Method for detecting faulty equipment

Abstract

One aspect of the invention relates to a method (100) for detecting faulty equipment from among a plurality of items of equipment of a device using a dynamic global decision tree. The detection method (100) comprises: - at a current node of the global decision tree, displaying (110) the question relating to a failure of the device respectively associated with the current node; - in response to the question being displayed (110), receiving (120) an item of information relating to a characteristic of at least one item of equipment from among the plurality of items of equipment; - determining (130) at least one item of equipment to be checked from among the plurality of items of equipment of the device according to the received item of information; - selecting (140) at least one decision tree from among the one or more decision trees associated respectively with the at least one item of equipment to be checked; and - updating (150) the global decision tree by replacing a lower tree structure of the current node with at least one portion of the at least one selected decision tree.

Description

DESCRIPTION TITLE: Method for detecting faulty equipment TECHNICAL FIELD OF THE INVENTION

[0001] The technical field of the invention is that of the detection of faulty equipment and more specifically the detection of faulty equipment of an aircraft.

[0002] The present invention relates to a method for detecting faulty equipment. TECHNOLOGICAL BACKGROUND OF THE INVENTION

[0003] Troubleshooting and maintenance operations in various industrial sectors are often lengthy and complex. For example, troubleshooting an aircraft engine can last several days and require numerous tasks, particularly in identifying the source of the failure. Detecting the faulty equipment in an aircraft engine failure is generally complex and requires a large number of tests and checks. These troubleshooting operations typically require the intervention of a maintenance technician. To optimize the troubleshooting process, for example, in terms of time and material resources, it is well-established that maintenance technicians need support to identify the faulty equipment among the various components of the system being troubleshooted.

[0004] Troubleshooting assistance, in the form of a process for detecting faulty equipment in a device, can be classified into two groups. The first group of fault detection processes is of the "Manual Troubleshooting" type. These "Manual Troubleshooting" processes are based on predefined decision trees. A predefined decision tree is a decision tree whose architecture and passage conditions are fixed and defined at the time of its design. Because they are based on a predefined decision tree or a sequence of predefined decision trees, these processes allow for the establishment of a well-structured, standardized, and readable troubleshooting procedure. This ensures that each step of the procedure is followed consistently and provides a methodical approach to resolving the fault. These processes can therefore be These are used to effectively train maintenance operators and improve the repeatability of detection diagnostics. Once maintenance operators are familiar with the predefined decision tree (or sequence of predefined decision trees), they can apply the troubleshooting process consistently and rigorously, thus minimizing potential errors and oversights.

[0005] To determine the order of questions within the decision tree(s) used, several factors are considered, such as the probability of failure of the equipment involved, the complexity of the question posed at each step, and the accessibility of the equipment. Questions relating to the most accessible or easily diagnosable equipment are often placed at the top of the decision tree, i.e., the first questions displayed or information provided in the troubleshooting procedure. This allows for more efficient diagnosis by quickly eliminating potential causes when they are easily verifiable. However, these "Troubleshooting Manual" type processes have a major drawback: their specific and fixed nature.In other words, since the decision trees used, as well as their sequence, are predefined and fixed, a discrepancy with the actual situation can exist and degrade the assistance provided by the detection process. For example, the accessibility of specific equipment on an aircraft engine can be influenced by various ancillary equipment and specific configurations, such as the placement of an auxiliary particulate filter near the air intake of an aircraft engine. In this configuration, the troubleshooting procedure, and therefore the detection process, must take this specific placement into account. Within a "Troubleshooting Manual" type process, this means that the order of questions must consider the arrangement of the ancillary equipment and provide specific instructions for handling it.No current method for detecting faulty equipment based on a fixed decision tree can handle all possible configurations. Therefore, detecting a faulty device using current "Troubleshooting Manual" type methods requires adjustments or adaptations to the standard procedure to account for these limitations.

[0006] The second group of methods for detecting faulty equipment represents any other approach, for example a neural network approach, or for example a more widespread approach based on "Bayesian networks". Bayesian network-based methods allow for the representation of cause-and-effect relationships between different variables within a device troubleshooting procedure. For example, Bayesian networks can model the various possible causes of a problem and their associated indicators or symptoms. The conditional probabilities between these different variables are then quantified, facilitating diagnosis. These Bayesian network-type methods are therefore particularly well-suited for diagnosing and resolving complex problems.

