An after-sales work order intelligent allocation and processing system based on a knowledge graph

Abstract

The application provides a kind of based on knowledge graph's after-sales work order intelligent distribution and processing system, it is related to work order distribution technical field, the system includes double graph construction, work order fact positioning, cognitive conflict identification, cognitive verification path generation, processing strategy generation, work order distribution and graph update module;According to historical after-sales work order, construct fault fact graph and maintenance cognitive graph, establish the mapping relationship of fault fact node and maintenance cognitive node;According to the work order to be processed, locate target fault fact node, identify the conflict of fault cause, processing measure or processing result between associated maintenance cognitive node, generate cognitive conflict event and verification path;Execute verification to determine target maintenance cognitive node, generate work order processing strategy including maintenance mode, flow mode and processing priority, determine processing personnel, processing node and flow path, and update graph according to actual processing process and result;The application can improve the accuracy and efficiency of work order distribution and processing.

Description

Technical Field

[0001] This invention relates to the field of work order allocation technology, and in particular to an intelligent allocation and processing system for after-sales work orders based on knowledge graphs. Background Technology

[0002] With the widespread adoption of intelligent manufacturing and smart terminal equipment, enterprises typically establish intelligent after-sales work order management systems. These systems provide unified management of customer-submitted fault descriptions, equipment operation logs, equipment version information, historical maintenance records, and maintenance personnel's handling results. They also extract information such as fault phenomena, causes, handling measures, and maintenance personnel to construct a knowledge graph. When a new after-sales work order enters the system, the knowledge graph retrieves similar historical maintenance cases, generates fault analysis results and handling strategies, and assigns maintenance personnel, selects maintenance methods, and processes the work order based on these strategies, achieving intelligent allocation and processing of after-sales work orders.

[0003] However, the aforementioned technologies typically use historical maintenance conclusions directly as knowledge relationships in the knowledge graph for subsequent work order processing, without distinguishing between the actual operating facts of the equipment and the maintenance personnel's understanding of the cause of the fault. This results in both types of information being stored together in the same knowledge graph. When historical maintenance understanding changes, the system struggles to identify which knowledge relationships originate from objective equipment facts and which only stem from the technical judgment of the maintenance personnel at the time. Consequently, outdated maintenance understanding is continuously applied to subsequent work order processing. For example, in a smart air conditioner after-sales scenario, a compressor cycle protection phenomenon occurs under software version V1.2. The field engineer replaces the compressor based on the fault symptoms at the time and simultaneously upgrades the control program. The equipment returns to normal, and the work order is ultimately closed as "compressor fault." Based on this, the knowledge graph establishes a knowledge relationship of "no cooling - compressor fault." Subsequent research and development revealed that the fault was actually caused by an abnormal configuration of protection parameters in software version V1.2. Upgrading the software would restore normal operation, but because the original knowledge relationships had already been used to generate processing strategies for multiple subsequent work orders, newly arriving work orders of the same type still prioritized generating a compressor replacement strategy and continued to feed new repair results back to the knowledge graph. This reinforced erroneous repair perceptions, leading to remote processing work orders being incorrectly upgraded to on-site repair work orders. Software issues were continuously assigned to hardware repair personnel, ultimately affecting the processing paths, resource allocation, and workflow outcomes of subsequent work orders. This demonstrates that existing technologies lack a mechanism to distinguish between objective equipment conditions and repair perceptions and to utilize both to generate work order processing strategies. This makes it easy for repair perceptions to continuously deviate from the true state of the equipment, resulting in after-sales work order processing decisions being influenced by historical erroneous perceptions for a long time. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent allocation and processing system for after-sales work orders based on knowledge graphs, aiming to solve the problems mentioned in the background art.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] A knowledge graph-based intelligent allocation and processing system for after-sales work orders, the system comprising:

[0007] The dual-graph construction module is used to construct a fault fact graph and a maintenance cognition graph based on the equipment operation facts and maintenance cognition information in historical after-sales work orders, and to establish a one-to-one or one-to-many cognitive mapping relationship between fault fact nodes and maintenance cognition nodes in the two graphs.

[0008] The work order fact location module is used to determine the corresponding target fault fact node in the fault fact map based on the equipment operation facts in the after-sales work order to be processed.

[0009] The cognitive conflict identification module is used to obtain the corresponding maintenance cognitive node based on the target fault fact node through cognitive mapping relationship. When there are multiple corresponding maintenance cognitive nodes and there is a conflict among the maintenance cognitive nodes, a cognitive conflict event is generated.

[0010] The cognitive verification path generation module is used to generate corresponding cognitive verification paths based on cognitive conflict events. The cognitive verification path includes at least one verification node, each verification node corresponds to a maintenance cognitive node, and the verification conditions and verification dependencies of the corresponding maintenance cognitive node are recorded.

[0011] The processing strategy generation module is used to verify each verification node according to the cognitive verification path, determine candidate maintenance cognitive nodes based on the verification results, and determine the target maintenance cognitive node and the work order processing strategy corresponding to the after-sales work order to be processed based on the strategy evaluation value of the corresponding processing measures of the candidate maintenance cognitive nodes.

[0012] The work order allocation module is used to determine the target personnel, target processing nodes and work order flow paths according to the work order processing strategy, and to control the after-sales work orders to be processed to complete the work order processing according to the work order flow path and generate processing results.

[0013] The graph update module is used to update the maintenance cognition nodes and cognition mapping relationships in the maintenance cognition graph based on the actual processing process and the actual processing results of the after-sales work orders to be processed. When it is confirmed that there are new or corrected equipment operation facts, the fault fact nodes in the fault fact graph are updated simultaneously, so that the updated dual graphs can participate in the generation of work order processing strategies for subsequent after-sales work orders to be processed.

