Method and system for accurately determining fault of electric propulsion system

Abstract

The invention discloses a fault accurate determination method and system of an electric propulsion system. The method comprises the following steps: acquiring historical fault data of each first component in the ship electric propulsion system, and constructing a fault tree of the ship electric propulsion system according to the historical fault data; according to the fault tree, through causal analysis, constructing a first Petri network model; when any first fault event occurs, initializing each first library in a first Petri network model according to the first fault event occurring at the current moment; according to the current operation data of the first components and the first libraries, triggering transitions in a first Petri network model are activated, and a first Petri network model corresponding to the next moment is obtained; and determining a first fault event occurring at the next moment according to each first place in the first Petri network model corresponding to the next moment. According to the invention, the fault blocking response speed after the fault event occurs can be improved, and meanwhile, the subsequent fault prediction accuracy is improved.

Description

Technical Field

[0001] The present application relates to the field of fault detection, and in particular to a method and system for accurately determining faults of an electric propulsion system. Background Art

[0002] As the complexity of ship electric propulsion systems continues to increase, the need for fault diagnosis and prediction is becoming increasingly important. Traditional fault diagnosis methods mainly rely on manual experience or are based on simple rules and threshold detection, and are often difficult to deal with multiple faults and multi-variable relationships in complex systems. Since the ship electric propulsion system involves multiple interrelated key components, such as generators, transmission lines, motors, control systems, etc., and the ship electric propulsion system has limited component installation space compared to other power systems, when one of the components fails, it may trigger a series of chain reactions, causing other normally operating components to fail in turn, thereby causing a global failure of the entire system.

[0003] The prior art usually monitors the operating data of each component in real time when the ship's electric propulsion system is operating normally, and predicts the upcoming fault events. When an abnormality occurs for the first time, an alarm is immediately issued to prevent the fault from occurring. However, when the above method fails to predict and some fault events have occurred in the ship's electric propulsion system but do not cause the ship's electric propulsion system to be paralyzed, the prediction method of the prior art often has a poor prediction effect on the occurrence of subsequent faults. At this time, if the subsequent fault prediction cannot be performed for the chain reaction of the fault, the fault will continue to spread until the entire ship's electric propulsion system is paralyzed. Summary of the invention

[0004] The present application provides a method and system for accurately determining faults of an electric propulsion system to solve the technical problem of how to quickly and accurately determine the next fault that may occur when a fault event has occurred in the ship's electric propulsion system, thereby preventing the fault from propagating between different components.

[0005] In order to solve the above technical problems, in a first aspect, an embodiment of the present application provides a method for accurately determining a fault of an electric propulsion system, comprising: Acquire historical fault data of each first component in the ship electric propulsion system, and construct a fault tree of the ship electric propulsion system according to the historical fault data; the nodes of the fault tree include a plurality of first fault events of each first component and potential propagation paths between the first fault events; each of the potential propagation paths includes a plurality of the first fault events; According to the fault tree, a first Petri network model is constructed through causal analysis; wherein the first Petri network model includes a first library representing each of the first fault events and a plurality of transitions obtained through causal analysis for representing the transfer conditions of the first fault events; the first library and the transitions are connected by arcs obtained through the potential propagation path construction; When any of the first fault events occurs, initializing each of the first places in the first Petri network model according to the first fault event currently occurring; According to the current operation data of each of the first components and each of the first places, activating the triggerable transitions in the first Petri network model, and obtaining the first Petri network model corresponding to the next moment; A first fault event occurring at a next moment is determined according to each of the first places in the first Petri network model corresponding to the next moment.

[0006] Compared with the prior art, the embodiments of the present application have the following beneficial effects: by constructing a fault tree to systematically identify the failure modes of each component and their potential propagation paths, and further constructing a Petri network model based on the fault tree, the network structure of the Petri network model has better interpretability and accuracy; at the same time, because the Petri network model requires lower computing costs and higher real-time response efficiency than the network model based on deep learning, it is possible to block the propagation of the fault more promptly when a fault occurs in the ship's electric propulsion system; finally, in order to improve the accuracy of the Petri network model's prediction of the system state at the next moment, the triggering conditions of the transition are improved through a causal analysis method, thereby improving the accuracy of the Petri network model's state transfer, and further improving the accuracy of the prediction of the fault event at the next moment.

[0007] In some embodiments of the first aspect of the present application, constructing a first Petri network model according to the fault tree through causal analysis includes: Constructing a first library corresponding to each of the first fault events; By causal analysis, combining the historical operation data of the first components corresponding to the first fault events connected in each of the potential propagation paths, constructing the transitions between the first places corresponding to the first fault events connected; The first place corresponding to the connected first fault event is connected to the transition between the first place corresponding to the connected first fault event to construct an arc.

[0008] Compared with the prior art, the above embodiment has the following beneficial effects: since each component may have more than one type of fault event, and different fault events are caused by different causes, the first fault event is modeled as the first library place of the Petri network model, rather than modeling each component in the system as the first library place of the Petri network model. This can effectively improve the simulation accuracy of the fault propagation process, while reducing the computational complexity of subsequent transitions constructed based on causal analysis, and improving the response speed of the Petri network model.