[0007] These "Bayesian network" methods, however, have drawbacks, particularly related to the flexibility of the question order. This flexibility can make the troubleshooting procedure less structured and therefore more difficult for the maintenance technician to follow. Furthermore, designing Bayesian networks is very complex and time-consuming, as it is always necessary to simulate each possible failure scenario to determine the resulting procedures. Thus, in practice, these "Bayesian network" methods are reserved for complex troubleshooting cases, that is, those requiring consideration of a multitude of indicators or symptoms that are difficult for the maintenance technician to access. Another drawback of these "Bayesian network" methods is that modifying a conditional probability within the Bayesian network can have unpredictable effects on all related troubleshooting procedures.Because conditional probabilities are used to detect all possible causes, unforeseen adjustments can lead to unexpected changes in diagnoses and resolution recommendations. This necessitates careful validation and a thorough understanding of the implications of each conditional probability modification.

[0008] Therefore, there is a need to provide a troubleshooting assistance process that at least partially resolves the drawbacks of currently known solutions. SUMMARY OF THE INVENTION

[0009] The invention offers a solution to the problems mentioned above, based on a dynamic global decision tree. Indeed, at a current node, The global decision tree is updated based on the response received by the computer. Updating the global decision tree involves replacing a lower-level tree at the current node with decision trees selected based on the information received. Thus, at each current node, the global decision tree is adapted and updated according to the latest information received. For example, for each displayed question, the user can choose to answer the question or select a piece of equipment to check. In the first alternative, the user can choose to provide information in response to the question. Then, a list of equipment to check is determined from this provided information. In the second alternative, the user can directly select a piece of equipment to check from a list of equipment to check, determined, for example, during the previous iteration.Finally, for both alternatives, the next node can be determined.

[0010] One aspect of the invention relates to a computer-implemented method for detecting a faulty piece of equipment among a plurality of equipment in a device using a dynamic global decision tree, wherein each piece of equipment among the plurality of equipment in the device is associated with a respective decision tree from among a plurality of decision trees, each decision tree comprising a set of nodes, each node being respectively associated with a question relating to a failure of the device, the detection method comprising the steps of: at a current node of the global decision tree, displaying the question relating to a failure of the device respectively associated with the current node, in response to the display of the question, receiving information relating to a characteristic of at least one piece of equipment among the plurality of equipment.determining at least one piece of equipment to be checked from among the plurality of equipment in the device based on the information received, selecting at least one decision tree from among the decision tree(s) associated respectively with the at least one piece of equipment to be checked, and, update the global decision tree by replacing a lower tree of the current node with at least a portion of the at least one selected decision tree.

[0011] In the invention, the detection method is reliable and understandable by everyone. Furthermore, the detection method is optimized in terms of time and computational resources. Finally, the detection method is adaptive since the overall decision tree is updated using multiple decision trees. It is indeed possible to delete, modify, or add one or more decision trees based on the response received by the computer.

[0012] In addition to the features mentioned in the preceding paragraph, the method according to one aspect of the invention may have one or more complementary features from the following, considered individually or in all technically possible combinations: the method further includes a step of determining a next node based on the information received, the next node being a node of the updated global decision tree; the steps of displaying the question, receiving information, determining at least one piece of equipment to be checked, selecting at least one decision tree, and updating the overall decision tree of the method are iterated until the faulty equipment is detected, and wherein the current node of a current iteration is the next node of a previous iteration; each piece of equipment among the plurality of equipment in the device is associated with a probability of failure.The process also includes updating the probability of failure of at least one piece of equipment among the plurality of equipment in the device based on the information received; the selection of at least one decision tree or the updating of the overall decision tree is carried out based on an accessibility criterion for at least one piece of equipment and / or the updated probability of failure of at least one piece of equipment. at least a portion of at least one selected decision tree includes a node designated as a mandatory next node of the current node in the overall decision tree; the method further includes identifying, among the sets of nodes in at least one selected decision tree, nodes, called redundant nodes, associated with the displayed question; at least a portion of at least one selected decision tree includes only nodes other than the redundant nodes; each node in the plurality of decision trees is associated with a respective status indication among: Paused, Successfully completed the faulty equipment detection, or When the failure to detect the equipment is complete, the selection of at least one decision tree is performed by excluding decision trees that include nodes associated with the status indication "Completed with failure to detect the equipment is complete", or at least a portion of the selected decision tree is determined by excluding nodes associated with the status indication "Completed with failure to detect the equipment is complete", the updated global decision tree is displayed.

