monitoring data shows an anomaly; Indicates the first When an anomaly is found in a monitoring data point, the first one in the fault tree... The second parameter of each leaf node corresponds to the fault type.

[0095] Then, the occurrence probabilities of all fault types when all monitoring data show anomalies are summed and linearly normalized to obtain the new occurrence probabilities of all fault types in the fault tree. The new occurrence probabilities of fault types are then weighted and adjusted using similarity weights to obtain the weighted occurrence probabilities of the fault types.

[0096] As an optional embodiment, the specific method for calculating the weighted probability of occurrence for any fault type is as follows:

[0097]

[0098] In the formula, This represents the weighted probability of occurrence of the fault type; Indicates the probability of a new fault type occurring; The similarity weights represent the fault types.

[0099] For example, consider the case where the probabilities of all fault types occurring when all monitoring data shows anomalies are summed. For instance, monitoring data K1 has leaf nodes n1 and n2 in the fault tree, and monitoring data K2 has leaf nodes n1, n3, and n4. When monitoring data K1 is abnormal, the probability of fault types corresponding to its leaf nodes n1 and n2 is 0.1 and 0.9, respectively. When monitoring parameter K2 is abnormal, the probability of fault types corresponding to its leaf nodes n1, n3, and n4 is 0.1, 0.3, and 0.6, respectively. After summing the probabilities of fault types corresponding to all leaf nodes, the probabilities of fault types corresponding to leaf nodes n1, n2, n3, and n4 are 0.2, 0.9, 0.3, and 0.6, respectively, and then linear normalization is performed.

[0100] Thus, the weighted probability of occurrence is obtained through the above method.

[0101] Step S004: Use the weighted probability of occurrence of fault types and combine it with the fault tree to diagnose the faults of the pressure-stabilizing water pump.

[0102] Specifically, the fault tree is sorted in descending order based on the weighted probability of each fault type. When using the fault tree for fault analysis, the leaf nodes that appear earlier in the tree are checked first.

[0103] After sorting the fault types in the fault tree in descending order, the earlier a fault type appears in the sorted order, the more likely it is to be the fault type corresponding to the current pressure-stabilizing water pump failure. Therefore, it is more important to prioritize troubleshooting the pressure-stabilizing water pump when diagnosing its faults.

[0104] By following the above steps and optimizing the fault screening sequence, the efficiency of fault diagnosis can be greatly increased, enabling the rapid screening of accurate and genuine fault types, thus avoiding the need for simple screening based on historical experience or frequency, as is the case with traditional methods.

[0105] In addition, one embodiment of the present invention provides a system for diagnosing operational faults of electromechanical equipment, comprising: at least one memory for storing a program; and at least one processor for loading the program to execute steps S001 to S004 in the method for diagnosing operational faults of electromechanical equipment.

[0106] A storage medium storing processor-executable instructions, which, when executed by a processor, are used to implement steps S001 to S004 in the method for diagnosing operational faults of electromechanical equipment.

[0107] The aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors can be microprocessors or any conventional processor. It is worth noting that the processor can be a processor supporting Advanced Reduced Instruction Set Machines (ARM) architecture.

[0108] The storage medium can be volatile memory or non-volatile memory, or may include both. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which serves as an external cache. Many forms of RAM are available by way of example, but not limitation. Examples include Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Sync Link DRAM (SLDRAM), and Direct Rambus RAM (DRRAM).

[0109] This invention extracts fault feature vectors from historical maintenance records and combines them with disturbance testing to obtain enhanced fault feature vectors. It uses similarity weights to match the differences between the two, and dynamic disturbance testing amplifies the weak features of early faults (such as limited fluctuations in sensor data). Combined with historical feature vectors, it can identify early faults that are difficult to capture using traditional thresholding methods. Disturbance testing verifies the reliability of sensor data (fault characterization parameters), effectively eliminating environmental noise interference and avoiding misdiagnosis. Furthermore, based on fault tree analysis, it limits the scope of similarity weight calculation, focusing only on fault types related to the current abnormal monitoring data, and combines the historical frequency (statistical probability) of fault types with actual... The time similarity weight (dynamic matching degree) is used to calculate the weighted occurrence probability, which provides physical correlation constraints through fault tree and avoids interference from irrelevant fault types. The weighted probability takes into account both historical patterns and real-time features, improving the reliability of diagnosis, and has a stronger analytical capability for multiple coexisting or related faults (such as pressure fluctuations caused by abnormal motor power). Furthermore, by quantifying the priority of each fault type through the weighted occurrence probability, the elements in the feature vector are matched one-to-one with the monitoring device, which can locate the key fault source, thereby prioritizing the investigation of high-weight fault types, reducing blind maintenance time, and effectively improving the efficiency of fault diagnosis for pressure-stabilizing water pumps.