[0009] In some embodiments of the first aspect of the present application, the causal analysis is combined with the historical operation data of the first component corresponding to the first fault event connected in each of the potential propagation paths to construct the transition between the first places corresponding to the first fault events connected, including: The first storage places on both sides of each transition include the first input storage place on one side and the first output storage place on the other side; the historical operation data include: the first historical operation data corresponding to the first component when the first fault event occurs and the second historical operation data corresponding to the first component when the first fault event does not occur; For the transition to be constructed, preprocessing the first historical operation data corresponding to the first input library and the second historical operation data corresponding to the first output library; A first regression model is fitted according to the preprocessed first historical operation data and the preprocessed second historical operation data, and the corresponding transition to be constructed is constructed according to the first regression model.

[0010] Compared with the prior art, the above embodiment has the following beneficial effects: when the transition is triggered, the input library representing the transition is experiencing the corresponding first fault event, and the first fault event corresponding to the first input library causes the occurrence of the first fault event corresponding to the subsequent first output library. Therefore, the factors affecting the operation state of the first component corresponding to the subsequent first output library include not only the first component corresponding to the first input library, but also the operation data of the first component corresponding to the first output library that is about to fail. Therefore, when fitting the first regression model, multiple factors are comprehensively considered to improve the accuracy of the first regression model. At the same time, the triggering conditions of the transition in the Petri network model are improved through the first regression model, thereby improving the accuracy of the Petri network model state transfer, and further improving the accuracy of the next moment fault event prediction.

[0011] In some embodiments of the first aspect of the present application, fitting a first regression model according to the preprocessed first historical operation data and the preprocessed second historical operation data, and constructing the corresponding transition to be constructed according to the first regression model, includes: Using the preprocessed first historical operation data and the preprocessed second historical operation data as input variables of the first regression model, and using the probability of the first component corresponding to the first output library having the first fault event corresponding to the first output library as an output variable, to fit the first regression model; The transition to be constructed is constructed according to a preset first threshold and the first regression model.

[0012] Compared with the prior art, the above embodiment has the following beneficial effects: by fitting the relationship between the operating data of the first component corresponding to the first library on both sides of the transition and the probability of occurrence of the first fault event corresponding to the first output library, combined with the judgment operation of the first threshold, the accuracy of the transition for fault propagation simulation is improved.

[0013] In some embodiments of the first aspect of the present application, activating triggerable transitions in the first Petri network model according to current operating data of each of the first components and each of the first places, and obtaining the first Petri network model corresponding to the next moment, includes: Selecting a transition that has a token in the first input place and has no token in the first output place from all the transitions as a first transition; wherein when a token exists in the first place, it indicates that a first fault event corresponding to the first place is occurring; Inputting the current operation data of the first component corresponding to the first fault event corresponding to the first library on both sides of the first transition into the corresponding first regression model; When the output value of the first regression model is greater than the first threshold, the first transition corresponding to the first transition is activated to add a token in the first output library of the first transition.

[0014] Compared with the prior art, the above embodiment has the following beneficial effects: when a first fault event occurs, it is determined whether there is a token in all the first input locations of the transition, and no token exists in the first output location. If so, it is proved that the transition has the basic conditions for triggering. Subsequently, the probability of the occurrence of the first fault event corresponding to the first input location is evaluated through the operation data of the first component corresponding to the first location on both sides of the transition, and the first threshold is combined to determine whether the transition needs to be triggered, thereby ensuring the effectiveness of the transition triggering. When the transition is triggered, a token is added to the subsequent first output location to prevent the repeated triggering of the transition during subsequent fault prediction, resulting in an error in the state transfer of the Petri network model.

[0015] In some embodiments of the first aspect of the present application, when any first transition is activated, it also includes: maintaining the tokens in the first places on both sides of the first transition unchanged until the first fault events on both sides corresponding to the first places are resolved.

[0016] Compared with the prior art, the above embodiment has the following beneficial effects: in order to improve the accuracy of the Petri network model in the state transfer process, usually in the traditional Petri network model, when the transition is triggered, the token in the first input library of the transition will change. However, since fault propagation is a transient process and the occurrence of a fault event is a continuous process, in order to make the Petri network model better simulate the fault propagation process of the ship's electric propulsion system, the token change strategy in the first library after the transition is triggered is improved to improve the accuracy of fault propagation prediction.

[0017] In some embodiments of the first aspect of the present application, determining the first fault event occurring at the next moment based on each of the first library locations in the first Petri network model corresponding to the next moment includes: determining that the first fault event corresponding to the first library location in which the token exists in the first Petri network model corresponding to the next moment is the first fault event occurring at the next moment.

[0018] Compared with the prior art, the above embodiment has the following beneficial effects: each time the first Petri network model is activated by transition, the entire first Petri network model undergoes a state transfer. At this time, by counting whether each first library contains a token, the first fault event that occurs at the next moment is determined, thereby improving the response speed of fault prediction at the next moment during the fault propagation process.

