Production line exception processing method, device, equipment and medium

Abstract

The present application relates to the technical field of industrial automation, and particularly relates to a production line abnormality processing method and device, computer equipment and medium. In response to a production abnormality event of a production line, multi-dimensional state perception data corresponding to the production abnormality event is obtained; an abnormality processing decision tree corresponding to the production abnormality event is constructed according to the multi-dimensional state perception data; an optimal processing path for solving the production abnormality event is determined based on the abnormality processing decision tree; a target abnormality processing strategy corresponding to the optimal processing path is determined, and the target abnormality processing strategy is executed. The present application can break through the limitations of traditional static mapping mechanisms in dealing with'same symptom different causes' and'multiple abnormality concurrency' scenarios, realize adaptive response and accurate disposal of complex abnormality scenarios, significantly improve the intelligent level and response efficiency of production line abnormality processing, and provide reliable protection for stable operation of the production line.

Description

Technical Field

[0001] This invention relates to the field of industrial automation technology, and in particular to a production line anomaly handling method, production line anomaly handling device, computer equipment, and computer-readable storage medium. Background Technology

[0002] In modern intelligent manufacturing systems, production lines are highly integrated production systems whose operational efficiency and stability directly impact product quality and production costs. For example, a packaging production line is a combination of automated or semi-automated equipment used to complete the entire product packaging process, covering all stages from material input to finished product output. Its core functions include sorting, filling, sealing, labeling, and palletizing, and it is widely used in industries such as food, pharmaceuticals, chemicals, and electronics. However, in actual operation, production lines such as packaging lines often experience production interruptions due to equipment failures, material abnormalities, or environmental interference, leading to decreased efficiency. Summary of the Invention

[0003] This invention provides a production line anomaly handling method, a production line anomaly handling device, a computer device, and a computer-readable storage medium, which can achieve adaptive response and precise handling of complex anomaly scenarios, significantly improve the intelligence level and response efficiency of production line anomaly handling, and provide a reliable guarantee for the stable operation of the production line.

[0004] In a first aspect, the production line anomaly handling method provided by the present invention includes:

[0005] In response to production anomaly events on the production line, acquire multi-dimensional status perception data corresponding to the production anomaly events;

[0006] Based on multidimensional state perception data, construct an anomaly handling decision tree corresponding to production anomaly events;

[0007] Based on the anomaly handling decision tree, the optimal handling path for resolving production anomalies is determined;

[0008] Determine the target anomaly handling strategy corresponding to the optimal processing path, and execute the target anomaly handling strategy.

[0009] Secondly, the production line anomaly handling device provided by the present invention includes:

[0010] The multi-dimensional sensing module is used to respond to production anomaly events on the production line and acquire multi-dimensional status sensing data corresponding to the production anomaly events.

[0011] The decision tree construction module is used to construct an exception handling decision tree corresponding to production anomaly events based on multi-dimensional state perception data.

[0012] The path determination module is used to determine the optimal processing path for resolving production anomalies based on the anomaly handling decision tree.

[0013] The operation execution module is used to determine the target exception handling strategy corresponding to the optimal processing path and execute the target exception handling strategy.

[0014] Optionally, in one embodiment, the decision tree construction module is used to take the production anomaly event as the root node of the anomaly handling decision tree; input multi-dimensional state-aware data into a pre-trained splitting rule prediction model to obtain the node splitting rules of the anomaly handling decision tree; recursively split the root node according to the node splitting rules to obtain multi-level branch nodes of the anomaly handling decision tree; for each final-level branch node, the anomaly handling strategy that meets the conditions of the final-level branch node is taken as the leaf node corresponding to the final-level branch node.

[0015] Optionally, in one embodiment, the decision tree construction module is used to generate a set of feasible anomaly handling strategy candidates based on multidimensional state-aware data; for each final-level branch node, anomaly handling strategies that meet the conditions of the final-level branch node are selected from the candidate set and used as the leaf nodes corresponding to the final-level branch node.

[0016] Optionally, in one embodiment, the path determination module is used to: for each processing path in the anomaly handling decision tree, score the processing path on multiple different utility dimensions, and weight and fuse the scoring results of each utility dimension according to the weight coefficient of each utility dimension to obtain the comprehensive utility score of the processing path; and determine the optimal processing path for resolving production anomalies based on the comprehensive utility scores of each processing path.

[0017] Optionally, in one embodiment, the path determination module is used to obtain the historical optimal processing path corresponding to historical multidimensional state perception data similar to the multidimensional state perception data; based on the comprehensive utility score of each processing path and the historical comprehensive utility score of the historical optimal processing path, the optimal processing path for resolving production anomalies is determined from the historical optimal processing path and each processing path of the anomaly handling decision tree.

