Mechanical arm vacuum loop fault diagnosis and self-healing system and method

Abstract

The invention discloses a mechanical arm vacuum circuit fault diagnosis and self-healing system and method, relates to the technical field of industrial robot adsorption execution, and solves the problems that vacuum circuit abnormity lacks a traceable diagnosis mechanism, fault recognition is inaccurate and no self-healing strategy support exists. In combination with AND gate / OR gate trigger logic and a minimum cut set recursive expansion technology, the vacuum system has a complete and clear fault conduction path, so that system-level anomaly can be accurately mapped to a bottom layer physical root cause, the structured traceability which cannot be achieved by a traditional method is realized, deviation caused by artificial experience judgment is effectively reduced, and the reliability of the system is improved. And the fault positioning accuracy and diagnosis efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of industrial robot adsorption actuation technology, specifically to a fault diagnosis and self-healing system and method for a robotic arm vacuum circuit. Background Technology

[0002] Robotic arms are widely used in smart manufacturing, precision material handling, and electronic packaging, with many end-effector tasks relying on vacuum adsorption. A vacuum circuit, typically composed of a vacuum pump assembly, valve assembly, vacuum piping, pressure detection module, and control module, is a key subsystem ensuring the reliability of the robotic arm's adsorption and the stability of its movements. However, with increasing work cycles and more complex operating environments, vacuum circuits are prone to various failures over long-term use, including pump performance degradation, valve sticking, minor leaks in the piping, sensor misalignment, and abnormal control response. Abnormalities in the vacuum circuit can directly lead to adsorption failure, material falling, movement deviations, and even production line shutdowns; therefore, the ability to diagnose and repair these issues has a decisive impact on equipment reliability.

[0003] In existing technologies, abnormal monitoring of vacuum circuits generally adopts single-point threshold judgment or simple diagnostic methods based on empirical rules, which have significant shortcomings. Vacuum systems lack a systematic fault tracing mechanism, making it difficult to establish a hierarchical correspondence between system functional failures and underlying physical causes. It is also impossible to construct a logical transmission chain from the final manifestation to the core module and then to the underlying components, resulting in fault location relying on manual experience, low accuracy, and high troubleshooting costs. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a fault diagnosis and self-healing system and method for robotic arm vacuum circuits, which solves the problems of lack of traceable diagnostic mechanisms for vacuum circuit anomalies, inaccurate fault identification, and lack of self-healing strategy support.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for fault diagnosis and self-healing of a robotic arm's vacuum circuit, comprising the following steps: Step 1: Based on all potential causes of overall functional loss, disassemble the vacuum loop fault from top to bottom, establish a node system of top-level events, first-level intermediate events, second-level intermediate events, and bottom events, configure logic gates, and use the minimum cut set method to extract all the simplest bottom event combinations that can trigger the top event; Step 2: When an abnormal event is detected, diagnosis begins from the first-level event, trigger combination vectors are extracted, real-time parameters are collected and compared with historical normal parameters, normal nodes are eliminated layer by layer to obtain the bottom-level suspected node set, and the fault type is classified. Step 3: Construct a fault vector of suspected nodes, parameter deviation features, fault type, and quantitative chemical condition data, match it with the historical fault database, and select the self-healing strategy to be executed based on the similarity.

[0006] As a further aspect of the present invention: the first-level intermediate events include valve system failure, vacuum pump system failure, pipeline system failure, pressure detection module failure, and control module failure affecting abnormalities.

[0007] As a further aspect of the present invention: the logic gate includes an OR gate and an AND gate, wherein the OR gate indicates that the failure of the parent node is triggered when any child node occurs, and the AND gate indicates that the failure of the parent node is triggered only when at least two specified child nodes occur simultaneously.

[0008] As a further aspect of the present invention: the minimum cut set method expands through recursive nodes to extract all the simplest bottom event combinations that can trigger the top-level event, and each simplest bottom event combination constitutes a complete fault propagation path from the bottom event to the top-level event.