[0014] The above-described solution of the present invention has at least the following beneficial effects:

[0015] This invention constructs a fault fact map and a maintenance cognition map separately, and establishes a cognitive mapping relationship between fault fact nodes and maintenance cognition nodes. It separates and stores equipment operation facts and maintenance cognition information, so that equipment operation facts remain stable as objective facts in the long term, and maintenance cognition can be maintained and updated independently. This avoids the direct solidification of historical maintenance cognition into equipment facts, thereby distinguishing between the objective operating status of equipment and the technical judgment of maintenance personnel, and providing an independent data foundation for subsequent maintenance cognition verification.

[0016] Furthermore, this invention obtains multiple corresponding maintenance cognition nodes based on the target fault fact node, generates cognitive conflict events for conflicting maintenance cognitions, and constructs a cognitive verification path including verification nodes, verification conditions, and verification dependencies. The conflicting maintenance cognitions are verified step by step, so that the maintenance cognitions no longer directly participate in the generation of work order processing strategies, but complete the authenticity verification before the generation of work order processing strategies. This reduces the possibility of historically erroneous maintenance cognitions continuing to participate in subsequent work order processing decisions and improves the reliability of the target maintenance cognition.

[0017] Furthermore, after obtaining candidate maintenance cognition nodes, this invention calculates a strategy evaluation value by combining the processing resources and processing efficiency of the corresponding processing measures, and determines the target maintenance cognition node and the corresponding work order processing strategy based on the strategy evaluation value. This ensures that the work order processing strategy is not only based on verified maintenance cognition, but also comprehensively considers maintenance resource allocation and processing efficiency, achieving synergistic optimization of maintenance cognition reliability and processing execution efficiency. This reduces the problem of unreasonable selection of maintenance methods, processing personnel, and work order flow paths due to erroneous maintenance cognition, and improves the intelligent allocation and processing efficiency of after-sales work orders.

[0018] Furthermore, this invention updates the maintenance cognitive graph and cognitive mapping relationship based on the work order processing process and results, and simultaneously updates the fault fact graph when new or corrected facts about equipment operation occur. This ensures that the updates to maintenance cognition and objective facts about equipment act on their respective graphs, preventing changes in maintenance cognition from directly affecting the facts about equipment operation. At the same time, the updated dual graphs continuously participate in the generation of subsequent work order processing strategies, forming a closed-loop update mechanism based on the co-evolution of equipment operation facts and maintenance cognition. This improves the adaptability of the knowledge graph to subsequent after-sales work order processing and prevents the continuous reinforcement of historical erroneous maintenance cognition from affecting subsequent work order processing decisions. Attached Figure Description

[0019] Figure 1 This is an architecture diagram of an intelligent after-sales work order allocation and processing system based on knowledge graph, provided by an embodiment of the present invention.

[0020] Figure 2 This is a flowchart of a strategy evaluation process in an intelligent after-sales work order allocation and processing system based on knowledge graphs, provided by an embodiment of the present invention. Detailed Implementation

[0021] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0022] like Figure 1 As shown, an embodiment of the present invention proposes an intelligent allocation and processing system for after-sales work orders based on a knowledge graph. The system includes:

[0023] The dual-graph construction module is used to construct a fault fact graph and a maintenance cognition graph based on the equipment operation facts and maintenance cognition information in historical after-sales work orders, and to establish a one-to-one or one-to-many cognitive mapping relationship between fault fact nodes and maintenance cognition nodes in the two graphs.

[0024] The work order fact location module is used to determine the corresponding target fault fact node in the fault fact map based on the equipment operation facts in the after-sales work order to be processed.

[0025] The cognitive conflict identification module is used to obtain the corresponding maintenance cognitive node based on the target fault fact node through cognitive mapping relationship. When there are multiple corresponding maintenance cognitive nodes and there is a conflict among the maintenance cognitive nodes, a cognitive conflict event is generated.

[0026] The cognitive verification path generation module is used to generate corresponding cognitive verification paths based on cognitive conflict events. The cognitive verification path includes at least one verification node, each verification node corresponds to a maintenance cognitive node, and the verification conditions and verification dependencies of the corresponding maintenance cognitive node are recorded.

[0027] The processing strategy generation module is used to verify each verification node according to the cognitive verification path, determine candidate maintenance cognitive nodes based on the verification results, and determine the target maintenance cognitive node and the work order processing strategy corresponding to the after-sales work order to be processed based on the strategy evaluation value of the corresponding processing measures of the candidate maintenance cognitive nodes.

[0028] The work order allocation module is used to determine the target personnel, target processing nodes and work order flow paths according to the work order processing strategy, and to control the after-sales work orders to be processed to complete the work order processing according to the work order flow path and generate processing results.

[0029] The graph update module is used to update the maintenance cognition nodes and cognition mapping relationships in the maintenance cognition graph based on the actual processing process and the actual processing results of the after-sales work orders to be processed. When it is confirmed that there are new or corrected equipment operation facts, the fault fact nodes in the fault fact graph are updated simultaneously, so that the updated dual graphs can participate in the generation of work order processing strategies for subsequent after-sales work orders to be processed.

[0030] In this embodiment of the invention, the invention constructs a fault fact map and a maintenance cognition map respectively, and establishes a cognitive mapping relationship between the two. When maintenance cognitions conflict, a cognitive verification path is generated to verify the conflicting maintenance cognitions. In combination with processing resources and processing efficiency, a work order processing strategy is generated. This realizes the separate management and collaborative application of equipment operation facts and maintenance cognitions, avoids the continuous participation of historical erroneous maintenance cognitions in subsequent work order processing decisions, and improves the rationality of after-sales work order processing strategy generation and intelligent work order allocation.

[0031] In a preferred embodiment of the present invention, the processing procedure of the dual-map construction module includes:

[0032] Retrieve multiple completed historical after-sales work orders and extract equipment operation facts and maintenance knowledge information from each historical after-sales work order; among them, equipment operation facts describe the actual operating status of the equipment when the failure occurred; maintenance knowledge information includes the cause of the failure, the handling measures and the handling results for the corresponding failure.