[0013] Another aspect of the invention relates to a computer program product comprising instructions which, when the program is executed by a computer, cause the computer to implement the process according to the invention.

[0014] An additional aspect of the invention relates to a system for detecting faulty equipment comprising means adapted to carry out the detection process according to the invention.

[0015] The invention and its various applications will be better understood by reading the following description and examining the accompanying figures. BRIEF DESCRIPTION OF THE FIGURES

[0016] The figures are presented for illustrative purposes only and are in no way limiting to the invention. Figure 1 is a synoptic diagram illustrating the steps of an example of the detection process according to the invention. Figure 2 is a representation of an example of a graphical interface compatible with the detection method according to the invention. DETAILED DESCRIPTION

[0017] Unless otherwise specified, the same element appearing on different figures has a unique reference.

[0018] Figure 1 is a block diagram illustrating the steps of an example of process 100 according to the invention. The mandatory steps of the example of process 100 are indicated by a solid rectangle and the optional steps are indicated by a dashed rectangle.

[0019] Process 100 is computer-implemented. "Computer-implemented" means that all, or virtually all, of the steps in Process 100 are executed by at least one computer, processor, or similar system. Thus, steps are carried out by the computer, possibly fully or semi-automatically. In some examples, the triggering of at least some of the steps in these processes may be achieved through user-computer interaction, referred to in this application as user interaction. The level of user interaction required may depend on the intended level of automation and be balanced against the need to implement the user's wishes. In some examples, this level may be user-defined and / or predefined. User interaction may be performed using a mouse or other device.When the computer includes a touchscreen, user interaction can be performed using that touchscreen.

[0020] A typical example of a computer implementation of Process 100 is to run Process 100 with a system adapted for this purpose. The system may include a processor coupled with memory and a graphical user interface, the memory having stored a computer program containing instructions to implement the process. The memory may It can also store data, for example in the form of a database. Memory is any hardware suitable for such storage, possibly comprising several distinct physical parts. The system may also include one or more means of user interaction or one or more means of measuring physical values ​​for one or more components of the device being troubleshooted.

[0021] Method 100 is a method for detecting faulty equipment. In this application, the term "faulty" means the temporary or permanent loss of the qualities required to perform a function. The term "faulty" in this application is synonymous with "defective" or "broken down." The faulty equipment to be detected is a component of a device. The device comprises a plurality of components. For example, the device is a component of an aircraft, such as an aircraft engine. The method can be implemented iteratively. In one example, steps 110 to 170 of Method 100 can be repeated until the faulty equipment is detected and, if necessary, repaired. In another example, steps 110 to 170 of Method 100 can be repeated until the device is functioning normally, and thus all faulty components of the device are detected and, if necessary, repaired.At each iteration of steps 110 to 170 of process 100, a question regarding the device malfunction is displayed, and information relating to a characteristic of at least one piece of equipment within the device is provided in response. This response can, for example, be provided by the maintenance operator responsible for troubleshooting the device. Alternatively, this response can be provided by a system, such as a database, a measurement system, or a diagnostic system for the equipment. This response can, for example, be derived from external data such as flight sensor data or documentation of previously performed maintenance. Thus, some iterations of steps 110 to 170 of process 100 can be performed automatically, that is, without interaction with the maintenance operator.

[0022] The 100 method is based on a decision tree, called the global decision tree. The term "global" here means that the global decision tree is updated from a plurality of decision trees. A decision tree is a decision support tool corresponding to a set of choices that can be represented classically in the graphical form of a tree, or in the form of steps. with associated questions and cross-references. The global decision tree comprises a set of nodes connected by transitions. Each node in the global decision tree is connected to at least one other node in the decision tree. Between two iterations of process 100, the global decision tree is traversed from top to bottom, that is, from the root node to one of the leaf nodes. It should be noted that it is possible, between two iterations of process 100, to jump between two nodes that are not a parent and child node. For example, it is possible to move from a current node to a sibling node of the current node; a sibling node is a node that shares the same parent node as the current node. This jump from a current node to a sibling node of the current node can correspond to a change in the equipment to be checked, i.e., modifying the next piece of equipment to be checked from the list of equipment to be checked.It is also possible to move back up the global decision tree, for example, by returning to the parent node of the current node or to a more distant node in the decision tree, such as any node other than one of the child nodes of the current node. The root node is the node in the global decision tree that does not have a parent node. A leaf node is a node that does not have a child node. Every node in the decision tree, other than the leaf nodes, has a sub-tree comprising all of that node's child nodes. In other words, every node in the decision tree, apart from the leaf nodes, can be considered a root node of another decision tree corresponding to a sub-section of the global decision tree. Within process 100, each node is associated with a question to be displayed in order to detect a faulty piece of equipment among the plurality of equipment in the system.An answer to the question is expected in order to update the global decision tree and determine the next node, which will become the current node for the next iteration of process 100. The current node and the next node are therefore nodes of the global decision tree.