[0018] Optionally, in one embodiment, the operation execution module is further configured to acquire execution effect data of the production line and acquire the deviation between the execution effect data and the expected effect data corresponding to the target anomaly handling strategy;

[0019] When the deviation is greater than or equal to a preset threshold, the multi-dimensional perception module is also used to reacquire the current multi-dimensional state perception data of the production anomaly event, the decision tree construction module is also used to reconstruct the anomaly handling decision tree corresponding to the production anomaly event based on the current multi-dimensional state perception data, the path determination module is also used to re-determine the optimal handling path for resolving the production anomaly event based on the reconstructed anomaly handling decision tree, and the operation execution module is also used to determine the target anomaly handling strategy corresponding to the re-determined optimal handling path and execute the target anomaly handling strategy until the deviation is less than the preset threshold.

[0020] Optionally, in one embodiment, the production line anomaly handling device provided by the present invention further includes an execution feedback module, which is used to update the weight coefficients of each utility dimension according to the execution effect data when the deviation is less than a preset threshold.

[0021] Thirdly, the computer device provided by the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the production line abnormality handling method provided by the present invention.

[0022] Fourthly, the computer-readable storage medium provided by the present invention stores a computer program, which, when executed by a processor, implements the production line anomaly handling method provided by the present invention.

[0023] This invention provides a production line anomaly handling solution. In response to production anomaly events on the production line, it acquires multi-dimensional state perception data corresponding to the anomaly event; based on the multi-dimensional state perception data, it constructs an anomaly handling decision tree corresponding to the production anomaly event; based on the anomaly handling decision tree, it determines the optimal handling path to resolve the production anomaly event; it determines the target anomaly handling strategy corresponding to the optimal handling path and executes the target anomaly handling strategy. Thus, by dynamically constructing the anomaly handling decision tree when a production anomaly event occurs on the production line and deriving the optimal handling path in real time, it overcomes the limitations of traditional static mapping mechanisms in dealing with scenarios of "same symptoms but different causes" and "multiple anomalies occurring concurrently." It achieves adaptive response and precise handling of complex anomaly scenarios, significantly improving the intelligence level and response efficiency of production line anomaly handling, and providing a reliable guarantee for the stable operation of the production line. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1This is a flowchart illustrating the production line anomaly handling method provided in an embodiment of the present invention;

[0026] Figure 2 yes Figure 1 Detailed process diagram of S120;

[0027] Figure 3 This is an example diagram of the exception handling decision tree constructed in an embodiment of the present invention;

[0028] Figure 4 yes Figure 1 Detailed process diagram of S130;

[0029] Figure 5 yes Figure 4 Detailed process diagram of S1320;

[0030] Figure 6 This is a schematic diagram of the production line abnormality handling device provided in an embodiment of the present invention;

[0031] Figure 7 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0032] To make the technical problems solved, the technical solutions, and the beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0033] It should be understood that, when used in this specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0034] It should also be understood that the term “and / or” as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0035] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."

[0036] Furthermore, in the description of this invention and the appended claims, the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0037] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of the invention include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0038] In related technologies, the following handling methods are typically adopted when facing production anomalies that may occur on the production line:

[0039] 1. Rigid Interlock Protection Mode: Based on the "IF-THEN" hard-coded logic of the programmable logic controller. For example, detecting a "film failure" signal immediately triggers an emergency stop for the entire line. This mode responds quickly, but the handling strategy is absolute and lacks flexibility, which can easily cause cascading shutdowns of non-faulty units, leading to unnecessary production interruptions and reducing overall equipment efficiency.

[0040] 2. Tiered Alarm Manual Intervention Mode: The system issues alarm messages of different levels based on preset thresholds. Subsequent processing relies entirely on the operator's experience for judgment and action. This mode depends on manual response, and its processing efficiency is greatly affected by the operator's skill level. The processing quality is unstable and difficult to replicate, and there is a risk of response delay.

[0041] 3. Static expert system-based processing mode: This mode inputs the experience rules of domain experts into a knowledge base, and the system provides processing suggestions through pattern matching. While this mode achieves partial solidification of experience, the knowledge base is difficult to maintain, has poor scalability, and struggles to adapt to dynamic needs such as production line process changes or new fault types.

[0042] As can be seen from the above, the essence of related technologies all belong to the "static mapping" paradigm, that is, a fixed correspondence of "abnormality type → handling action" is predefined. This paradigm cannot cope with the complex scenarios of "same symptoms but different causes" (such as "poor sealing" in a packaging production line may originate from temperature, pressure, material or equipment wear) and "multiple abnormalities occurring concurrently" in industrial sites. To this end, the present invention provides a production line abnormality handling method, a production line abnormality handling device, a computer device, and a computer-readable storage medium. The production line abnormality handling method can be executed by the production line abnormality handling device or by a computer device integrating the production line abnormality handling device. In response to a production abnormality event on the production line, multi-dimensional state perception data corresponding to the production abnormality event is acquired; based on the multi-dimensional state perception data, an abnormality handling decision tree corresponding to the production abnormality event is constructed; based on the abnormality handling decision tree, the optimal handling path for resolving the production abnormality event is determined; the target abnormality handling strategy corresponding to the optimal handling path is determined, and the target abnormality handling strategy is executed.