[0033] The equipment operation facts in each historical after-sales work order are classified, and the equipment operation facts that belong to the same fault handling process and can jointly represent the corresponding fault status are associated to form corresponding fault fact data; fault fact nodes are established according to each fault fact data, and the corresponding equipment operation facts are stored in the fault fact nodes;

[0034] When different historical after-sales work orders contain the same equipment operation facts, the corresponding equipment operation facts are associated with the same fault fact node; when there are differences in equipment operation facts, corresponding fault fact nodes are established separately; multiple fault fact nodes and the association relationships formed between nodes based on equipment operation facts together constitute a fault fact map;

[0035] The maintenance knowledge information in each historical after-sales work order is classified, and the cause of failure, handling measures and handling results corresponding to the same fault handling are associated to form maintenance knowledge data; maintenance knowledge nodes are established according to each maintenance knowledge data, and the cause of failure, handling measures and handling results are stored as maintenance knowledge information in the corresponding maintenance knowledge nodes;

[0036] Further extract the equipment model, software version and operating environment from each historical after-sales work order, and store them as applicable feature information of the corresponding maintenance knowledge node.

[0037] When the causes of failure, handling measures, and handling results in different historical after-sales work orders are all the same, the corresponding maintenance knowledge information is associated with the same maintenance knowledge node; when at least one of the causes of failure, handling measures, or handling results is different, corresponding maintenance knowledge nodes are established respectively; multiple maintenance knowledge nodes and the maintenance knowledge information and applicable feature information stored in the corresponding maintenance knowledge nodes together constitute a maintenance knowledge graph;

[0038] Based on the correspondence between equipment operation facts and maintenance knowledge information in each historical after-sales work order, the corresponding fault fact nodes and maintenance knowledge nodes are determined, and a cognitive mapping relationship is established between the corresponding fault fact nodes and maintenance knowledge nodes. Among them, a fault fact node can establish a one-to-one cognitive mapping relationship with a maintenance knowledge node, or it can establish a one-to-many cognitive mapping relationship with multiple maintenance knowledge nodes respectively, so as to reflect the situation that the same fault fact may correspond to different maintenance knowledge.

[0039] When adding a new historical after-sales work order, extract equipment operation facts, maintenance knowledge information, and applicable characteristic information from the new historical after-sales work order; if a corresponding fault fact node exists in the fault fact graph, associate the new equipment operation fact with the corresponding fault fact node; otherwise, create a new fault fact node; if a corresponding maintenance knowledge node exists in the maintenance knowledge graph, associate the new maintenance knowledge information and applicable characteristic information with the corresponding maintenance knowledge node; otherwise, create a new maintenance knowledge node; based on the correspondence between equipment operation facts and maintenance knowledge information in the new historical after-sales work order, establish a cognitive mapping relationship between the corresponding fault fact node and the maintenance knowledge node.

[0040] In this embodiment of the invention, by constructing a fault fact graph and a maintenance cognition graph respectively, and establishing a one-to-one or one-to-many cognitive mapping relationship between fault fact nodes and maintenance cognition nodes, decoupled management of equipment operation facts and maintenance cognition information is achieved. When a pending after-sales work order enters the system, the corresponding fault fact node can be quickly located based on the equipment operation facts, and the associated maintenance cognition node can be further obtained. This provides a data foundation for subsequent cognitive conflict identification, cognitive verification path generation, and work order processing strategy generation, improving the retrieval efficiency and accuracy of historical maintenance cognition.

[0041] In a preferred embodiment of the present invention, the processing procedure of the work order fact location module includes:

[0042] Obtain pending after-sales work orders and extract equipment operation facts from them; equipment operation facts include at least one of equipment status and fault logs, and each equipment operation fact consists of at least one fault fact item and its corresponding fact value; wherein, equipment status may include at least one of equipment operating parameters, component status, alarm status and fault phenomenon; fault logs may include at least one of fault codes, alarm codes, abnormal event records and operation logs;

[0043] Standardize the equipment operation facts in the after-sales work orders to be processed, convert fault fact items with the same meaning but different representations into preset standard fault fact items, and convert the corresponding fact values ​​into data formats that match the fault fact map to form fault fact data to be located;

[0044] Based on the fault fact items in the fault fact data to be located, query the fault fact nodes that store the same fault fact items in the fault fact graph, and determine the fault fact nodes obtained from the query as the fault fact nodes to be matched.

[0045] Read the equipment operation facts stored in each fault fact node to be matched, and match the fault fact items and fact values ​​corresponding to the after-sales work orders to be processed with the corresponding fault fact items and fact values ​​in each fault fact node to be matched, so as to obtain the fact matching results corresponding to each fault fact node to be matched.

[0046] Specifically, for discrete fact values, when the fact value in the after-sales work order to be processed is the same as or has a preset equivalence relationship with the corresponding fact value in the fault fact node to be matched, the corresponding fault fact item is determined to match; for continuous fact values, when the fact value in the after-sales work order to be processed is within the preset numerical range corresponding to the fault fact node to be matched, the corresponding fault fact item is determined to match; for log text type fact values, when the fault code is the same, or the fault feature extracted from the log text is the same as the fault feature stored in the fault fact node to be matched, the corresponding fault fact item is determined to match.

[0047] Based on the matching results of each fault fact item, determine the fact matching result corresponding to each fault fact node to be matched; the fact matching result shall include at least the matched fault fact items, the unmatched fault fact items, and the number or proportion of matched fault fact items.

[0048] The fault fact nodes that meet the preset node positioning conditions are identified as target fault fact nodes. The node positioning conditions include: all the preset necessary fault fact items are matched, and the number or matching ratio of the matched fault fact items reaches the preset positioning threshold.

[0049] When only one fault fact node to be matched meets the node location conditions, the fault fact node is determined as the target fault fact node; when multiple fault fact nodes to be matched meet the node location conditions, the fault fact node with the highest matching degree is determined as the target fault fact node based on the number of matched fault fact items, the matching ratio, or the importance of preset fault fact items corresponding to each fault fact node to be matched.