[0009] As a further aspect of the present invention: the extraction of the trigger combination vector specifically includes: when a certain level event is used as the diagnostic entry point, obtaining all possible child node trigger combinations; The real-time parameter acquisition specifically includes: acquiring multiple consecutive sets of running parameters for each node in the trigger combination vector and calculating the average value; The comparison with historical normal parameters includes comparing the real-time parameter mean with the parameter range established based on historical normal data; if the value exceeds the range, it is determined to be an abnormal node.

[0010] As a further aspect of the present invention: the stepwise removal of normal nodes specifically includes: for the retained abnormal nodes, if they are not bottom events, then iteratively using them as new parent nodes, repeatedly executing the process of extracting trigger combination vectors, comparing parameters and removing nodes, until the bottom event is reached or all nodes are removed.

[0011] As a further aspect of the present invention: the classification of fault types specifically includes: based on the number of nodes in the underlying suspected node set and their logical relationship in the fault tree, the faults are classified into single faults C1, compound faults C2, or multiple single faults coexisting C3. Among them, single fault C1 corresponds to a single node abnormality and meets the OR gate triggering condition, compound fault C2 corresponds to multiple node abnormalities and meets the AND gate or specific OR gate triggering condition, and multiple single faults coexisting C3 corresponds to multiple node abnormalities but no logical connection.

[0012] As a further aspect of the present invention: the step of selecting the self-healing strategy based on similarity includes: when the matching similarity is greater than or equal to a preset threshold, directly executing the self-healing strategy of the best matching historical case; when the matching similarity is lower than the preset threshold, attempting to execute multiple suspected self-healing strategies in descending order of matching degree; if all attempted strategies are ineffective, triggering a disassembly command for the abnormal component.

[0013] A fault diagnosis and self-healing system for the vacuum circuit of a robotic arm, including: The fault tree construction module is used to construct the fault tree model of the vacuum circuit, store the node system, logic gate relationships and fault paths; The data acquisition module is used to collect the operating parameters and quantitative data of the vacuum circuit in real time. The hierarchical diagnostic module is used to perform anomaly monitoring, trigger combination vector extraction, parameter comparison analysis, and fault type identification. Self-healing execution module: used to construct fault vectors, match historical cases, and execute self-healing strategies; The data storage module is used to store historical normal operating condition parameter benchmarks and self-healing strategies corresponding to historical fault case libraries.

[0014] This invention provides a system and method for fault diagnosis and self-healing of a robotic arm's vacuum circuit. Compared with the prior art, it has the following advantages: (1) This invention constructs a hierarchical fault tree model and combines AND gate / OR gate triggering logic and minimum cut set recursive expansion technology to enable the vacuum system to have a complete and clear fault propagation path. In this way, system-level anomalies can be accurately mapped to the underlying physical root cause, achieving the structured tracing capability that traditional methods cannot achieve, effectively reducing the deviation caused by human experience judgment, and improving the accuracy of fault location and diagnostic efficiency. (2) This invention constructs a comprehensive fault vector consisting of the final suspected node, parameter deviation features, fault type and real-time operating conditions, and performs similarity matching with a preset fault database. In high matching scenarios, the best historical self-healing strategy can be directly reused, and in low matching scenarios, the suspected strategy can be executed step by step according to similarity sorting. This forms an automatic repair mechanism that can be closed-loop, adaptive and continuously optimized. This mechanism greatly shortens the fault recovery time, reduces manual intervention and ineffective disassembly, and enables the vacuum circuit to have intelligent self-healing capabilities that traditional technologies do not have. Attached Figure Description