[0023] The overall decision tree is dynamic. In fact, the overall decision tree evolves during the implementation of process 100. For example, one or more nodes of the overall decision tree can be deleted, replaced, or added during process 100. Alternatively, transitions between nodes of the overall decision tree can also be deleted, replaced, or added during process 100. The overall decision tree is updated using multiple decision trees. Among all the decision trees used, A plurality of decision trees is associated with the plurality of equipment in the system. For example, one or more decision trees may be specific to a piece of equipment in the system. In another example, each piece of equipment in the system is associated with a decision tree. In a further example, some decision trees, or at least portions of decision trees, may be generic and therefore common to all or part of all the equipment in the system.

[0024] An initial global decision tree can be obtained by the computer before the first global decision tree update performed during process 100. The term "obtain" in this application means "receive" or "generate." For example, an initial global decision tree might consist of a single node that displays a default question. An example of a default question could be an error message provided by the device being troubleshooted or an instruction, such as one designed to ensure user safety or device integrity. Alternatively, the initial global decision tree could consist of multiple nodes. For example, the initial global decision tree could include all the decision trees of the plurality of decision trees. Finally, regardless of the number of nodes, the initial global decision tree could be specific to the device being troubleshooted.

[0025] Step 110 of the process involves displaying a question about a device failure. This step 110 is performed at a current node of the global decision tree. In other words, for a given current node of the decision tree, a question about a device failure is displayed during step 110. Thus, when multiple iterations of process 100 are performed, at each transition to a new current node of the global decision tree, a question about a device failure is displayed during step 110. The displayed question can relate to various aspects useful for the troubleshooting procedure. For example, the displayed question can concern: a diagnosis to be performed, such as replacing equipment with other normally functioning equipment or restarting equipment; the repeatability of the failure; or the occurrence of one or more alert messages. details on the conditions under which the failure first occurred, details on the operational context, for example climatic conditions, availability of a diagnostic device, configuration in which the device is installed, availability or not of accessory equipment, or difficulty in accessing or seeing equipment of the device.

[0026] Displaying the question about a device failure (110) can be done using a computer display means implementing method 100 or a display means connected to the computer implementing method 100.

[0027] Step 120 of the process involves receiving information relating to a characteristic of at least one piece of equipment from among the plurality of equipment. This information is received in response to the question displayed in step 110. The information received may be a free-text answer. Alternatively, the information received may be one of a set of predetermined answers to the displayed question.The information received may relate to: the result of a diagnosis carried out, the repeatability of the failure or of obtaining one or more alert messages, the conditions of occurrence of the failure the first time it was detected, the operational context, for example climatic conditions, the possible availability of a diagnostic device, the configuration in which the device is installed, the availability or not of accessory equipment, or the difficulty of accessing or visibility of a piece of equipment of the device.

[0028] Step 130 of the process involves determining at least one piece of equipment to be checked from among the plurality of equipment in the device. Thus, in step 130, it is possible that at least one piece of equipment among the plurality of equipment in the device may be considered as not being faulty, for example, when a piece of equipment has been inspected and its normal operation has been verified. This determination is made based on the information received. For example, if the information received includes an indication of an alert message or an error code provided by the device or a piece of equipment in the device, the piece(s) of equipment to be checked can be deduced from this. In another example, when the information received is the result of a diagnostic confirming that a piece of equipment is functioning correctly, this piece of equipment can be removed from the list of at least one piece of equipment to be checked.

[0029] Step 140 of the process involves selecting at least one decision tree associated with at least one piece of equipment to be checked. This selection can also be performed as an alternative to step 120. When step 140 is performed as an alternative to step 120, step 130 may also be omitted. For example, the user can explicitly designate the equipment to be checked using a button. For instance, for a piece of equipment to be checked, the decision tree(s) associated with that equipment are selected. Conversely, if a piece of equipment is considered not to require checking, the decision tree(s) associated with that equipment are not selected. Other predetermined or non-predetermined selection criteria may be used. Furthermore, when generic decision trees are used in process 100, these trees may or may not be selected, according to predefined or non-predefined rules.