[0050] After identifying the target fault fact node, the node identifier of the target fault fact node and the corresponding equipment operation facts are sent to the cognitive conflict identification module, so that the cognitive conflict identification module can obtain the maintenance cognitive node associated with the target fault fact node through the cognitive mapping relationship corresponding to the target fault fact node.

[0051] In this embodiment of the invention, by standardizing the equipment operation facts in the after-sales work order to be processed, and by performing node query and fact matching in the fault fact map based on the fault fact items and corresponding fact values, the target fault fact node corresponding to the current actual operation of the equipment can be determined, providing an accurate fault fact location basis for subsequent acquisition of associated maintenance cognitive nodes, identification of cognitive conflicts, and generation of cognitive verification paths.

[0052] In a preferred embodiment of the present invention, the processing procedure of the cognitive conflict identification module includes:

[0053] Obtain the target fault fact node determined by the work order fact location module, and query the cognitive mapping relationship corresponding to the target fault fact node to obtain the maintenance cognitive node associated with the target fault fact node;

[0054] When the target fault fact node is associated with only one maintenance cognition node, it is determined that there is no cognitive conflict between multiple maintenance cognition nodes, and the maintenance cognition node is sent to the processing strategy generation module or the subsequent preset processing module to generate the work order processing strategy corresponding to the after-sales work order to be processed based on the maintenance cognition node.

[0055] When the target fault fact node is associated with at least two maintenance cognition nodes, the maintenance cognition information stored in each maintenance cognition node is extracted respectively; the maintenance cognition information includes at least the fault cause, handling measures and handling results, wherein the handling results are the historical handling results formed after the corresponding handling measures are executed on the corresponding historical after-sales work orders;

[0056] Based on the type of maintenance cognitive information, the fault causes, handling measures and historical handling results corresponding to each maintenance cognitive node are compared to determine whether there are inconsistencies in maintenance cognitive information between each maintenance cognitive node.

[0057] Specifically, the causes of failures corresponding to each maintenance cognitive node are compared. When the causes of failures corresponding to at least two maintenance cognitive nodes are different, or the failure formation mechanisms represented by each cause of failure are inconsistent, it is determined that there is a conflict of causes of failures between the corresponding maintenance cognitive nodes, and the conflict of causes of failures is identified as the corresponding conflict type.

[0058] Compare the handling measures corresponding to each maintenance cognition node; when the handling measures corresponding to at least two maintenance cognition nodes are different, or the maintenance objects, maintenance operations or handling parameters corresponding to each handling measure are inconsistent, it is determined that there is a conflict of handling measures between the corresponding maintenance cognition nodes, and the conflict of handling measures is determined as the corresponding conflict type.

[0059] Compare the historical processing results corresponding to each maintenance cognition node; when the historical processing results corresponding to at least two maintenance cognition nodes are different, or the equipment recovery status, fault elimination status or historical fault recurrence status after implementing the corresponding processing measures are inconsistent, it is determined that there is a conflict of processing results between the corresponding maintenance cognition nodes, and the conflict of processing results is determined as the corresponding conflict type.

[0060] When the fault causes, handling measures, and historical handling results corresponding to each maintenance cognition node are consistent, it is determined that there is no maintenance cognition conflict between the maintenance cognition nodes. In this case, the corresponding maintenance cognition node can be sent to the subsequent processing module as a conflict-free maintenance cognition node without generating a cognition conflict event.

[0061] When there is at least one conflict among the causes of failure, handling measures, and handling results, the maintenance cognitive node in conflict is determined according to the corresponding conflict type; wherein, for each conflict type, the maintenance cognitive node in which the corresponding maintenance cognitive information is inconsistent is determined as the conflicting maintenance cognitive node corresponding to that conflict type.

[0062] When there are at least two types of conflicts among maintenance cognitive nodes, record the conflict type corresponding to each conflict and establish the association between each conflict type and the corresponding conflict maintenance cognitive node, so that the same maintenance cognitive node can be associated with multiple conflict types it participates in.

[0063] Extract the applicable feature information corresponding to each maintenance cognitive node that has a conflict; the applicable feature information includes at least the equipment model, software version and operating environment, and is used to characterize the equipment conditions and operating conditions to which the corresponding maintenance cognitive information is applicable;

[0064] For each maintenance knowledge node that is in conflict, the node identifier, the corresponding maintenance knowledge information, and the applicable characteristic information of the maintenance knowledge node are associated to maintain the correspondence between the cause of the fault, the handling measures, the historical handling results and their applicable scope.

[0065] The cognitive conflict event is generated by associating the conflicting maintenance cognitive nodes, the identified conflict types, and the applicable characteristic information corresponding to each maintenance cognitive node. The cognitive conflict event includes at least the target fault fact node identifier, the conflicting maintenance cognitive node identifier, the conflict type corresponding to each maintenance cognitive node, and the applicable characteristic information.

[0066] Among them, the target fault fact node identifier is used to indicate the equipment operation fact corresponding to this cognitive conflict; the maintenance cognitive node identifier where the conflict occurred is used to determine the object that needs to be verified for maintenance cognition in the future; the conflict type is used to determine the verification basis to be used for subsequent verification; and the applicable feature information is used to determine the applicable conditions for maintenance cognition corresponding to each maintenance cognitive node.

[0067] When multiple maintenance cognitive nodes have inconsistencies only in some maintenance cognitive information, only the conflict type corresponding to the inconsistent maintenance cognitive information and the maintenance cognitive nodes involved in the conflict are written into the cognitive conflict event; maintenance cognitive nodes that do not participate in the corresponding conflict type are not identified as the verification objects corresponding to that conflict type.

[0068] After a cognitive conflict event is generated, it is sent to the cognitive verification path generation module, so that the cognitive verification path generation module can establish corresponding verification nodes and generate cognitive verification paths according to the maintenance cognitive node that has conflicted, the conflict type, and the applicable feature information corresponding to each maintenance cognitive node.