[0015] Figure 1 This is a flowchart of the method of the present invention. Figure 2 This is a system flowchart of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1 Please see Figure 1 This application provides a method for fault diagnosis and self-healing of a robotic arm's vacuum circuit, comprising the following steps: Step 1: Construct a vacuum loop fault tree model. Decompose the system from top to bottom with the loss of system function as the top event to form a node system containing first-level intermediate events, second-level intermediate events, and bottom events. Configure the triggering logic of AND gates and OR gates for each parent node. Extract the simplest combination of bottom events that trigger the top event using the minimum cut set method to form the fault path. Step 2: Perform hierarchical diagnosis and localization based on the model. When an anomaly is detected, start from the preliminary judgment of the superior event, iteratively obtain its trigger combination vector, collect multiple sets of real-time parameters for each node in the vector to calculate the mean, and compare it with the historical normal working condition parameter range to remove normal nodes and retain abnormal nodes until the bottom suspected node set FN is obtained. Step 3: Determine the fault type based on the number of nodes in FN and their logical relationships, and calculate the parameter deviation characteristics of each node; execute fault self-healing, integrate the suspected node set, parameter deviation characteristics, fault type and real-time quantitative chemical condition data to construct a comprehensive fault vector, and perform similarity matching with historical cases in the preset fault database. If the matching degree is higher than or equal to the threshold, the best historical self-healing strategy is directly executed; otherwise, multiple suspected strategies are tried in order of matching degree until the anomaly is eliminated. If all are ineffective, component disassembly is triggered.

[0018] Example 2 This illustrates yet another embodiment of the invention, and further describes the contents of Embodiment 1 in detail.

[0019] The robotic arm's vacuum circuit includes: vacuum generation, pressure regulation, vacuum transmission, vacuum control, and adsorption execution; Step 1: S11: First, take the vacuum loop operation as the top event T, and dissect all potential causes that lead to the overall loss of the vacuum system's function. Gradually obtain the top-down node system through the transmission path of final failure - core module failure - sub-module failure - underlying physical root cause. Level 1 intermediate events (M1-M5): correspond to the failure of five core modules: valve system failure, vacuum pump system failure, pipeline system failure, pressure detection module failure, and control module failure. Secondary intermediate events (M11-M1N, M21-M2N…M51-M5N): These are further broken down under each primary module. For example, a vacuum pump system failure can be broken down into abnormal pump speed, degradation of the pump's internal seal, abnormal pump current, etc.; a valve system failure can be broken down into valve core jamming, valve body wear, response time delay, etc. Bottom events (M11-M11N…M5K1-M5KN): Further breakdown to the root cause that can be directly pointed to the physical unit, such as a sensor misalignment, a local leak in a vacuum pipeline, a valve core jamming at a specific temperature, or suction vibration caused by wear of a pump bearing, etc. Through the above decomposition, a complete node set structure from top to bottom is obtained.

[0020] S12: Configure corresponding triggering logic for each parent node, including: OR gate: The failure of a parent node can be triggered by the occurrence of a single child node; for example, the failure of the pipeline module (M21) can be triggered by the occurrence of a single pipeline leak. AND gate: The parent node will fail only if at least two child nodes occur simultaneously; for example, the pump module will fail (M1) only if both low pump speed (M11) and internal seal wear (M12) occur simultaneously.

[0021] After configuring all parent-child relationships and logic gates, the minimal cut set method is used to recursively expand the nodes and extract all the simplest combinations of basic events that can trigger the top event T, such as: {M111} {M121, M131} {M211, M212, M213} … Each set of minimal cut sets forms a traceable fault path (bottom node - ... first-level node - T).

[0022] Specifically, this step, through a top-down node decomposition method and a minimum cut set recursive expansion technique, can solve the problems of ambiguous hierarchical causes and unclear triggering paths in vacuum system failures. It combines the transmission path from final failure to core module failure to sub-module failure to underlying physical root cause with OR / AND gate triggering logic, which can simultaneously construct a complete node system and precise fault triggering rules. The top-level events and underlying physical root causes form a hierarchical complementarity, providing a logical basis for fault tracing. This step combines node hierarchical decomposition with triggering logic configuration to achieve visualization and traceable extraction of vacuum system failure paths.