[0043] Please refer to Figure 1 This is a flowchart illustrating a production line anomaly handling method disclosed in an embodiment of the present invention, as shown below. Figure 1 As shown, the process for handling production line anomalies can be as follows:

[0044] In S110, in response to production anomaly events on the production line, multi-dimensional state perception data corresponding to the production anomaly events is acquired.

[0045] A production line is an industrial facility that integrates various automated equipment, sensors, and control systems to efficiently manufacture products. Abnormal events that occur during its operation are often highly complex and uncertain; the same abnormal phenomenon may be caused by the superposition of failures in different stages. For example, taking "incomplete sealing" in a packaging production line as an example, its causes may involve multiple dimensions such as fluctuations in heat sealing temperature, insufficient air pressure, material thickness deviations, or wear and tear on the actuators.

[0046] The following uses a computer device as an example to illustrate the specific implementation process of the production line anomaly handling method provided by this invention.

[0047] When a computer device detects a production anomaly on the production line, it randomly triggers data acquisition to obtain multi-dimensional state perception data corresponding to the anomaly. This multi-dimensional state perception data is used to characterize the complete state features of the anomaly in a multi-dimensional space, including but not limited to different dimensions such as anomaly diagnosis input, production context information, equipment status information, and resource constraint information.

[0048] For example, when a production anomaly occurs on the production line, the computer equipment can obtain the preliminary diagnostic results of the upstream intelligent diagnostic system as input for anomaly diagnosis. These diagnostic results include, but are not limited to: the anomaly type used for standardized classification of the production anomaly; the root cause probability distribution reflecting the distribution of the root causes of the anomaly; the diagnostic confidence level indicating the intelligent diagnostic system's confidence in its own diagnostic results; and the expected failure evolution time reflecting the time window for the anomaly to deteriorate to an uncontrollable state, etc.

[0049] Computer equipment can obtain current production context information from the production management system of the production line, including but not limited to: the model and specifications of the products currently being produced, the urgency of the production task, the value of remaining materials, the production plan progress, etc.

[0050] Computer equipment can collect real-time equipment status information through sensor networks deployed at various stages of the production line, including but not limited to the operating mode of each equipment unit, real-time load, health index of key components, availability status of backup equipment units, etc.

[0051] Computer equipment can obtain resource constraint information from the production management system of the production line, including but not limited to: real-time energy costs, safety regulations, and the number, location, and skills of available maintenance personnel, etc.

[0052] For example, taking the "instantaneous deviation in filling accuracy" anomaly event in a beverage bottling production line as an example, after detecting the production anomaly event, the computer equipment obtains the following multi-dimensional state perception data corresponding to the production anomaly event:

[0053] Anomaly diagnosis input: "Instantaneous fluctuation of metering pump", root cause probability distribution: electrical interference: 60%, gas content in material: 30%, other: 10%; confidence level C=75%.

[0054] Production context information: High-value new product is being produced, task urgency P_order="high", current batch is 80% complete.

[0055] Equipment status information: There is a standby metering pump available, and the switchover will take 2 minutes; the current pump health index H = "good".

[0056] Resource constraint information: Engineers with the relevant skills are on duty.

[0057] In S120, an anomaly handling decision tree corresponding to production anomaly events is constructed based on multidimensional state perception data.

[0058] After acquiring multi-dimensional state perception data corresponding to a production anomaly, the computer equipment constructs a unique anomaly handling decision tree in real time based on the acquired multi-dimensional state perception data. This tree is used to dynamically evaluate multiple possible processing paths. The anomaly handling decision tree has the anomaly type as its root node, with downward branches covering different dimensions such as anomaly diagnostic input, production context information, equipment status information, and resource constraint information. Each leaf node corresponds to a feasible anomaly handling strategy, consisting of an ordered sequence of processing actions. For example, the anomaly handling strategy corresponding to a leaf node could be: "Fine-tuning parameter P gain by 5% → Observing for 3 production cycles → If the indicator does not recover, switching to the backup pump → Recording the diagnostic request."

[0059] In S130, the optimal processing path for resolving production anomalies is determined based on the anomaly handling decision tree.

[0060] As shown above, after constructing the exception handling decision tree corresponding to the production anomaly, the computer equipment traverses all the exception handling strategies corresponding to the leaf nodes and selects the processing path with the highest utility as the optimal processing path to resolve the production anomaly. The method of utility evaluation is not specifically limited here and can be designed according to actual production needs.

[0061] In S140, the target anomaly handling strategy corresponding to the optimal processing path is determined and the target anomaly handling strategy is executed.