[0069] In this embodiment of the invention, by acquiring multiple maintenance cognition nodes associated with the target fault fact node, and comparing the fault causes, handling measures and historical handling results corresponding to each maintenance cognition node, it is possible to identify cognitive conflicts between different maintenance cognitions corresponding to the same equipment operation fact; by further associating conflicting maintenance cognition nodes, conflict types and applicable characteristic information, it is possible to provide a clear data foundation for subsequently determining the verification object, verification basis and verification conditions, and avoid unverified conflicting maintenance cognitions directly participating in the generation of work order processing strategies.

[0070] In a preferred embodiment of the present invention, the processing procedure of the cognitive verification path generation module includes:

[0071] Acquire cognitive conflict events generated by the cognitive conflict identification module, and extract target fault fact nodes, conflicting maintenance cognitive nodes, conflict types corresponding to each maintenance cognitive node, and applicable feature information from the cognitive conflict events;

[0072] Establish corresponding verification nodes for each maintenance cognition node that has a conflict, so that one verification node corresponds to one maintenance cognition node; write the node identifier, maintenance cognition information, applicable feature information and corresponding conflict type of the maintenance cognition node into the corresponding verification node;

[0073] When the same maintenance cognitive node participates in at least two conflict types, only one verification node is established for that maintenance cognitive node, and the content to be verified and the verification conditions corresponding to each conflict type are recorded in the verification node.

[0074] Based on the conflict type corresponding to each maintenance cognition node, the content to be verified for the corresponding verification node is determined. Specifically, when the conflict type is a fault cause conflict, the fault cause stored in the maintenance cognition node is determined as the content to be verified; when the conflict type is a handling measure conflict, the handling measure stored in the maintenance cognition node is determined as the content to be verified; when the conflict type is a handling result conflict, the handling result stored in the maintenance cognition node is determined as the content to be verified.

[0075] The corresponding verification basis is determined according to the conflict type. When the conflict type is a fault cause conflict, the equipment status and fault log in the pending after-sales work order are used as the verification basis, and the equipment operation facts stored in the fault fact node mapped by the maintenance cognition node are combined to verify the fault cause corresponding to the maintenance cognition node.

[0076] Among them, based on the cause of the fault to be verified, the status parameters, fault codes, alarm information, abnormal status of components or log characteristics that can characterize whether the cause of the fault is valid are determined, and the corresponding data are extracted from the equipment status and fault log of the after-sales work order to be processed.

[0077] When the conflict type is a conflict of handling measures, the equipment model, software version and operating environment corresponding to the maintenance cognition node are used as the verification basis. The equipment model, software version and operating environment are matched with the equipment model, software version and operating environment corresponding to the after-sales work order to be processed, so as to verify whether the corresponding handling measures are applicable to the equipment corresponding to the current after-sales work order to be processed.

[0078] The operating environment includes at least one of the following: the network environment, temperature and humidity environment, load status, external connection environment, system configuration environment, and the device usage scenario.

[0079] When the conflict type is a conflict of processing results, the historical processing process, post-processing equipment status and historical fault recurrence records corresponding to the maintenance cognition node are used as the verification basis, and the processing results recorded in the maintenance cognition node are verified in combination with the historical verification data in the corresponding historical after-sales work orders.

[0080] Among them, the historical processing process is used to characterize whether the corresponding processing measures have been executed according to the predetermined steps; the post-processing equipment status is used to characterize whether the equipment has returned to the preset operating state after the processing measures have been executed; and the historical fault recurrence record is used to characterize whether the same fault has occurred again after the corresponding processing measures have been executed.

[0081] For each verification node, the verification field is extracted based on the content to be verified and the corresponding verification basis; the verification field is a data field used to determine whether the maintenance knowledge information is valid or applicable to the after-sales work order to be processed.

[0082] Filter the key verification fields directly related to the corresponding conflict type from the extracted verification fields; for conflict of cause of failure, key verification fields include fields used to characterize the equipment status or fault log features; for conflict of handling measures, key verification fields include equipment model field, software version field, and operating environment field; for conflict of handling result, key verification fields include historical handling steps field, post-handling equipment status field, and fault recurrence status field.

[0083] Obtain the actual field values ​​corresponding to each key verification field, and match the actual field values ​​with the preset field values, preset value ranges, preset states or preset features required by the corresponding maintenance cognition nodes to obtain the verification results corresponding to each key verification field;

[0084] Validation expressions are constructed based on each key validation field, the logical relationship between key validation fields, and their corresponding preset matching requirements, and the validation expressions are used as validation conditions for the corresponding validation nodes.

[0085] The validation expression includes at least one field judgment expression; when the validation condition includes multiple key validation fields, the field judgment expressions are connected using logical AND, logical OR, or logical NOT relationships according to the conditions for the establishment of the corresponding maintenance knowledge information.

[0086] For example, when there is a conflict in the cause of the fault, and the fault cause corresponding to the maintenance cognition node is evaporator frosting, "evaporator surface is in a frosted state" and "evaporator temperature is lower than the preset temperature threshold" can be constructed as field judgment expressions respectively, and corresponding verification expressions can be constructed through logical AND relation.

[0087] When there is a conflict in the handling measures, and the handling measures corresponding to the maintenance cognition node are applicable to the preset equipment model, preset software version and preset operating environment, the equipment model matching results, software version matching results and operating environment matching results can be used to construct the corresponding verification expression through logical AND relation.

[0088] For conflicting processing results, a corresponding verification expression can be constructed based on whether the historical processing steps are complete, whether the equipment status after processing reaches the preset recovery state, and whether there are records of the same fault recurrence within the preset time range.

[0089] When the same verification node corresponds to multiple conflict types, corresponding verification conditions are generated for each conflict type, and the generated multiple verification conditions are recorded in the same verification node;

[0090] Determine the execution input and verification result for each verification condition; when the execution of one verification condition requires the verification result of another verification condition, determine the verification condition that provides the verification result as the pre-verification condition, determine the verification condition that uses the verification result as the post-verification condition, and establish a verification dependency relationship between the pre-verification condition and the post-verification condition.