[0110] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for diagnosing operational faults in electromechanical equipment, characterized in that, The method includes the following steps: Several monitoring data of the pressure-stabilizing water pump are obtained using a monitoring device, the fault types and historical maintenance records of the pressure-stabilizing water pump are collected, and a fault tree is established based on the monitoring data and fault types. The historical maintenance records contain several maintenance events. By utilizing the similarity of monitoring data before and after maintenance for different fault types in all maintenance events in historical maintenance records, fault feature vectors for each fault type are obtained; disturbance tests are conducted on pressure-stabilizing water pumps, and the changes in monitoring data before and after the disturbance tests are analyzed to obtain fault characterization parameters and test vectors of the monitoring device; the test vectors are adjusted using the fault characterization parameters to obtain fault feature enhancement vectors; and the similarity weights of fault types are obtained through the fault feature vectors and fault feature enhancement vectors of the fault types. Based on the frequency of occurrence of each fault type in historical maintenance records, and combined with similarity weights, the weighted probability of occurrence of each fault type is obtained. Fault diagnosis of pressure-stabilized water pumps is performed by using the weighted probability of occurrence of fault types and combining it with fault trees. The acquisition of the fault feature vectors for each fault type includes: constructing a feature performance vector for a single maintenance event based on the DTW distance between any monitoring data collected before and after the maintenance event for each fault type in a single maintenance event; and obtaining the average feature performance vector of all maintenance events corresponding to each fault type as the fault feature vector for the fault type. The obtained fault characterization parameters and test vectors of the monitoring device include: When the monitoring data corresponding to each monitoring device of the pressure-stabilizing water pump exceeds the preset threshold range, the monitoring data is regarded as abnormal monitoring data, and a disturbance test item is set to test the pressure-stabilizing water pump; based on the difference in fluctuation amplitude between the disturbance monitoring data obtained from any monitoring device and the monitoring data before the test, the fault characterization parameter of the monitoring device is obtained, and the difference in fluctuation amplitude is negatively correlated with the fault characterization parameter. The disturbance monitoring data and pre-test monitoring data obtained from all monitoring devices with abnormal monitoring data are combined with the method for obtaining the feature performance vector to obtain the feature performance vector, which is denoted as the test vector.

2. The method for diagnosing operational faults in electromechanical equipment according to claim 1, characterized in that, The acquisition of the feature representation vector includes: For any fault type in any maintenance event, the degree of difference of the monitoring data is obtained based on the DTW distance between any monitoring data collected before and after the maintenance event for the fault type. An array of the degree of difference of all monitoring data for the corresponding fault type for the maintenance event is obtained as the feature representation vector of the maintenance event.

3. The method for diagnosing operational faults in electromechanical equipment according to claim 1, characterized in that, The method for adjusting the test vector using fault characterization parameters to obtain the fault feature enhancement vector includes the following specific methods: By multiplying the corresponding elements of the monitoring device in the test vector using the fault characterization parameters of any monitoring device, the adjusted values ​​of the corresponding elements of the monitoring device in the test vector are obtained, thereby obtaining the fault feature enhancement vector.

4. The method for diagnosing operational faults in electromechanical equipment according to claim 1, characterized in that, The specific method for obtaining the similarity weight of the fault type through the fault feature vector and the fault feature enhancement vector of the fault type includes: Obtain the fault type corresponding to all leaf nodes in the fault tree for the anomaly monitoring data, and denote it as the first fault type. Use the cosine similarity between the fault feature vector and the fault feature enhancement vector of the first fault type as the similarity weight of the first fault type.

5. The method for diagnosing operational faults in electromechanical equipment according to claim 1, characterized in that, The method for obtaining the weighted probability of occurrence of each fault type based on the frequency of occurrence of each fault type in historical maintenance records, combined with similarity weights, includes the following specific methods: The total frequency of all fault types that occur when any monitoring data in the historical maintenance records is abnormal is recorded as the first parameter. Any fault type that occurs when any monitoring data is abnormal is recorded as the target fault type. The frequency of the target fault type is recorded as the second parameter of the target fault type. The ratio of the second parameter to the first parameter is recorded as the probability of occurrence of the target fault type. The probabilities of occurrence of all fault types when all monitoring data are abnormal are superimposed and linearly normalized to obtain the new probability of occurrence of all fault types in the fault tree. The new probability of occurrence of fault types is weighted and adjusted using similarity weights to obtain the weighted probability of occurrence of the fault type.

6. The method for diagnosing operational faults in electromechanical equipment according to claim 1, characterized in that, The specific method for diagnosing faults in pressure-stabilized water pumps by using the weighted probability of fault types in conjunction with fault trees includes: The fault trees are sorted in descending order based on the weighted probability of each fault type, and the sorting result is used as the order for troubleshooting.

7. The method for diagnosing operational faults in electromechanical equipment according to claim 1, characterized in that, The specific methods for establishing a fault tree based on monitoring data and fault type are as follows: Based on the data types collected by all monitoring devices, a fault tree analysis method is used to construct a fault tree for the pressure-stabilizing water pump. The tree is constructed from top to bottom, with the top node representing each monitoring data point and the leaf nodes of each monitoring data point corresponding to several fault types.

8. A fault diagnosis system for electromechanical equipment, characterized in that, include: At least one memory for storing programs; At least one processor is configured to load the program to execute the operational fault diagnosis method for electromechanical equipment as described in any one of claims 1-7.

9. A storage medium storing processor-executable instructions, characterized in that: The processor-executable instructions, when executed by the processor, are used to implement the operational fault diagnosis method for electromechanical equipment as described in any one of claims 1-7.

Images