[0062] In this embodiment of the invention, after determining the optimal processing path, the computer device further determines the exception handling strategy corresponding to the optimal processing path, denoted as the target exception handling strategy, and decomposes it into an executable atomic operation instruction sequence, which is then sent to the relevant execution units for execution in sequence.

[0063] For example, in response to the "instantaneous deviation in filling accuracy" event, the determined target anomaly handling strategy is "fine-tune parameter P gain by 5% → observe for 3 production cycles → if the indicator does not recover, switch to the standby pump → record the diagnostic request". At this time, the computer equipment decomposes the strategy into atomic operation instructions such as "adjust the PID controller P parameter to 1.05 times the original value", "start the cycle monitoring timer", "call the pump switching module instruction", and "generate log entries and upload them to the MES system", which are then sent to the PLC, SCADA and DCS systems for execution in sequence to realize the automated closed-loop control of anomaly handling.

[0064] Alternatively, in one embodiment, please refer to Figure 2 Based on multidimensional state-aware data, an anomaly handling decision tree corresponding to production anomaly events is constructed, including:

[0065] In S1210, the production anomaly event is used as the root node of the anomaly handling decision tree;

[0066] In S1220, multidimensional state-aware data is input into a pre-trained splitting rule prediction model to obtain the node splitting rules of the anomaly handling decision tree.

[0067] In S1230, the root node is recursively split according to the node splitting rules to obtain multi-level branch nodes of the anomaly handling decision tree.

[0068] In S1240, for each final-level branch node, the exception handling strategy that meets the conditions of the final-level branch node is used as the leaf node corresponding to the final-level branch node.

[0069] It should be noted that, in this embodiment of the invention, a splitting rule prediction model is pre-trained, and the training method of this splitting rule prediction model can be as follows:

[0070] Multidimensional state-aware data and labeled processing paths corresponding to historical production anomalies are collected as training samples. A supervised learning method is used to train the splitting rule prediction model until a preset stopping condition is met. This preset stopping condition can be configured according to actual needs, including but not limited to the convergence of the model loss function, the matching degree between the node splitting rules output by the splitting rule prediction model and the labeled processing paths reaching a matching degree threshold, and the number of iterations reaching an upper limit threshold, etc. Furthermore, it should be noted that the specific architecture of the splitting rule prediction model in this embodiment is not limited; machine learning models such as decision trees, random forests, gradient boosting trees, or deep neural networks can be used, as long as they can learn the mapping relationship between state features and processing paths based on historical data.

[0071] Correspondingly, when dynamically constructing an anomaly handling decision tree based on multidimensional state-aware data, the computer equipment inputs the multidimensional state-aware data of the current production anomaly event into the pre-trained splitting rule prediction model, and the splitting rule prediction model outputs the node splitting rule corresponding to the current production anomaly event.

[0072] Subsequently, the computer equipment uses the production anomaly event as the root node and recursively splits the root node according to the node splitting rules output by the splitting rule prediction model, generating multi-level branch nodes that conform to the current operating conditions layer by layer. For each final-level branch node, the computer equipment uses the anomaly handling strategy that meets the conditions of that final-level branch node as the corresponding leaf node, thereby dynamically generating an anomaly handling decision tree that matches the production anomaly event. This decision tree can reflect the current state characteristics and optimal handling logic of the production line in real time, ensuring the accuracy and adaptability of the anomaly handling strategy.

[0073] For example, when the beverage bottling production line in the above embodiment experiences an abnormal event of "instantaneous exceeding of filling accuracy tolerance," the computer equipment takes "instantaneous exceeding of filling accuracy tolerance" as the root node, combines it with the currently collected multi-dimensional state perception data, inputs it into the pre-trained splitting rule prediction model, and outputs the node splitting rule for this abnormal event:

[0074] First-level split: Divide into two based on "task urgency". If it is high, proceed to the branch "prioritize production"; otherwise, proceed to the branch "prioritize quality".

[0075] The second level of splitting: Under the "prioritize production" branch, a binary split is performed based on the "root cause certainty". If the root cause certainty is high, the "attempt to recover" branch is entered; otherwise, the "switch to backup branch" is entered.

[0076] Based on the above splitting rules, the computer equipment recursively constructs an anomaly handling decision tree. Due to the high urgency of the task, the "prioritize production" branch is the main branch, while the "prioritize quality" branch, with its estimated lower utility in the production context, is directly pruned. This ultimately generates a final-level branch node containing "root cause certainty." For this final-level branch node, the computer equipment configures the anomaly handling strategy that meets the "attempt recovery" condition as its A-leaf node, and the anomaly handling strategy that meets the "switch to standby" condition as its B-leaf node, thus forming an anomaly handling decision tree for the current operating condition. Figure 3 As shown.