[0091] For example, when the verification of the processing results requires the processing measures to pass the applicability verification as a prerequisite, the verification conditions corresponding to the conflict of processing measures are determined as the pre-verification conditions, the verification conditions corresponding to the conflict of processing results are determined as the post-verification conditions, and a verification dependency relationship is established from the verification conditions of processing measures to the verification conditions of processing results.

[0092] Based on the verification dependencies between each verification condition, the execution order of the corresponding verification conditions and verification nodes is determined; for verification conditions with verification dependencies, the preceding verification conditions are executed first, followed by the subsequent verification conditions; for verification conditions without verification dependencies, the execution order between them is not restricted, and the corresponding verification conditions are allowed to be executed in parallel.

[0093] When a verification condition in one verification node depends on the verification result of a verification condition in another verification node, a verification dependency relationship is established between the corresponding verification nodes; when the verification conditions in two verification nodes do not have a verification result reference relationship, a verification dependency relationship is not established between the two verification nodes.

[0094] Based on each verification node, the verification conditions recorded in each verification node, and the verification dependencies between verification nodes, the verification nodes are connected in the determined execution order to generate a cognitive verification path.

[0095] The cognitive verification path records at least the maintenance cognitive node, the content to be verified, the verification basis, the verification conditions, the preceding verification node, and the following verification node for each verification node; for verification nodes that do not have preceding verification nodes, they are identified as the starting verification node that can be directly executed.

[0096] After generating the cognitive verification path, the cognitive verification path is sent to the processing strategy generation module, so that the processing strategy generation module executes the verification conditions corresponding to each verification node according to the verification dependency relationship between the verification nodes, and determines the maintenance cognitive node that passes the verification as a candidate maintenance cognitive node.

[0097] Among them, maintenance cognitive nodes that fail verification are not identified as candidate maintenance cognitive nodes and do not participate in subsequent strategy evaluation; when there are multiple verified maintenance cognitive nodes, the processing strategy generation module further evaluates the strategy based on the verification status, verification path, processing resources and processing efficiency of each candidate maintenance cognitive node.

[0098] In this embodiment of the invention, by determining corresponding verification criteria for conflicts in fault causes, handling measures, and handling results, and generating verification conditions in the order of extracting verification fields, filtering key verification fields, matching verification results, and constructing verification expressions, each maintenance cognitive node can obtain an executable verification method corresponding to its conflict type. Furthermore, by establishing verification dependencies based on the verification result reference relationships between verification conditions, the execution order of each verification condition and verification node can be determined, and verification nodes without verification dependencies can be executed in parallel, thereby improving the accuracy and execution efficiency of the cognitive verification path.

[0099] like Figure 2 As shown, in a preferred embodiment of the present invention, the processing procedure of the processing strategy generation module includes:

[0100] Obtain the cognitive verification path generated by the cognitive verification path generation module, and read the verification nodes, verification conditions corresponding to each verification node, and verification dependencies between verification nodes recorded in the cognitive verification path;

[0101] The execution order of each verification node and verification condition is determined based on the verification dependencies between verification nodes. For verification conditions with verification dependencies, the pre-verification conditions are executed first, and the corresponding post-verification conditions are executed after obtaining the verification results of the pre-verification conditions. For verification conditions without verification dependencies, the corresponding verification conditions are executed separately or in parallel.

[0102] Execute the verification conditions recorded in each verification node, and record the execution status and verification result of each verification condition; wherein, the execution status of the verification condition includes at least an executed status and an unexecuted status, and the verification result includes at least a verification requirement met and a verification requirement not met;

[0103] When a verification node records at least two verification conditions, each verification condition in the verification node is executed respectively, and the verification status of the corresponding maintenance cognition node is determined based on the verification results of each verification condition.

[0104] For the i-th maintenance cognition node, the total number of verification conditions recorded in its corresponding verification node is denoted as Ni, the number of verification conditions that meet the verification requirements is denoted as Ai, and the number of verification conditions actually executed during the cognition verification process is denoted as Ei.

[0105] When a post-validation condition is not executed because its pre-validation condition does not meet the validation requirements, the post-validation condition is recorded as an unexecuted validation condition and included in the total number of validation conditions Ni, but not included in the number of validation conditions actually executed Ei and the number of validation conditions that meet the validation requirements Ai.

[0106] Based on the number of verification conditions Ai that meet the verification requirements and the total number of verification conditions Ni, determine the verification evaluation information corresponding to the i-th maintenance cognition node; based on the number of verification conditions Ei that are actually executed and the total number of verification conditions Ni, determine the path evaluation information corresponding to the i-th maintenance cognition node.

[0107] Verification evaluation information is used to characterize the degree to which the maintenance cognitive information of the corresponding maintenance cognitive node meets the verification requirements; path evaluation information is used to characterize the degree of actual execution of the verification conditions in the corresponding cognitive verification path.

[0108] The verification pass conditions are used to determine whether each maintenance cognition node has passed the verification. The verification pass conditions may include: all the preset necessary verification conditions are met, and the number of verification conditions that meet the verification requirements or the corresponding verification evaluation information reaches the preset verification threshold.

[0109] Maintenance cognitive nodes that meet the verification pass conditions will be identified as candidate maintenance cognitive nodes; maintenance cognitive nodes that do not meet the verification pass conditions will not be identified as candidate maintenance cognitive nodes and will not participate in the calculation of subsequent strategy evaluation values.

[0110] For each candidate maintenance awareness node, obtain the historical after-sales work order corresponding to the candidate maintenance awareness node, and extract the personnel type, processing node, spare parts type, tool type, processing time, and work order flow record used when executing the corresponding processing measures from the historical after-sales work order;

[0111] When multiple historical after-sales work orders adopt the same or equivalent handling measures, the types of personnel handling, handling nodes, spare parts types, tool types, handling time and work order flow records corresponding to each historical after-sales work order can be statistically analyzed, and historical handling data of corresponding candidate maintenance cognition nodes can be formed based on the statistical results.