[0030] An optional step 150 of the process involves identifying, among the sets of nodes in at least one selected decision tree, nodes associated with the displayed question. Nodes associated with the displayed question are called redundant nodes. Redundant nodes can indeed be present in different decision trees across the plurality of decision trees used in process 100. Thus, redundant nodes for which an answer has been obtained are filtered out to be removed from the overall decision tree. In this example, at least a portion of the at least one selected decision tree therefore includes only nodes other than the redundant nodes. The size of the overall decision tree is thus optimized. The duration of the detection process 100 is also reduced since, When the information has been received for a redundant node, the same question is no longer displayed.

[0031] In an example consistent with the previous examples, at least a portion of the selected decision tree includes only nodes other than those belonging to the lower hierarchy of the redundant nodes. This example can be implemented, for instance, when redundant nodes have an identical lower hierarchy. Thus, the number of nodes in the updated global decision tree is minimized.

[0032] Step 160 of the process involves updating the global decision tree. Updating the global decision tree in step 160 consists of replacing the current node's lower tree with at least a portion of at least one selected decision tree. Thus, the global decision tree's lower tree is updated in step 160. To perform the replacement of the current node's lower tree, the selected decision tree(s) can be used partially or in their entirety. In one example, the selected decision trees can be used in their entirety, that is, without modification, to replace the current node's lower tree. In another example, some of the selected decision trees can be modified before being used to replace the lower tree.For example, modifying a selected decision tree might involve isolating one or more portions of the selected decision tree to use only that portion or those portions in replacing the lower tree of the current node. In this latter example, the update 160 of the global decision tree might include, for each selected decision tree, a substep of determining at least one portion of the selected decision tree.

[0033] In an example consistent with the previous examples, determining at least a portion of at least one selected decision tree, during an example in step 160, includes a node designated as a mandatory next node of the current node in the global decision tree. For example, one or more nodes in the plurality of decision trees can be associated with a first metadata indicating a mandatory transition from a first node to a second node. Thus, when the current node in the global decision tree is associated with the first metadata, then this first metadata indicates the The next node is mandatory. This next node is therefore, in this example, a criterion for determining at least one portion of the at least one selected decision tree.

[0034] In an example consistent with the previous one, the mandatory next node instruction can be associated with a piece of equipment in the device. Thus, when the equipment being verified is the device equipment associated with this mandatory next node instruction, then this instruction is considered valid and is complied with. Conversely, when the equipment being verified is not the device equipment associated with this mandatory next node instruction, then this instruction is considered invalid and is not complied with. Alternatively, the mandatory next node instruction can be generic, that is, associated with no device equipment. In this case, this mandatory next node instruction is always complied with.

[0035] An optional step 170 of the process involves determining the next node, which then becomes the next current node—that is, the current node for the next iteration of process 100. The next node can be one of the child nodes of the current node or a node within the current node's lower hierarchical level. The next node can also be a node in the updated global decision tree that is not within the current node's lower hierarchical level, for example, the current node's parent node or a node located above the current node in the global decision tree. This might, for instance, correspond to going back in the device troubleshooting procedure, for example, to recheck or modify an answer provided to a previous question, which was therefore displayed for a previous current node. Finally, the next node can be any node in the updated global decision tree.

[0036] The determination of the next node (170) is based on the information received. For example, the received information may explicitly designate the next node. The next node may also be inferred from the received information. In this case, calculations may be necessary. Furthermore, when a mandatory next node is associated with a current node, the next node is that mandatory next node.