[0019] In some embodiments of the first aspect of the present application, the acquiring historical fault data of each first component in the ship electric propulsion system, and constructing a fault tree of the ship electric propulsion system according to the historical fault data, includes: The event that the ship electric propulsion system stops operating due to a fault is taken as the top event of the fault tree; extracting each of the first fault events of each of the first components from the historical fault data, and using the first fault events as sub-events of the fault tree; According to the occurrence order of each of the sub-events when the top event in the historical fault data occurs, the potential propagation path is constructed to obtain the fault tree.

[0020] Compared with the prior art, the above embodiment has the following beneficial effects: since the first fault event will eventually reach the event that the ship's electric propulsion system stops running due to a fault if the propagation process is not blocked during the propagation process, the event that the ship's electric propulsion system stops running due to a fault can be used as the top event of the fault tree, and each first fault event can be used as a sub-event. Further, through the statistical analysis method, the order and frequency of faults in the historical fault data are used as the basis to construct the connection relationship between each sub-event, thereby obtaining the potential propagation path of the first fault event, providing an accurate construction basis for the subsequent Petri network model.

[0021] In a second aspect, an embodiment of the present application further provides a system for accurately determining faults of an electric propulsion system, comprising: a first data acquisition module, a network construction module, an initialization module, a model state transfer module, and a fault determination module; The first data acquisition module is used to acquire historical fault data of each first component in the ship electric propulsion system, and construct a fault tree of the ship electric propulsion system according to the historical fault data; the nodes of the fault tree include a number of first fault events of each first component and potential propagation paths between the first fault events; each of the potential propagation paths includes a number of the first fault events; The network construction module is used to construct a first Petri network model according to the fault tree through causal analysis; wherein the first Petri network model includes a first library representing each of the first fault events and a number of transitions obtained through causal analysis and used to represent the transfer conditions of the first fault events; the first library and the transitions are connected by arcs obtained through the potential propagation path construction; The initialization module is used to initialize each of the first libraries in the first Petri network model according to the first fault event currently occurring when any of the first fault events occurs; The model state transfer module is used to activate the triggerable transition in the first Petri network model according to the current operation data of each of the first components and each of the first libraries, and obtain the first Petri network model corresponding to the next moment; The fault determination module is used to determine the first fault event occurring at the next moment according to each of the first libraries in the first Petri network model corresponding to the next moment.

[0022] In some embodiments of the second aspect of the present application, constructing a first Petri network model according to the fault tree through causal analysis includes: Constructing a first library corresponding to each of the first fault events; By causal analysis, combining the historical operation data of the first components corresponding to the first fault events connected in each of the potential propagation paths, constructing the transitions between the first places corresponding to the first fault events connected; The first place corresponding to the connected first fault event is connected to the transition between the first place corresponding to the connected first fault event to construct an arc. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic flow chart of a method for accurately determining a fault of an electric propulsion system provided in some embodiments of the present application; Figure 2 A fault tree diagram provided in some embodiments of the present application; Figure 3 This is a schematic diagram of the state transfer process of the first Petri network model in some embodiments of the present application; Figure 4 This is a schematic diagram of the structure of a system for accurately determining faults of an electric propulsion system provided in some embodiments of the present application. DETAILED DESCRIPTION

[0024] The prior art usually monitors the operating data of each component in real time when the ship's electric propulsion system is operating normally, and predicts the upcoming fault events. When an abnormality occurs for the first time, an alarm is immediately issued to prevent the fault from occurring. However, when the above method fails to predict and some fault events have occurred in the ship's electric propulsion system but do not cause the ship's electric propulsion system to be paralyzed, the prediction method of the prior art often has a poor prediction effect on the occurrence of subsequent faults. At this time, if the subsequent fault prediction cannot be performed for the chain reaction of the fault, the fault will continue to spread until the entire ship's electric propulsion system is paralyzed.

[0025] In order to solve the above technical problems, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0026] Embodiment 1 Please refer to Figure 1 , a method for accurately determining faults of an electric propulsion system provided in an embodiment of the present application, includes S10 to S50, specifically: S10: Acquire historical fault data of each first component in the ship electric propulsion system, and construct a fault tree of the ship electric propulsion system based on the historical fault data; the nodes of the fault tree include a number of first fault events of each first component and potential propagation paths between the first fault events; each of the potential propagation paths includes a number of the first fault events.

[0027] Preferably, in some embodiments of the present application, the historical fault data includes all fault events that occurred in various components when the ship's electric propulsion system fails and stops operating, as well as the time points when these fault events occurred.

[0028] Further, in some embodiments of the present application, the acquiring of historical fault data of each first component in the ship electric propulsion system and constructing a fault tree of the ship electric propulsion system according to the historical fault data includes: The event that the ship electric propulsion system stops operating due to a fault is taken as the top event of the fault tree; extracting each of the first fault events of each of the first components from the historical fault data, and using the first fault events as sub-events of the fault tree; According to the occurrence order of each of the sub-events when the top event in the historical fault data occurs, the potential propagation path is constructed to obtain the fault tree.