[0023] Step Two: S21: When the system detects an abnormal event and preliminarily determines that it belongs to a certain level of event (e.g., M1), then M1 is used as the diagnostic entry point, the current parent node is used, and the corresponding trigger combination vector F1 is extracted. For example, the combined vector F1 corresponding to M1 might contain: Single trigger combination (OR gate form): F1={M11} or {M12} or {M13}; Multiple sub-trigger combination (AND gate form): F1={M11 & M12}, {M12 & M13}; At this point, all child nodes in F1 are extracted, and the running parameter status of each child node within a certain event period is confirmed to determine whether the node is truly abnormal.

[0024] S22: To ensure timely diagnosis and avoid misjudgment caused by only transient data, collect 10 consecutive sets of real-time parameters (such as pressure, flow rate, temperature, valve position, and current) associated with all nodes in F1 and calculate their mean μ_s; at the same time, read the mean μ_b and variance σ_b of the historical 30 sets of normal operating conditions corresponding to the node in the system as a baseline reference. For each node, the real-time mean μ_s is compared with the allowable variance σ_b under the normal operating condition mean μ_b. If μ_s ∈ [μ_b-Kσ_b, μ_b+Kσ_b], where K is the rated coefficient, the node is considered to be in normal condition and is removed from F1. If μ_s exceeds the interval, the node is considered to be in abnormal condition and is retained as an abnormal node. After verification, if all nodes in F1 are removed, it can be determined that the current fault event does not belong to type M1, and the diagnostic entry is switched to other first-level events. S23: Determine the abnormal nodes retained from the previous layer. If the node is not a bottom-level event, continue to use it as the parent node to obtain whether the new trigger combination vector F2 is an OR gate combination or an AND gate combination. Continue to execute: collect 10 consecutive sets of parameters for all nodes in F2, compare them with the historical 30 sets of normal parameter intervals, remove the normal ones and retain the abnormal ones. If the bottom-level event is still not reached, continue to obtain the next layer vector F3 until all paths reach the bottom-level event or are all removed. Finally, a set of retained bottom-level nodes FN is obtained. This set FN contains all bottom-level events that still show abnormality after passing the layer-by-layer parameter test. These events constitute the final suspected node set of this fault.

[0025] S24: Count all bottom-level abnormal nodes in the FN set to obtain the number of abnormal nodes N, and make a judgment based on the fault tree logic gates: If N=1 and its parent node is an OR gate combination, then it is determined to be a single fault type C1; If N≥2 and these nodes have parent-child association / cooperative triggering relationship, and simultaneously satisfy the corresponding triggering rules of AND gate or OR gate, then it is determined to be a composite fault C2; If N≥2 and there are no transmission links or common minimum cut sets among these nodes, then it is judged as multiple single faults coexisting C3; Finally, the final fault type identifier C*∈{C1, C2, C3} for each fault node in FN is obtained.

[0026] S25: Obtain the parameter deviation characteristics of each underlying node in the final FN. The parameter deviation characteristics are: the average value of 10 real-time parameters μ_s is higher or lower than the average value of 30 historical normal operating conditions μ_b.

[0027] Specifically, this step, through hierarchical trigger vector extraction and multi-set parameter statistical verification, can solve the problems of misjudgment of transient data and inaccurate fault node location. By combining the calculation of the average value of 10 sets of parameters in real time with the comparison of the range of 30 sets of normal operating condition parameters in history, it can simultaneously achieve the preliminary screening of abnormal nodes and the elimination of misjudged nodes. The progressive vector extraction and parameter verification form a synergy, providing a quantitative basis for fault type determination. This step combines the fault tree hierarchical transmission rules with parameter statistical verification to achieve the locking of the bottom suspected node set, accurate classification of fault types, and quantification of parameter deviation features.