[0077] Among them, the anomaly handling strategy corresponding to leaf node A may include ["fine-tune pump control parameters +5%", "observe for 5 cycles", "escalate alarm if not recovered"];

[0078] The anomaly handling strategies corresponding to leaf node B may include ["parallel start of standby pump preheating", "switch to standby pump (estimated time 2 min)"], "put the original pump into offline diagnostic mode".

[0079] Optionally, in one embodiment, before setting the exception handling strategy that meets the conditions of the final-level branch node as the leaf node corresponding to the final-level branch node for each final-level branch node, the method further includes:

[0080] Based on multidimensional state-aware data, generate a set of feasible anomaly handling strategy candidates;

[0081] For each last-level branch node, the exception handling strategy that meets the conditions of that last-level branch node is used as the corresponding leaf node, including:

[0082] For each final-level branch node, select the exception handling strategy that meets the conditions of the final-level branch node from the candidate set, and use it as the leaf node corresponding to the final-level branch node.

[0083] It should be noted that computer equipment maintenance has an atomic action library, which contains various executable atomic operations that match the production line, such as "adjusting valve opening", "starting and stopping equipment units", and "switching control modes".

[0084] In this embodiment of the invention, after acquiring multi-dimensional state-aware data corresponding to a production anomaly event, the computer device matches executable atomic operations from an atomic action library based on the multi-dimensional state-aware data, combines them to generate candidate anomaly handling strategies, and includes them in an anomaly handling strategy candidate set for subsequent screening. Specifically, the computer device can screen all technically feasible atomic operations from the atomic action library that comply with safety specifications and are executable in the current production context, and combine them to generate candidate anomaly handling strategies.

[0085] Correspondingly, for each final-level branch node, anomaly handling strategies that meet the different branch conditions of the final-level branch node are selected from the candidate set and configured as the leaf nodes of the final-level branch node to achieve precise matching between strategies and operating conditions.

[0086] Alternatively, in one embodiment, please refer to Figure 4 Based on the anomaly handling decision tree, the optimal handling path for resolving production anomalies is determined, including:

[0087] In S1310, for each processing path in the anomaly handling decision tree, the processing path is scored on multiple different utility dimensions, and the scores of each utility dimension are weighted and fused according to the weight coefficients of each utility dimension to obtain the comprehensive utility score of the processing path.

[0088] In S1320, the optimal processing path for resolving production anomalies is determined based on the comprehensive utility score of each processing path.

[0089] Embodiments of the present invention also provide an optional multi-objective utility evaluation mechanism to ensure the comprehensiveness and scientific nature of the processing path selection.

[0090] For each processing path in the anomaly handling decision tree, the computer device scores it on multiple different utility dimensions, including but not limited to time cost, economic cost, risk cost, and quality impact.

[0091] Among them, the time cost dimension assesses the impact of the time required to execute the processing path on the production recovery cycle; the economic cost dimension quantifies expenditures such as material waste, energy consumption, and maintenance costs; the risk cost dimension measures the probability of processing failure and the risk of secondary failures that may be caused; and the quality impact dimension assesses the potential impact of the processing path on the quality indicators of the current batch and subsequent products.

[0092] For each processing path in the exception handling decision tree, after evaluating the utility scores of each dimension of the processing path, the computer device performs a weighted fusion of the scores of each utility dimension based on the weight coefficients of each utility dimension to obtain the comprehensive utility score of the processing logic, which can be expressed as:

[0093] U = Σ (w_i * u_i(X_i));

[0094] Where U represents the overall utility score of the processing path, w_i represents the weight coefficient of the i-th utility dimension, u_i(X_i) represents the scoring function of the processing path under the i-th utility dimension, and X_i represents the evaluation variable under the utility dimension.

[0095] It should be noted that the embodiments of the present invention do not limit the allocation of weight coefficients for different utility dimensions, and can be dynamically adjusted according to the needs of actual production scenarios to adapt to optimization objectives under different working conditions. For example, in emergency recovery scenarios, the weight of the time cost dimension can be increased, while in quality-sensitive production, the proportion of the quality impact dimension can be increased, and so on.

[0096] As shown above, after evaluating the comprehensive utility score of each processing path in the anomaly handling decision tree, the computer device sorts the processing paths according to their comprehensive utility scores and selects the one with the highest score as the optimal processing path.

[0097] Alternatively, in one embodiment, please refer to Figure 5 Based on the comprehensive utility score of each processing path, the optimal processing path for resolving production anomalies is determined, including:

[0098] In S13210, the historical optimal processing path corresponding to historical multidimensional state perception data similar to the multidimensional state perception data is obtained;

[0099] In S13220, based on the comprehensive utility score of each processing path and the historical comprehensive utility score of the historical best processing path, the optimal processing path for resolving production anomalies is determined from the historical best processing path and the processing paths of the anomaly processing decision tree.