[0029] Compared with the prior art, the above embodiment has the following beneficial effects: since the first fault event will eventually reach the event that the ship's electric propulsion system stops running due to a fault if the propagation process is not blocked during the propagation process, the event that the ship's electric propulsion system stops running due to a fault can be used as the top event of the fault tree, and each first fault event can be used as a sub-event. Further, through the statistical analysis method, the order and frequency of faults in the historical fault data are used as the basis to construct the connection relationship between each sub-event, thereby obtaining the potential propagation path of the first fault event, providing an accurate construction basis for the subsequent Petri network model.

[0030] Preferably, in some embodiments of the present application, constructing the potential propagation path and obtaining the fault tree according to the occurrence order of each of the sub-events when the top event in the historical fault data occurs includes: S11: extracting all sub-events that occurred before each top event and the order in which the sub-events occurred from the historical fault data; S12: Determine the logical relationship between the sub-events according to the order in which the sub-events occur, and construct an initial fault tree according to the logical relationship. At this time, the propagation path formed by the fault tree includes a large number of repeated sub-events; S13: Simplify the fault tree constructed in S12 through Boolean algebra theorem to obtain a fault tree with non-repeated sub-events.

[0031] like Figure 2 As shown, Figure 2 a is the fault tree before simplification, where Represents the top event, For sub-events, For logical AND, is the logical OR. Figure 2 b is the fault tree simplified by Boolean algebra theorem. Figure 2 a, where due to is a necessary condition for the top event to occur, so the left branch can be simplified. and The branches formed together may represent another first fault event in actual operation. ,Right now and When they occur together, it will lead to occur. Figure 2 The fault tree in the figure is only an example. The fault tree corresponding to the actual ship electric propulsion system is more complex.

[0032] S20: According to the fault tree, a first Petri network model is constructed through causal analysis; wherein the first Petri network model includes a first library representing each of the first fault events and a number of transitions obtained through causal analysis for representing the transfer conditions of the first fault events; the first library and the transitions are connected by arcs obtained through the potential propagation path construction.

[0033] Furthermore, in some embodiments of the present application, constructing a first Petri network model according to the fault tree through causal analysis includes: Constructing a first library corresponding to each of the first fault events; By causal analysis, combining the historical operation data of the first components corresponding to the first fault events connected in each of the potential propagation paths, constructing the transitions between the first places corresponding to the first fault events connected; The first place corresponding to the connected first fault event is connected to the transition between the first place corresponding to the connected first fault event to construct an arc.

[0034] Since each component may have more than one type of fault event and different fault events have different causes, modeling the first fault event as the first place of the Petri network model instead of modeling each component in the system as the first place of the Petri network model can effectively improve the simulation accuracy of the fault propagation process, while reducing the computational complexity of subsequent transitions constructed based on causal analysis and improving the response speed of the Petri network model.

[0035] Preferably, in some embodiments of the present application, the first Petri network model also includes a second place, wherein the second place represents the normal operating status of each first component.

[0036] In order to better illustrate the structure of the first Petri network model, Figure 2 The fault tree shown in the figure gives the corresponding first Petri network model, assuming The corresponding part is the first part , The corresponding part is the first part , The corresponding part is the first part , then the structure of the first Petri network model is as follows Figure 3 The first library includes , corresponding to Figure 2 The first fault event in The second library is , corresponding to the first component , first component and the first component . Figure 3 There are also libraries , represents whether the ship's electric propulsion system stops operating due to a fault. Further, Figure 3 The first Petri network model shown in FIG. 1 also includes several transitions, namely , where the improved changes include , ,in, and are all changes in the embedded regression model, but and The difference is, The output variable of the corresponding regression model is the probability of the ship's electric propulsion system failing and stopping operation, and the input variables only include the first component and the first component The current running data of The corresponding regression model is also obtained through training with the above-mentioned input and output related historical operation data. Figure 3When a token (ie ·) appears in the first place, it means that the first fault event corresponding to the first place is occurring, and when the second place If a token appears in the , it means that each first component is operating normally, and the tokens of the second place and the first place cannot appear.

[0037] Further, in some embodiments of the present application, the causal analysis is combined with the historical operation data of the first components corresponding to the first fault events connected in each of the potential propagation paths to construct the transitions between the first places corresponding to the first fault events connected, including: The first storage places on both sides of each transition include the first input storage place on one side and the first output storage place on the other side; the historical operation data include: the first historical operation data corresponding to the first component when the first fault event occurs and the second historical operation data corresponding to the first component when the first fault event does not occur; For the transition to be constructed, preprocessing the first historical operation data corresponding to the first input library and the second historical operation data corresponding to the first output library; A first regression model is fitted according to the preprocessed first historical operation data and the preprocessed second historical operation data, and the corresponding transition to be constructed is constructed according to the first regression model.