[0028] Step 3: Integrate the parameter deviation characteristics and fault type identifier C* of the nodes in the FN set, and supplement them with real-time collected quantitative work condition data (work load level, ambient temperature and humidity, robotic arm work posture) to construct a fault vector based on the final suspected node, parameter deviation characteristics, fault type identifier C*, and quantitative work condition data. The fault vector is matched with the combined vectors of historical fault cases in the preset fault database. The similarity formula is used to obtain the fit between the fault vector and the combined vectors of historical fault cases. If the matching similarity is greater than or equal to 85%, the combined vector of the historical fault case with the highest matching degree is selected, its historical self-healing strategy is read, and this self-healing strategy is directly used for self-healing. If the matching similarity is less than 85%, the combined vectors of the three historical fault cases with the highest matching degree are obtained, and their historical self-healing strategies are read. Three self-healing strategies are obtained and marked as suspected fault self-healing strategies. At the same time, these three self-healing strategies are executed in order of similarity from high to low until the abnormality disappears and the parameters return to normal. If none of the three self-healing strategies can repair the abnormality, the underlying components of the abnormality are disassembled.

[0029] Specifically, this step addresses the issues of insufficient adaptability of self-healing strategies and low efficiency of fault repair by integrating multi-dimensional data and fault database similarity matching technology. By combining suspected node information, parameter deviation features, fault type identification, and quantitative chemical condition data, it can simultaneously achieve comprehensive construction of fault vectors and accurate adaptation of historical cases. Direct invocation of high-matching strategies and orderly verification of suspected strategies complement each other, providing precise solution support for fault repair. This step combines similarity threshold determination (≥85%) with hierarchical self-healing execution logic to achieve closed-loop processing of automated self-healing of vacuum system faults and targeted disassembly of abnormal components.

[0030] Furthermore, refer to Figure 2As shown, a fault diagnosis and self-healing system for the vacuum circuit of a robotic arm is proposed to achieve any of the above-mentioned fault diagnosis and self-healing methods, including: The fault tree construction module is used to build and maintain the vacuum loop fault tree model. It stores a complete node system including top-level events, intermediate events and bottom events, configures the logical gate relationships between each node, and generates all possible fault paths by analyzing the minimum cut set method. The data acquisition module is used to collect the operating parameters and quantitative condition data of the vacuum circuit in real time through multi-source sensors, and to perform real-time preprocessing and feature extraction on the collected data. The operating parameters include pressure, flow rate, temperature, valve position, and current parameters, and the quantitative condition data includes the working load level, ambient temperature and humidity, and robotic arm working posture information. The hierarchical diagnostic module is used to perform system anomaly monitoring, extract trigger combination vectors at each level, identify abnormal nodes through parameter statistical analysis and comparison, and complete the automatic classification and location of fault types based on the fault tree logical relationship. The self-healing execution module is used to construct a multi-dimensional fault feature vector, intelligently match it with the historical fault case library, adaptively select and execute the corresponding self-healing strategy based on the matching results, and monitor the self-healing effect in real time. The data storage module is used to systematically store historical normal operation parameter benchmarks, fault case feature vectors and corresponding self-healing strategy schemes, and to provide data support and services to each module.

[0031] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0032] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for fault diagnosis and self-healing of a robotic arm's vacuum circuit, characterized in that, Includes the following steps: Step 1: Based on all potential causes of overall functional loss, disassemble the vacuum loop fault from top to bottom, establish a node system of top-level events, first-level intermediate events, second-level intermediate events, and bottom events, configure logic gates, and use the minimum cut set method to extract all the simplest bottom event combinations that can trigger the top event; Step 2: When an abnormal event is detected, diagnosis begins from the first-level event, trigger combination vectors are extracted, real-time parameters are collected and compared with historical normal parameters, normal nodes are eliminated layer by layer to obtain the bottom-level suspected node set, and the fault type is classified. Step 3: Construct a fault vector of suspected nodes, parameter deviation features, fault type, and quantitative chemical condition data, match it with the historical fault database, and select the self-healing strategy to be executed based on the similarity.