[0112] The personnel resource information required to implement the corresponding processing measures shall be determined based on the type of personnel being processed; the personnel resource information may include at least one of the following: the professional type, skill level, number of personnel, and personnel occupancy status of the personnel being processed;

[0113] The information on the processing nodes required to execute the corresponding processing measures is determined based on the processing nodes. The processing node information may include at least one of the following: the number of processing nodes, node type, node collaboration relationship, and node resource usage.

[0114] Determine the spare parts resource information required to implement the corresponding treatment measures based on the spare parts type; the spare parts resource information may include at least one of the following: spare parts type, spare parts quantity, spare parts acquisition difficulty, and spare parts occupancy cost.

[0115] Determine the tool resource information required to implement the corresponding treatment measures based on the tool type; the tool resource information may include at least one of the following: tool type, tool quantity, special tool requirements, and tool occupancy status;

[0116] Personnel resource information, processing node information, spare parts resource information, and tool resource information are collectively identified as the processing resources for the corresponding candidate maintenance cognition nodes, and the corresponding resource consumption parameters are determined based on the actual occupancy or resource consumption level of each type of processing resource.

[0117] Among them, the resource consumption parameters correspond to the personnel resource parameters, processing node parameters, spare parts resource parameters and tool resource parameters in the strategy evaluation value calculation formula, so that the four types of resources, personnel, processing nodes, spare parts and tools, participate in the calculation of resource consumption information respectively.

[0118] Based on the processing time in historical after-sales work orders, determine the time consumption parameters required to execute the corresponding handling measures; the processing time can be the duration from when the work order enters the first processing node to when the fault handling is completed, or it can be the cumulative value of the time consumed by each actual maintenance operation;

[0119] Based on the work order flow records in the historical after-sales work orders, determine the corresponding flow consumption parameters; the work order flow records may include at least one of the following: the number of processing nodes the work order passes through, the number of flow between nodes, the number of reassignments, the waiting time, and the flow time between nodes;

[0120] The time consumption parameter and the circulation consumption parameter are jointly determined as the processing timeliness data of the corresponding candidate maintenance cognition node, and the corresponding processing timeliness information is determined based on the processing timeliness data;

[0121] Among them, the time consumption parameter corresponds to the processing time parameter in the strategy evaluation value calculation formula, and the circulation consumption parameter corresponds to the work order circulation parameter in the strategy evaluation value calculation formula, so that the processing time and work order circulation status can be used together to calculate the processing timeliness information.

[0122] Since different types of resource consumption parameters and processing time parameters have different data units and value ranges, before calculating resource consumption information and processing time information, the personnel resource parameters, processing node parameters, spare parts resource parameters, tool resource parameters, processing time parameters and work order flow parameters are normalized respectively.

[0123] For evaluation parameters such as personnel occupancy, number of processing nodes, spare parts consumption, tool occupancy, processing time, and work order circulation consumption, the smaller the value, the better the processing strategy. A predetermined reverse normalization method is used to process them so that the larger the normalized parameter value, the less resource consumption or the higher the processing efficiency of the corresponding candidate maintenance cognition node.

[0124] When the maximum and minimum values ​​of the same evaluation parameter are the same in each candidate maintenance cognition node, the normalized parameter value corresponding to each candidate maintenance cognition node is determined as a preset unified value according to the established calculation rules, so as to avoid the situation where the denominator is zero in the normalization calculation process.

[0125] Based on the normalized personnel resource parameters, processing node parameters, spare parts resource parameters, and tool resource parameters, the resource consumption information corresponding to each candidate maintenance cognition node is determined according to the established resource consumption information calculation formula.

[0126] Based on the normalized processing time parameters and work order flow parameters, the processing time information corresponding to each candidate maintenance cognition node is determined according to the established processing time information calculation formula.

[0127] Based on the verification evaluation information, path evaluation information, resource consumption information and processing time information corresponding to each candidate maintenance cognition node, the strategy evaluation value corresponding to each candidate maintenance cognition node is calculated according to the established strategy evaluation value calculation formula.

[0128] Among them, the verification evaluation information is used to characterize the degree of verification matching between the maintenance cognitive information corresponding to the candidate maintenance cognitive node and the current after-sales work order to be processed; the path evaluation information is used to characterize the degree of execution completeness of the corresponding cognitive verification path; the resource consumption information is used to characterize the comprehensive resource consumption of personnel, processing nodes, spare parts and tools required to execute the corresponding processing measures; and the processing time information is used to characterize the processing time required to execute the corresponding processing measures and the comprehensive efficiency of the work order flow process.

[0129] Compare the strategy evaluation values ​​corresponding to each candidate maintenance knowledge node, and determine the candidate maintenance knowledge node with the highest strategy evaluation value as the target maintenance knowledge node;

[0130] When at least two candidate maintenance cognitive nodes have the same strategy evaluation value, the candidate maintenance cognitive node with higher verification evaluation information is given priority to be determined as the target maintenance cognitive node; when both the strategy evaluation value and the verification evaluation information are the same, the candidate maintenance cognitive node with higher processing timeliness information is given priority to be determined as the target maintenance cognitive node.

[0131] When a unique target maintenance cognitive node cannot be determined based on the strategy evaluation value, verification evaluation information, and processing time information, the target maintenance cognitive node can be determined from the corresponding candidate maintenance cognitive nodes according to the preset selection rules. The preset selection rules may include giving priority to candidate maintenance cognitive nodes that have been used more times in the past, have had their most recent successful processing time, or have a low fault recurrence rate.

[0132] After identifying the target maintenance awareness node, read the processing measures recorded in the target maintenance awareness node, and determine the maintenance method corresponding to the after-sales work order to be processed based on the processing measures; the maintenance method is used to indicate at least one of the maintenance operations, maintenance steps and processing parameters that need to be performed to process the after-sales work order to be processed.