[0038] Since when the transition is triggered, the input library representing the transition is experiencing the corresponding first fault event, and the first fault event corresponding to the first input library causes the occurrence of the first fault event corresponding to the subsequent first output library, the factors affecting the operation state of the first component corresponding to the subsequent first output library include not only the first component corresponding to the first input library, but also the operation data of the first component corresponding to the first output library that is about to fail. Therefore, when fitting the first regression model, multiple factors are comprehensively considered to improve the accuracy of the first regression model. At the same time, the triggering conditions of the transition in the Petri network model are improved through the first regression model, thereby improving the accuracy of the Petri network model state transfer, and further improving the accuracy of the next moment fault event prediction.

[0039] Preferably, in some embodiments of the present application, the operating data of the first component includes but is not limited to signal data such as temperature, pressure, current, amplitude, etc.

[0040] Preferably, in some embodiments of the present application, when preprocessing the data, the data needs to be normalized without performing a denoising operation, so as to more accurately extract abnormal data.

[0041] Preferably, in some embodiments of the application, the fitting method of the first regression model includes but is not limited to linear regression, multivariate regression and nonlinear regression. The present application does not limit the specific method of fitting the first regression model, but only limits the input and output variables during fitting.

[0042] Further, in some embodiments of the present application, fitting a first regression model according to the preprocessed first historical operation data and the preprocessed second historical operation data, and constructing the corresponding transition to be constructed according to the first regression model, includes: Using the preprocessed first historical operation data and the preprocessed second historical operation data as input variables of the first regression model, and using the probability of the first component corresponding to the first output library having the first fault event corresponding to the first output library as an output variable, to fit the first regression model; The transition to be constructed is constructed according to a preset first threshold and the first regression model.

[0043] By fitting the relationship between the operating data of the first component corresponding to the first library on both sides of the transition and the probability of occurrence of the first fault event corresponding to the first output library, combined with the determination operation of the first threshold, the accuracy of the transition for fault propagation simulation is improved.

[0044] Preferably, in some embodiments of the present invention, the transition to be constructed is constructed according to a preset first threshold and the first regression model, specifically: Change activation judgment conditions:

[0045] in, is the first threshold, is the first regression model, as well as Respectively represent the operating data of the first component corresponding to the first location on both sides of the transition.

[0046] S30: When any of the first fault events occurs, initializing each of the first places in the first Petri network model according to the first fault event currently occurring.

[0047] The process of initializing each of the first places in the first Petri network model is to add a token to the first place according to the occurrence of the current first fault event.

[0048] S40: activating triggerable transitions in the first Petri network model according to the current operation data of each of the first components and each of the first places, and obtaining the first Petri network model corresponding to the next moment.

[0049] Further, in some embodiments of the present application, activating the triggerable transition in the first Petri network model according to the current operation data of each of the first components and each of the first places, and obtaining the first Petri network model corresponding to the next moment, includes: Selecting a transition that has a token in the first input place and has no token in the first output place from all the transitions as a first transition; wherein when a token exists in the first place, it indicates that a first fault event corresponding to the first place is occurring; Inputting the current operation data of the first component corresponding to the first fault event corresponding to the first library on both sides of the first transition into the corresponding first regression model; When the output value of the first regression model is greater than the first threshold, the first transition corresponding to the first transition is activated to add a token in the first output library of the first transition.

[0050] When the first fault event occurs, it is determined whether there is a token in the first input library of the transition, and no token exists in the first output library. If so, it proves that the transition has the basic conditions for triggering. Subsequently, the operating data of the first component corresponding to the first library on both sides of the transition is used to evaluate the probability of the first fault event corresponding to the first input library, and combined with the first threshold, it is determined whether the transition needs to be triggered, thereby ensuring the effectiveness of the transition triggering. When the transition is triggered, a token is added to the subsequent first output library to prevent the repeated triggering of the transition during subsequent fault prediction, resulting in errors in the state transfer of the Petri network model.

[0051] Preferably, in some embodiments of the present application, the first transition specifically refers to Figure 3 The transition that needs to be improved is shown in , that is, the storage places on both sides of the transition are the first storage places. Figure 3 middle The changes shown are also improved by the regression model, but due to The output place does not belong to the first place, and the problem solved by this application is how to block the occurrence of subsequent first fault events when the first fault event occurs. However, when When triggered, the subsequent event that the ship's electric propulsion system stops running due to a fault has been triggered, so the above-mentioned first transition does not include the transition in which the fault event represented by the two side libraries is not the first fault event.

[0052] S50: Determine a first fault event occurring at a next moment according to each of the first places in the first Petri network model corresponding to the next moment.

[0053] Further, in some embodiments of the present application, determining the first fault event occurring at the next moment based on each of the first library locations in the first Petri network model corresponding to the next moment includes: determining that the first fault event corresponding to the first library location in which the token exists in the first Petri network model corresponding to the next moment is the first fault event occurring at the next moment.

[0054] Each time the first Petri network model is activated, the state of the entire first Petri network model is transferred. At this time, by counting whether each first library contains a token, the first fault event that occurs at the next moment is determined, thereby improving the response speed of fault prediction at the next moment during the fault propagation process.