2. The method for fault diagnosis and self-healing of a robotic arm vacuum circuit according to claim 1, characterized in that, The first-level intermediate events include valve system failure, vacuum pump system failure, pipeline system failure, pressure detection module failure, and control module failure affecting anomalies.

3. The method for fault diagnosis and self-healing of a robotic arm vacuum circuit according to claim 1, characterized in that, The logic gates include OR gates and AND gates, wherein the OR gate indicates that the failure of the parent node is triggered when any child node occurs, and the AND gate indicates that the failure of the parent node is triggered only when at least two specified child nodes occur simultaneously.

4. The method for fault diagnosis and self-healing of a robotic arm vacuum circuit according to claim 1, characterized in that, The minimum cut set method expands through recursive nodes to extract all the simplest combinations of bottom events that can trigger the top-level event. Each combination of bottom events constitutes a complete fault propagation path from the bottom event to the top-level event.

5. The method for fault diagnosis and self-healing of a robotic arm vacuum circuit according to claim 1, characterized in that, The extraction of the trigger combination vector includes: when a certain level event is used as the diagnostic entry point, obtaining all possible child node trigger combinations; The real-time parameter collection includes: collecting multiple sets of continuous running parameters for each node in the trigger combination vector and calculating the average value; The comparison with historical normal parameters includes comparing the real-time parameter mean with the parameter range established based on historical normal data; if the value exceeds the range, it is determined to be an abnormal node.

6. The method for fault diagnosis and self-healing of a robotic arm vacuum circuit according to claim 1, characterized in that, The stepwise removal of normal nodes specifically includes: for the retained abnormal nodes, if they are not bottom events, then iteratively using them as new parent nodes, repeatedly executing the process of extracting trigger combination vectors, comparing parameters, and removing nodes, until the bottom event is reached or all nodes are removed.

7. The method for fault diagnosis and self-healing of a robotic arm vacuum circuit according to claim 1, characterized in that, The specific classification of fault types includes: based on the number of nodes in the underlying suspected node set and their logical relationship in the fault tree, faults are classified into single faults (C1), compound faults (C2), or multiple single faults coexisting (C3). Among them, single fault C1 corresponds to a single node abnormality and meets the OR gate triggering condition, compound fault C2 corresponds to multiple node abnormalities and meets the AND gate or specific OR gate triggering condition, and multiple single faults coexisting C3 corresponds to multiple node abnormalities but no logical connection.

8. The method for fault diagnosis and self-healing of a robotic arm vacuum circuit according to claim 1, characterized in that, The specific steps of selecting and executing a self-healing strategy based on similarity include: when the matching similarity is greater than or equal to a preset threshold, directly executing the self-healing strategy of the best matching historical case; when the matching similarity is lower than the preset threshold, attempting to execute multiple suspected self-healing strategies in descending order of matching degree; and if all attempted strategies are ineffective, triggering a disassembly command for the abnormal component.

9. A fault diagnosis and self-healing system for a robotic arm vacuum circuit, used to implement the method according to any one of claims 1 to 7, characterized in that, include: The fault tree construction module is used to construct the fault tree model of the vacuum circuit, store the node system, logic gate relationships and fault paths; The data acquisition module is used to collect the operating parameters and quantitative data of the vacuum circuit in real time. The hierarchical diagnostic module is used to perform anomaly monitoring, trigger combination vector extraction, parameter comparison analysis, and fault type identification. Self-healing execution module: used to construct fault vectors, match historical cases, and execute self-healing strategies; The data storage module is used to store historical normal operating condition parameter benchmarks and self-healing strategies corresponding to historical fault case libraries.

Images