[0100] When determining the optimal processing path, the computer equipment not only relies on the overall utility assessment of the current state, but also incorporates historically optimal processing paths as candidate solutions.

[0101] The computer equipment also calculates the similarity between the multidimensional state perception data of the current production anomaly and the recorded historical multidimensional state perception data, and selects the historical optimal processing path corresponding to the historical multidimensional state perception data whose similarity to the multidimensional state perception data reaches the similarity threshold.

[0102] For each historically optimal processing path selected, the computer device performs a confidence-weighted adjustment on its historical comprehensive utility score to obtain an adjusted historical comprehensive utility score. Specifically, the computer device determines the confidence-weighting coefficient based on the time decay factor and similarity coefficient of the historical multidimensional state-aware data. Historically optimal processing paths that are more recent and have higher similarity will receive a higher confidence-weighting coefficient, and their weighted historical comprehensive utility scores will have greater reference value in the current decision-making process.

[0103] Subsequently, the computer equipment sorts the adjusted historical comprehensive utility scores of each historical best processing path with the comprehensive utility scores of each processing path in the current anomaly handling decision tree, and selects the one with the highest score as the optimal processing path for the current production anomaly.

[0104] For example, when a production anomaly occurs for the first time on a production line, since there is no historical optimal handling path, the computer equipment only ranks and selects the best path based on the comprehensive utility score of each handling path in the current anomaly handling decision tree. It also records the multi-dimensional state-aware data of this anomaly handling and the selected optimal handling path, providing historical reference for subsequent similar events. When a similar anomaly occurs again, the computer equipment will automatically match historical data, filter out the historical optimal handling path recorded when it first occurred, and combine it with the comprehensive utility evaluation results under the current state to jointly rank and determine the optimal handling path.

[0105] The above method for determining the optimal processing path takes into account both historical experience and real-time evaluation results, improving the stability and adaptability of decision-making. In particular, it can effectively enhance the accuracy and response efficiency of anomaly handling in complex and ever-changing production environments.

[0106] Optionally, in one embodiment, after executing the target exception handling strategy corresponding to the optimal processing path, the method further includes:

[0107] Obtain execution performance data from the production line, and determine the deviation between the execution performance data and the expected performance data corresponding to the target anomaly handling strategy;

[0108] If the deviation is greater than or equal to the preset threshold, the current multidimensional state perception data of the production anomaly event is reacquired, and the target anomaly handling strategy is re-determined based on the current multidimensional state perception data. This process is iteratively executed until the deviation is less than the preset threshold.

[0109] This invention also provides an optional dynamic feedback mechanism, which continuously monitors the deviation between the execution effect and the expected goal to achieve dynamic optimization of the processing strategy.

[0110] The computer equipment also acquires execution performance data from the production line, which describes the actual changes in production status after the implementation of the target anomaly handling strategy. Subsequently, the computer equipment compares the execution performance data with the expected performance data corresponding to the target anomaly handling strategy, calculating the deviation between the two. If the deviation is greater than or equal to a preset threshold, the computer equipment determines that the target anomaly handling strategy has failed to effectively control the anomaly. At this point, the computer equipment reacquires multi-dimensional state perception data of the production anomaly event, recording it as the current multi-dimensional state perception data, and reconstructs the anomaly handling decision tree based on this current multi-dimensional state perception data. It then re-determines the optimal handling path and executes the corresponding handling strategy, repeating this process until the deviation is less than the preset threshold.

[0111] Optionally, in one embodiment, after executing the target exception handling strategy corresponding to the optimal processing path, the method further includes:

[0112] If the deviation is less than the preset threshold, the weight coefficients of each utility dimension will be updated based on the execution effect data.

[0113] In this embodiment of the invention, when the deviation is less than a preset threshold, it indicates that the current target anomaly handling strategy is effective. The computer device will dynamically adjust the weight coefficients of each utility dimension based on the actual execution effect data, strengthen the dimensions that contribute more to the decision, and weaken the secondary factors, thereby improving the accuracy of subsequent decisions.

[0114] For example, the execution performance data acquired by the computer device includes the actual utility score of the target anomaly handling strategy. The computer device uses the actual utility score as the dependent variable and the scoring results of each dimension as independent variables, and uses regression analysis to fit the contribution of each dimension's score to the actual utility, which can be expressed as:

[0115] Uactual=β0+β1*Xtime+β2*Xcost+β3*Xrisk+β4*Xquality;

[0116] Where Uactual represents the actual utility score, Xtime, Xcost, Xrisk, and Xquality represent the scoring structure of the four utility dimensions: time cost, economic cost, risk cost, and quality impact, respectively. β0 is a constant term, and β1 to β4 are the regression coefficients of the corresponding dimensions, reflecting their contribution to actual utility.

[0117] The computer equipment performs regression analysis based on the above regression model to obtain the regression coefficients β1 to β4 for each utility dimension, and uses them as updated weight coefficients for subsequent anomaly handling decisions.