[0037] In an example, consistent with the previous examples, the position of the root node of each selected decision tree is identified in the tree of global decision. Thus, when the verification of a new piece of equipment in the system begins, the root node of the selected decision tree associated with this new equipment becomes the current node of the global decision tree. The start of a verification of new equipment can be inferred from the information received. For example, if the information received indicates that the verification of a piece of equipment in the system is complete and that the verified equipment was not faulty, then it is possible to automatically initiate the start of a verification of a new piece of equipment in the system. The start of a verification of new equipment can also be explicitly indicated in the information received. For example, the maintenance operator can select equipment on a graphical interface as illustrated in Figure 2. The graphical interface 200 in Figure 2 includes three buttons 210 corresponding to three pieces of equipment "E1", "E2", and "E3" in a system.In Figure 2, it can be seen that one of the buttons 210, labeled "E3", is grayed out to indicate that it is disabled. Indeed, one or more buttons 210 may be disabled, making it impossible to select the corresponding equipment as the first item to be checked. This deactivation can be achieved in various ways, such as hiding the button or preventing any user interaction with it. This button (or these buttons) 210 can, for example, allow navigation from a current node to a sibling node in the overall decision tree. This button (or these buttons) 210 can, for example, allow the user to designate the equipment to be checked in step 140. The graphical interface 200 in Figure 2 also includes a means 220 for displaying the question "Q" and means 230 for user interaction, such as buttons, allowing the user to select an answer from a predefined list.In figure 2, three answers are possible: “R1”, “R2” or “R3”.

[0038] In an example consistent with the previous examples, the updated global decision tree 160 is displayed. For instance, the global decision tree can be displayed on a portion 240 of the graphical interface 200. Displaying the dynamic global tree allows the maintenance operator to view the remaining options, corresponding, for example, to the different branches of the lower hierarchy of the global decision tree. Displaying the dynamic global tree also allows the maintenance operator to track their progress in the fault detection process, for example, by viewing what has already been done and what remains to be done. User interactions, such as selecting a next node, are also possible via the display of the dynamic global decision tree. In this case, the information received explicitly indicates the next node. It is also possible that user interactions with certain parts of the global decision tree may be prohibited. For example, if only a portion of a selected decision tree has been retained for the update 160 of the global decision tree, the remainder of the selected decision tree may be displayed, for example, in gray. This grayed-out portion of the selected decision tree is, for example, unclickable; that is, none of the nodes in the remaining part of the selected decision tree can become the next node. Furthermore, the nodes of the updated global decision tree 160 may be displayed differently, for example, with a different color, depending on a characteristic related to the equipment associated with the node.For example, if a first node is associated with a first piece of equipment that has already been verified and is not faulty, a first display color can be used for that first node. In the same example, if a second node is associated with a second piece of equipment that has already been verified but is faulty, a second display color can be used for that second node. Finally, if a third node is associated with a third piece of equipment that has not been verified, a third display color can be used for that third node.

[0039] In an example consistent with the preceding examples, each piece of equipment among the plurality of equipment in the device is associated with a probability of failure. The probability of failure can be determined using previously performed reliability studies. In this example, method 100 also includes updating the probability of failure of at least one piece of equipment among the plurality of equipment in the device based on information received 120. The update of the probability of failure can also be performed based on data received by the computer. For example, when detection method 100 is used to detect a failing piece of equipment in an aircraft engine, flight data can be used to update the probability of failure of at least one piece of equipment in the aircraft engine.The failure probability update can be implemented, for example, before determining 130 which at least one piece of equipment to be checked from among the plurality of equipment in the device. Furthermore, in this same example, the selection 140 of the at least one decision tree is performed based on the updated failure probability of the at least one piece of equipment. For example, The selection of at least one decision tree (140) can be performed when the failure probability of at least one updated piece of equipment exceeds a predetermined first failure probability threshold. Alternatively, in the same example, the update of the overall decision tree (160) is performed based on the failure probability of at least one updated piece of equipment. Specifically, the determination of at least a portion of the selected decision tree can be performed based on the failure probability of at least one updated piece of equipment. For example, when the updated failure probability exceeds a second predetermined failure probability threshold, all nodes of the selected decision tree can be used to replace the lower branch of the current node in the overall decision tree.In this same example, when the updated probability of failure is lower than the second predetermined probability of failure threshold, a single portion of the selected decision tree can be used to replace the lower branch of the current node in the global decision tree. The probability of failure of a piece of equipment can be its relative probability of failure among the total probability of failure of all the equipment in the system. Thus, the sum of the failure probabilities determined for all the equipment in the system can equal 100%. Updating the probability of failure of a piece of equipment can be implemented using a second metadata element associated with a node in the global decision tree.For example, this second metadata may include: a first multiplication factor to apply to a probability of equipment failure when a first response is provided, and a second multiplication factor to apply to the probability of equipment failure when a second response is provided.