[0055] Furthermore, in some embodiments of the present application, when any first transition is activated, it also includes: maintaining the tokens in the first places on both sides of the first transition unchanged until the first fault events on both sides corresponding to the first places are resolved.

[0056] In order to improve the accuracy of the Petri network model in the state transfer process, usually in the traditional Petri network model, when the transition is triggered, the token in the first input library of the transition will change. However, since fault propagation is a transient process and the occurrence of a fault event is a continuous process, in order to make the Petri network model better simulate the fault propagation process of the ship's electric propulsion system, the token change strategy in the first library after the transition is triggered is improved to improve the accuracy of fault propagation prediction.

[0057] Preferably, in some embodiments of the present application, by real-time monitoring of the operating status of each first component of the ship electric propulsion system, when the first component changes from a fault state to a normal operation, a corresponding transition is triggered. Figure 3 In, when Corresponding to the first component The first failure event At the end, the transition is triggered Another case is triggering a transition , where changes When triggered, There should be a token inside, and The triggering condition is that after the first fault events of all the first components are repaired, the ship's electric propulsion system recovers from the fault stop state, then the activation .

[0058] In summary, the method for accurately determining faults of an electric propulsion system provided in the present application has the following beneficial effects: by constructing a fault tree to systematically identify the fault modes of each component and their potential propagation paths, and further constructing a Petri network model based on the fault tree, the network structure of the Petri network model has better interpretability and accuracy; at the same time, because the Petri network model has lower computing costs and higher real-time response efficiency than the network model based on deep learning, it is able to block the propagation of the fault more promptly when a fault occurs in the ship's electric propulsion system; finally, in order to improve the accuracy of the Petri network model's prediction of the system state at the next moment, the triggering conditions of the transition are improved through a causal analysis method, thereby improving the accuracy of the Petri network model's state transfer, and further improving the accuracy of the prediction of the fault event at the next moment.

[0059] Embodiment 2 refer to Figure 4 , which is a fault accurate determination system for an electric propulsion system provided in some embodiments of the present application, comprising: a first data acquisition module 11, a network construction module 12, an initialization module 13, a model state transfer module 14 and a fault determination module 15.

[0060] Furthermore, in some embodiments of the present application, the first data acquisition module 11 is used to acquire historical fault data of each first component in the ship electric propulsion system, and construct a fault tree of the ship electric propulsion system based on the historical fault data; the nodes of the fault tree include a number of first fault events of each first component and potential propagation paths between each first fault event; each of the potential propagation paths includes a number of first fault events; the network construction module 12 is used to construct a first Petri network model based on the fault tree through causal analysis; wherein the first Petri network model includes a first library representing each first fault event and a number of transitions representing the transfer conditions of the first fault event obtained through causal analysis; the first library and the The arc connection obtained by the transition through the potential propagation path construction; the first library and the arc connection obtained by the transition through the potential propagation path construction; the initialization module 13 is used to initialize each of the first libraries in the first Petri network model according to the first fault event occurring at the current moment when any of the first fault events occurs; the model state transfer module 14 is used to activate the triggerable transitions in the first Petri network model according to the current operating data of each of the first components and each of the first libraries, and obtain the first Petri network model corresponding to the next moment; the fault determination module 15 is used to determine the first fault event occurring at the next moment according to each of the first libraries in the first Petri network model corresponding to the next moment.

[0061] Furthermore, in some embodiments of the present application, the first Petri network model is constructed based on the fault tree through causal analysis, including: constructing a first library corresponding to each of the first fault events; constructing the transitions between the first libraries corresponding to the connected first fault events in combination with the historical operation data of the first components corresponding to the connected first fault events in each of the potential propagation paths through causal analysis; connecting the first libraries corresponding to the connected first fault events with the transitions between the first libraries corresponding to the connected first fault events to construct an arc.

[0062] Further, in some embodiments of the present application, the causal analysis is combined with the historical operation data of the first component corresponding to the first fault event connected in each of the potential propagation paths to construct the transition between the first storage places corresponding to the connected first fault events, including: wherein the first storage places on both sides of each transition include the first input storage place on one side and the first output storage place on the other side; the historical operation data include: the first historical operation data corresponding to the first component when the first fault event occurs and the second historical operation data corresponding to the first component when the first fault event does not occur; for the transition to be constructed, pre-process the first historical operation data corresponding to the first input storage place and the second historical operation data corresponding to the first output storage place; fit the first regression model according to the pre-processed first historical operation data and the pre-processed second historical operation data, and construct the corresponding transition to be constructed according to the first regression model.

[0063] Furthermore, in some embodiments of the present application, the first regression model is fitted according to the preprocessed first historical operation data and the preprocessed second historical operation data, and the corresponding transition to be constructed is constructed according to the first regression model, including: taking the preprocessed first historical operation data and the preprocessed second historical operation data as input variables of the first regression model, taking the probability of the first fault event corresponding to the first output library occurring in the first component corresponding to the first output library as the output variable, and fitting the first regression model; and constructing the transition to be constructed according to a preset first threshold and the first regression model.