[0118] For example, the initial weight coefficients are set to be equally distributed, i.e., β1=β2=β3=β4=0.25. As data on the effects of multiple anomaly handling executions accumulates, regression coefficients β1=0.35, β2=0.18, β3=0.22, and β4=0.25 are obtained. The computer equipment updates the weight coefficients of each utility dimension accordingly, making the subsequent decision-making model closer to actual production needs. Through continuous iterative optimization, the computer equipment adaptively adjusts the strategy evaluation criteria in a dynamic environment, significantly improving the efficiency and intelligence level of anomaly handling.

[0119] As can be seen from the above, the production line anomaly handling solution provided by this invention responds to production anomaly events on the production line by acquiring multi-dimensional state perception data corresponding to the production anomaly events; based on the multi-dimensional state perception data, it constructs an anomaly handling decision tree corresponding to the production anomaly events; based on the anomaly handling decision tree, it determines the optimal handling path for resolving the production anomaly events; it determines the target anomaly handling strategy corresponding to the optimal handling path and executes the target anomaly handling strategy. Thus, by dynamically constructing an anomaly handling decision tree when a production anomaly event occurs on the production line and deriving the optimal handling path in real time, it overcomes the limitations of traditional static mapping mechanisms in dealing with scenarios of "same symptoms but different causes" and "multiple anomalies occurring concurrently," achieving adaptive response and precise handling of complex anomaly scenarios. This significantly improves the intelligence level and response efficiency of production line anomaly handling, providing a reliable guarantee for the stable operation of the production line.

[0120] To facilitate better implementation of the above production line anomaly handling methods, this embodiment of the invention also provides a corresponding production line anomaly handling device. The meanings of the terms used are the same as in the above production line anomaly handling methods; for specific implementation details, please refer to the descriptions in the above method embodiments.

[0121] Please refer to Figure 6 The production line anomaly handling device may include a multi-dimensional perception module 210, a decision tree construction module 220, a path determination module 230, and an operation execution module 240. Detailed descriptions of each functional module are as follows:

[0122] The multi-dimensional sensing module 210 is used to respond to production abnormal events on the production line and acquire multi-dimensional status sensing data corresponding to the production abnormal events.

[0123] The decision tree construction module 220 is used to construct an exception handling decision tree corresponding to production anomaly events based on multidimensional state perception data.

[0124] The path determination module 230 is used to determine the optimal processing path for resolving production anomalies based on the anomaly handling decision tree.

[0125] The operation execution module 240 is used to determine the target exception handling strategy corresponding to the optimal processing path and execute the target exception handling strategy.

[0126] Optionally, in one embodiment, the decision tree construction module 220 is used to take the production anomaly event as the root node of the anomaly handling decision tree; input multi-dimensional state-aware data into a pre-trained splitting rule prediction model to obtain the node splitting rules of the anomaly handling decision tree; recursively split the root node according to the node splitting rules to obtain multi-level branch nodes of the anomaly handling decision tree; for each final-level branch node, the anomaly handling strategy that meets the conditions of the final-level branch node is taken as the leaf node corresponding to the final-level branch node.

[0127] Optionally, in one embodiment, the decision tree construction module 220 is used to generate a set of feasible anomaly handling strategy candidates based on multidimensional state perception data; for each final branch node, anomaly handling strategies that meet the conditions of the final branch node are selected from the candidate set and used as the leaf nodes corresponding to the final branch node.

[0128] Optionally, in one embodiment, the path determination module 230 is used to: for each processing path in the anomaly handling decision tree, score the processing path in multiple different utility dimensions, and weight and fuse the scoring results of each utility dimension according to the weight coefficient of each utility dimension to obtain the comprehensive utility score of the processing path; and determine the optimal processing path for resolving production anomalies based on the comprehensive utility scores of each processing path.

[0129] Optionally, in one embodiment, the path determination module 230 is used to obtain the historical optimal processing path corresponding to historical multidimensional state perception data similar to the multidimensional state perception data; and to determine the optimal processing path for resolving production anomalies from the historical optimal processing path and the processing paths of the anomaly processing decision tree based on the comprehensive utility score of each processing path and the historical comprehensive utility score of the historical optimal processing path.

[0130] Optionally, in one embodiment, the operation execution module 240 is further configured to acquire execution effect data of the production line and acquire the deviation between the execution effect data and the expected effect data corresponding to the target anomaly handling strategy;

[0131] When the deviation is greater than or equal to a preset threshold, the multi-dimensional perception module 210 is further used to reacquire the current multi-dimensional state perception data of the production anomaly event, and the decision tree construction module 220 is further used to reconstruct the anomaly handling decision tree corresponding to the production anomaly event based on the current multi-dimensional state perception data; the path determination module 230 is further used to re-determine the optimal handling path for resolving the production anomaly event based on the reconstructed anomaly handling decision tree; the operation execution module 240 is further used to determine the target anomaly handling strategy corresponding to the re-determined optimal handling path, and execute the target anomaly handling strategy until the deviation is less than the preset threshold.