[0040] In an example consistent with the previous examples, each node in the plurality of decision trees is associated with a respective status indication. The status can be the status of the current detection process or of a previously implemented detection process. For example, the set of nodes traversed during a process can be associated with a status indication at the end of a current process. Thus, a single node can have a list of status indications corresponding to the set of instances of the processes in which it A node has been traversed. The status indication can be stored using a third metadata element associated with each node in the plurality of decision trees. The status can be one of the following: Paused, Successfully completed the faulty equipment detection, or Ended with failure to detect faulty equipment.

[0041] Thus, it is possible to know, for each node in the plurality of decision trees, whether the traversal of that node was performed during a process that was paused, or completed successfully, or completed unsuccessfully. In this example, the selection of at least one decision tree can therefore be performed based on this status indication. For example, the selection of at least one decision tree can be performed by excluding decision trees containing only nodes associated with the status indication "Completed with failure to detect faulty equipment." Similarly, in this example, at least a portion of the selected decision tree can be determined based on this status indication, for example, by excluding decision trees containing nodes associated with the status indication "Completed with failure to detect faulty equipment."Thus, the updated decision tree 160 is optimized to contain only nodes associated with one or more statuses. In this example, the assistance provided by process 100 uses the history of troubleshooting operations already performed to prompt or require the maintenance operator to follow a specific sequence of steps that led to the detection of faulty equipment in the past. Furthermore, in this example, process 100 maintains traceability of the procedures performed by maintenance operators by storing user questions and answers.

Claims

DEMANDS

1. A computer-implemented method (100) for detecting a faulty piece of equipment among a plurality of equipment in a device using a dynamic global decision tree, wherein each piece of equipment among the plurality of equipment in the device is associated with a respective decision tree from among a plurality of decision trees, each decision tree comprising a set of nodes, each node being respectively associated with a question relating to a device failure, the detection method (100) comprising the steps of: - at a current node of the global decision tree, display (110) the question relating to a device failure respectively associated with the current node, - in response to the display (110) of the question, receipt (120) of information relating to a characteristic of at least one piece of equipment among the plurality of equipment, - determination (130) of at least one piece of equipment to be checked from among the plurality of equipment in the device based on the information received, - selection (140) of at least one decision tree from among the decision tree(s) associated respectively with the at least one piece of equipment to be checked, and - update (160) of the global decision tree by replacing a lower tree of the current node with at least a portion of the at least one selected decision tree.

2. A detection method (100) according to claim 1 further comprising a step of: - Determination (170) of a next node based on the information received, the next node being a node of the global decision tree updated (160).

3. A detection method (100) according to claim 2, wherein the steps of displaying (110) the question, receiving (120) information, determining (130) at least one piece of equipment to be checked, selecting (140) at least one decision tree, and updating (160) the overall decision tree of the method (100) are iterated until the faulty equipment is detected, and in which the current node of a current iteration is the next node of a previous iteration.

4. A detection method (100) according to any one of the preceding claims, wherein each piece of equipment among the plurality of equipment in the device is associated with a probability of failure, wherein the method (100) also includes an update of the probability of failure of at least one piece of equipment among the plurality of equipment in the device based on the information received (120), and wherein the selection (140) of at least one decision tree or the update (160) of the overall decision tree is carried out based on an accessibility criterion of at least one piece of equipment and / or the updated probability of failure of at least one piece of equipment. [Claims] A detection method (100) according to any one of the preceding claims in which at least a portion of at least one selected decision tree comprises a node designated as a mandatory next node of the current node of the global decision tree. [Claims] A detection method (100) according to any one of the preceding claims, further comprising: - identification (150), among the sets of nodes of at least one selected decision tree, of nodes, called redundant nodes, associated with the displayed question, in which at least a portion of at least one selected decision tree comprises only nodes other than the redundant nodes.

7. A detection method (100) according to the preceding claim, wherein each node of the plurality of decision trees is associated with a respective indication of a status among: Paused, Successfully completed the faulty equipment detection, or Ended with failure to detect faulty equipment. in which the selection (140) of at least one decision tree is carried out by excluding decision trees including nodes associated with the status indication "Completed with failure to detect faulty equipment", or in which at least a portion of the at least one selected decision tree is determined by excluding nodes associated with the status indication "Completed with failure to detect faulty equipment". [Claims] A detection method (100) according to any one of the preceding claims, wherein the updated global decision tree (160) is displayed.

9. A computer program comprising instructions which, when the program is executed by a computer, cause the computer to implement the method according to any one of the preceding claims.

10. A system for detecting faulty equipment comprising means adapted to carry out the detection method (100) according to any one of claims 1 to 8.

Images