[0064] Further, in some embodiments of the present application, according to the current operating data of each of the first components and each of the first libraries, the triggerable transitions in the first Petri network model are activated to obtain the first Petri network model corresponding to the next moment, including: selecting from all the transitions the transitions that have tokens in the corresponding first input libraries and have no tokens in the corresponding first output libraries as the first transition; wherein, when there is a token in the first library, it means that the first fault event corresponding to the first library is occurring; the current operating data of the first component corresponding to the first fault event in the first libraries on both sides of the first transition is input into the corresponding first regression model; when the output value of the first regression model is greater than the first threshold, activating the corresponding first transition to add a token in the first output library of the first transition.

[0065] Furthermore, in some embodiments of the present application, when any first transition is activated, it also includes: maintaining the tokens in the first places on both sides of the first transition unchanged until the first fault events on both sides corresponding to the first places are resolved.

[0066] Further, in some embodiments of the present application, determining the first fault event occurring at the next moment based on each of the first library locations in the first Petri network model corresponding to the next moment includes: determining that the first fault event corresponding to the first library location in which the token exists in the first Petri network model corresponding to the next moment is the first fault event occurring at the next moment.

[0067] Furthermore, in some embodiments of the present application, the historical fault data of each first component in the ship electric propulsion system is obtained, and a fault tree of the ship electric propulsion system is constructed based on the historical fault data, including: taking the event that the ship electric propulsion system stops running due to a fault as the top event of the fault tree; extracting each of the first fault events of each of the first components from the historical fault data, and taking the first fault event as a sub-event of the fault tree; and constructing the potential propagation path according to the order of occurrence of each of the sub-events when the top event in the historical fault data occurs, to obtain the fault tree.

[0068] It can be understood that the above-mentioned system item embodiments correspond to the method item embodiments of the present invention. The electric propulsion system fault accurate determination system provided by the embodiment of the present invention can implement any method item embodiment of the present invention, that is, the electric propulsion system fault accurate determination method provided in embodiment one.

[0069] Embodiment 3 Based on the above-mentioned embodiment of the method for accurately determining faults of the electric propulsion system, another embodiment of the present application provides a terminal device for accurately determining faults of the electric propulsion system, the terminal device for accurately determining faults of the electric propulsion system comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, the method for accurately determining faults of the electric propulsion system of any embodiment of the present application is implemented.

[0070] Exemplarily, in this embodiment, the computer program may be divided into one or more modules, and the one or more modules are stored in the memory and executed by the processor to complete the present application. The one or more modules may be a series of computer program instruction segments capable of completing specific functions, and the instruction segments are used to describe the execution process of the computer program in the fault accurate determination device of the electric propulsion system.

[0071] The electric propulsion system fault accurate determination device may be a computing device such as a desktop computer, a notebook, a palm computer, a cloud server, etc. The electric propulsion system fault accurate determination terminal device may include, but is not limited to, a processor and a memory.

[0072] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc. The processor is the control center of the fault accurate determination device of the electric propulsion system, and uses various interfaces and lines to connect the various parts of the fault accurate determination device of the entire electric propulsion system. The memory may be used to store the computer program and / or module, and the processor realizes various functions of the fault accurate determination device of the electric propulsion system by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function, etc.; the data storage area may store data created according to the use of the mobile phone, etc. In addition, the memory may include high-speed random access memory and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0073] Embodiment 4 Based on the above-mentioned embodiment of the method for accurately determining faults of the electric propulsion system, another embodiment of the present application provides a storage medium, wherein the storage medium includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the method for accurately determining faults of the electric propulsion system of any embodiment of the present application.

[0074] In this embodiment, the storage medium is a computer-readable storage medium, and the computer program includes computer program code, which may be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0075] The specific embodiments described above further describe the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of protection of the present application.

Claims

1. A method for accurately determining faults in an electric propulsion system, characterized in that: include: Acquire historical fault data of each first component in the ship electric propulsion system, and construct a fault tree of the ship electric propulsion system according to the historical fault data; The nodes of the fault tree include a plurality of first fault events of each of the first components and potential propagation paths between the first fault events; each of the potential propagation paths includes a plurality of the first fault events; According to the fault tree, a first Petri network model is constructed through causal analysis; wherein the first Petri network model includes a first library representing each of the first fault events and a plurality of transitions obtained through causal analysis for representing the transfer conditions of the first fault events; the first library and the transitions are connected by arcs obtained through the potential propagation path construction; When any of the first fault events occurs, initializing each of the first places in the first Petri network model according to the first fault event currently occurring; According to the current operation data of each of the first components and each of the first places, activating the triggerable transitions in the first Petri network model, and obtaining the first Petri network model corresponding to the next moment; A first fault event occurring at a next moment is determined according to each of the first places in the first Petri network model corresponding to the next moment.

2. A method for accurately determining faults of an electric propulsion system according to claim 1, characterized in that: According to the fault tree, a first Petri network model is constructed through cause and effect analysis, including: Constructing a first library corresponding to each of the first fault events; By causal analysis, combining the historical operation data of the first components corresponding to the first fault events connected in each of the potential propagation paths, constructing the transitions between the first places corresponding to the first fault events connected; The first place corresponding to the connected first fault event is connected to the transition between the first place corresponding to the connected first fault event to construct an arc.