[0132] Optionally, in one embodiment, the production line anomaly handling device provided by the present invention further includes an execution feedback module, which is used to update the weight coefficients of each utility dimension according to the execution effect data when the deviation is less than a preset threshold.

[0133] Specific limitations regarding the production line anomaly handling device can be found in the limitations of the production line anomaly handling method described above, and will not be repeated here. Each module in the aforementioned production line anomaly handling device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0134] In one embodiment, a computer device is provided, the internal structure of which can be shown as follows: Figure 7 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and the database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The network interface connects to external wireless clients, providing wireless network access services to the connected clients. When executed by the processor, the computer program implements the production line anomaly handling method provided by this invention.

[0135] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the production line abnormality handling method described in the above embodiment.

[0136] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by a processor, implements the production line anomaly handling method described above.

[0137] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0138] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0139] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A production line anomaly handling method characterized by comprising: include: In response to a production anomaly event on the production line, acquire multi-dimensional state perception data corresponding to the production anomaly event; The aforementioned production anomaly event is used as the root node of the anomaly handling decision tree; The multidimensional state-aware data is input into a pre-trained splitting rule prediction model to obtain the node splitting rules of the anomaly handling decision tree. According to the node splitting rules, the root node is recursively split to obtain the multi-level branch nodes of the anomaly handling decision tree; Based on the multidimensional state-aware data, a set of feasible anomaly handling strategy candidates is generated. For each terminal branch node, select the exception handling strategy that meets the conditions of the terminal branch node from the exception handling strategy candidate set, and use it as the leaf node corresponding to the terminal branch node. For each processing path in the anomaly handling decision tree, the processing path is scored on multiple different utility dimensions, and the scores of each utility dimension are weighted and fused according to the weight coefficients of each utility dimension to obtain the comprehensive utility score of the processing path. Obtain the historical optimal processing path corresponding to historical multidimensional state perception data similar to the multidimensional state perception data; Based on the comprehensive utility score of each processing path and the historical comprehensive utility score of the historical best processing path, the optimal processing path for resolving the production anomaly is determined from the historical best processing path and each processing path of the anomaly handling decision tree. Determine the target anomaly handling strategy corresponding to the optimal processing path, and execute the target anomaly handling strategy.

2. The process of claim 1, wherein, After executing the target exception handling strategy, the method further includes: Obtain the execution effect data of the production line, and obtain the deviation between the execution effect data and the expected effect data corresponding to the target anomaly handling strategy; If the deviation is greater than or equal to the preset threshold, the current multidimensional state perception data of the production anomaly event is reacquired, and the target anomaly handling strategy is re-determined based on the current multidimensional state perception data. This process is iteratively executed until the deviation is less than the preset threshold.

3. The process of claim 2, wherein, After obtaining the deviation between the execution effect data and the expected effect data corresponding to the target anomaly handling strategy, the method further includes: If the deviation is less than the preset threshold, the weight coefficients of each utility dimension are updated based on the execution effect data.

4. An abnormality processing device for a production line, characterized by comprising: include: The multi-dimensional sensing module is used to respond to production abnormal events on the production line and acquire multi-dimensional state sensing data corresponding to the production abnormal events. The decision tree construction module is used to take the production anomaly event as the root node of the anomaly handling decision tree; input the multi-dimensional state-aware data into a pre-trained splitting rule prediction model to obtain the node splitting rules of the anomaly handling decision tree; and recursively split the root node according to the node splitting rules to obtain the multi-level branch nodes of the anomaly handling decision tree. Based on the multidimensional state perception data, a feasible set of candidate anomaly handling strategies is generated; for each terminal branch node, anomaly handling strategies that meet the conditions of the terminal branch node are selected from the set of candidate anomaly handling strategies and used as the leaf node corresponding to the terminal branch node. The path determination module is used to score the utility of each processing path in the anomaly handling decision tree across multiple utility dimensions, and to weight and fuse the scores of each utility dimension according to the weight coefficients of each utility dimension to obtain the comprehensive utility score of the processing path; to obtain the historical optimal processing path corresponding to historical multidimensional state perception data similar to the multidimensional state perception data; and to determine the optimal processing path for resolving the production anomaly event from the historical optimal processing path and the processing paths of the anomaly handling decision tree based on the comprehensive utility scores of each processing path and the historical comprehensive utility scores of the historical optimal processing path. The operation execution module is used to determine the target anomaly handling strategy corresponding to the optimal processing path and execute the target anomaly handling strategy.

5. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the production line abnormality handling method according to any one of claims 1 to 3.

6. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the production line abnormality handling method according to any one of claims 1 to 3.

Images