3. A method for accurately determining faults of an electric propulsion system according to claim 2, characterized in that: The causal analysis is combined with the historical operation data of the first components corresponding to the first fault events connected in each of the potential propagation paths to construct the transitions between the first places corresponding to the first fault events connected. include: The first storage places on both sides of each transition include the first input storage place on one side and the first output storage place on the other side; the historical operation data include: the first historical operation data corresponding to the first component when the first fault event occurs and the second historical operation data corresponding to the first component when the first fault event does not occur; For the transition to be constructed, preprocessing the first historical operation data corresponding to the first input library and the second historical operation data corresponding to the first output library; A first regression model is fitted according to the preprocessed first historical operation data and the preprocessed second historical operation data, and the corresponding transition to be constructed is constructed according to the first regression model.

4. A method for accurately determining faults of an electric propulsion system according to claim 3, characterized in that: The fitting of a first regression model according to the preprocessed first historical operation data and the preprocessed second historical operation data, and constructing the corresponding transition to be constructed according to the first regression model, includes: Using the preprocessed first historical operation data and the preprocessed second historical operation data as input variables of the first regression model, and using the probability of the first component corresponding to the first output library having the first fault event corresponding to the first output library as an output variable, to fit the first regression model; The transition to be constructed is constructed according to a preset first threshold and the first regression model.

5. A method for accurately determining faults of an electric propulsion system according to claim 4, characterized in that: The activating triggerable transitions in the first Petri network model according to the current operation data of each of the first components and each of the first places, and obtaining the first Petri network model corresponding to the next moment, includes: Selecting a transition that has a token in the first input place and has no token in the first output place from all the transitions as a first transition; wherein when a token exists in the first place, it indicates that a first fault event corresponding to the first place is occurring; Inputting the current operation data of the first component corresponding to the first fault event corresponding to the first library on both sides of the first transition into the corresponding first regression model; When the output value of the first regression model is greater than the first threshold, the first transition corresponding to the first transition is activated to add a token in the first output library of the first transition.

6. A method for accurately determining faults of an electric propulsion system according to claim 5, characterized in that: When any first transition is activated, the method further includes: maintaining the tokens in the first places on both sides of the first transition unchanged until the first fault events on both sides corresponding to the first places are resolved.

7. A method for accurately determining faults of an electric propulsion system according to claim 5, characterized in that: Determining the first fault event occurring at the next moment according to each of the first library places in the first Petri network model corresponding to the next moment includes: determining that the first fault event corresponding to the first library place where the token exists in the first Petri network model corresponding to the next moment is the first fault event occurring at the next moment.

8. The method for accurately determining faults of an electric propulsion system according to claim 1, characterized in that: The acquiring of historical fault data of each first component in the ship electric propulsion system and constructing a fault tree of the ship electric propulsion system according to the historical fault data includes: The event that the ship electric propulsion system stops operating due to a fault is taken as the top event of the fault tree; extracting each of the first fault events of each of the first components from the historical fault data, and using the first fault events as sub-events of the fault tree; According to the occurrence order of each of the sub-events when the top event in the historical fault data occurs, the potential propagation path is constructed to obtain the fault tree.

9. A system for accurately determining faults of an electric propulsion system, characterized in that: include: A first data acquisition module, a network construction module, an initialization module, a model state transfer module, and a fault determination module; The first data acquisition module is used to acquire historical fault data of each first component in the ship electric propulsion system, and construct a fault tree of the ship electric propulsion system according to the historical fault data; the nodes of the fault tree include a number of first fault events of each first component and potential propagation paths between the first fault events; each of the potential propagation paths includes a number of the first fault events; The network construction module is used to construct a first Petri network model according to the fault tree through causal analysis; wherein the first Petri network model includes a first library representing each of the first fault events and a number of transitions obtained through causal analysis and used to represent the transfer conditions of the first fault events; the first library and the transitions are connected by arcs obtained through the potential propagation path construction; The initialization module is used to initialize each of the first libraries in the first Petri network model according to the first fault event currently occurring when any of the first fault events occurs; The model state transfer module is used to activate the triggerable transition in the first Petri network model according to the current operation data of each of the first components and each of the first libraries, and obtain the first Petri network model corresponding to the next moment; The fault determination module is used to determine the first fault event occurring at the next moment according to each of the first libraries in the first Petri network model corresponding to the next moment.

10. The electric propulsion system fault accurate determination system according to claim 9, characterized in that: According to the fault tree, a first Petri network model is constructed through cause and effect analysis, including: Constructing a first library corresponding to each of the first fault events; By causal analysis, combining the historical operation data of the first components corresponding to the first fault events connected in each of the potential propagation paths, constructing the transitions between the first places corresponding to the first fault events connected; The first place corresponding to the connected first fault event is connected to the transition between the first place corresponding to the connected first fault event to construct an arc.

Images