[0133] Obtain the processing resources corresponding to the target maintenance knowledge node, and determine the target processing resources required to execute the maintenance method based on the personnel type, processing node, spare parts type, and tool type required by the maintenance method;

[0134] Based on the types of personnel and processing nodes in the target processing resources, determine the processing nodes that the after-sales work orders to be processed need to go through; based on the processing sequence, data transmission, and business flow relationships between the processing nodes, generate the work order flow path corresponding to the after-sales work orders to be processed.

[0135] The work order flow path includes at least a starting processing node, one or more intermediate processing nodes, and an ending processing node; when the maintenance method only requires one processing node to complete, that processing node is used as both the starting processing node and the ending processing node.

[0136] Based on the processing time information corresponding to the target maintenance awareness node, and combined with the fault impact, service time limit, and current work order waiting status of the pending after-sales work orders, determine the processing priority of the pending after-sales work orders;

[0137] Among them, the shorter the processing time and the less work order circulation consumption corresponding to the target maintenance awareness node, the higher the processing efficiency of the corresponding maintenance method; under the same conditions, the corresponding processing priority can be configured for the pending after-sales work orders that adopt this maintenance method.

[0138] The determined repair methods, work order flow paths, and processing priorities are associated to generate work order processing strategies for pending after-sales work orders; the work order processing strategies include repair methods, work order flow methods, and priorities.

[0139] Among them, the repair method is used to indicate the specific repair content that needs to be performed on the pending after-sales work order; the work order flow method is used to indicate the processing nodes that the pending after-sales work order needs to go through and the flow order between each processing node; the priority is used to indicate the processing order of the pending after-sales work order relative to other after-sales work orders.

[0140] The generated work order processing strategy is sent to the work order allocation module, so that the work order allocation module can determine the target personnel and target processing nodes for the after-sales work orders to be processed based on the repair method, work order flow method and priority in the work order processing strategy, combined with the personnel type, skill information, processing node and current work status of each processing personnel, and complete the work order allocation and flow according to the work order flow method.

[0141] In this embodiment of the invention, by executing the verification conditions corresponding to each maintenance cognitive node according to the cognitive verification path, only maintenance cognitive nodes that meet the verification conditions are identified as candidate maintenance cognitive nodes, thus avoiding the participation of unverified or unverified maintenance cognitive nodes in the strategy evaluation. Furthermore, by combining verification evaluation information, path evaluation information, resource consumption information, and processing timeliness information to calculate the strategy evaluation value corresponding to each candidate maintenance cognitive node, the target maintenance cognitive node can be determined from multiple dimensions such as verification matching degree, path execution completeness, resource consumption, and processing efficiency. Based on the target maintenance cognitive node, a work order processing strategy including maintenance method, work order flow method, and priority is generated, thereby improving the matching degree between the work order processing strategy and the actual situation of the after-sales work orders to be processed, as well as the efficiency of after-sales work order allocation and processing.

[0142] In a preferred embodiment of the present invention, the process of calculating the strategy evaluation value corresponding to each candidate maintenance cognition node includes:

[0143] Let M be the number of candidate maintenance cognitive nodes, and let the policy evaluation value corresponding to the i-th candidate maintenance cognitive node be M. ,in: .

[0144] I. Verification Evaluation Value

[0145] Based on the number of verification nodes that meet the verification conditions in the cognitive verification path corresponding to the i-th candidate maintenance cognitive node, and the total number of verification nodes in the cognitive verification path, the verification evaluation value is calculated according to the following formula: , ;

[0146] In the formula: This represents the verification evaluation value corresponding to the i-th candidate maintenance cognition node. The larger the value, the higher the degree to which the corresponding candidate maintenance cognitive node meets the verification conditions; This represents the number of verification nodes that meet the verification conditions in the cognitive verification path corresponding to the i-th candidate maintenance cognitive node; This represents the total number of verification nodes in the cognitive verification path corresponding to the i-th candidate maintenance cognitive node.

[0147] II. Path Evaluation Value

[0148] Based on the number of verification nodes actually executed in the cognitive verification path corresponding to the i-th candidate maintenance cognitive node, and the total number of verification nodes in the cognitive verification path, the path evaluation value is calculated according to the following formula: , ;

[0149] In the formula: This represents the path evaluation value corresponding to the i-th candidate maintenance awareness node. The larger the value, the higher the actual execution level of the corresponding cognitive verification path; This represents the number of verification nodes actually executed according to the verification dependency relationship in the cognitive verification path corresponding to the i-th candidate maintenance cognitive node.

[0150] III. Resource Consumption Evaluation Value

[0151] From the historical after-sales work orders associated with the i-th candidate maintenance cognitive node, extract the personnel type, historical processing node, spare parts type, and tool type required to execute the corresponding processing measures, and determine the number of personnel types, the number of historical processing nodes, the number of spare parts types, and the number of tool types, respectively.

[0152] in: This represents the number of personnel types corresponding to the i-th candidate maintenance awareness node;

[0153] This represents the number of historical processing nodes corresponding to the i-th candidate maintenance awareness node;

[0154] This represents the number of spare parts types corresponding to the i-th candidate maintenance awareness node;

[0155] This represents the number of tool types corresponding to the i-th candidate maintenance cognitive node.

[0156] For any resource quantity parameter The corresponding resource evaluation parameters are calculated according to the following formula: ;

[0157] In the formula: This represents the resource evaluation parameters corresponding to the i-th candidate maintenance cognition node; j represents the sequence number used to compare each candidate maintenance knowledge node; This represents the maximum value among the same resource quantity parameters corresponding to all candidate maintenance knowledge nodes; This represents the minimum value among all candidate maintenance knowledge nodes corresponding to the same resource quantity parameter.

[0158] When the same resource quantity parameter corresponding to each candidate maintenance cognition node is not completely the same, the smaller the resource quantity, the larger the obtained resource evaluation parameter; when the same resource quantity parameter corresponding to each candidate maintenance cognition node is the same, the corresponding resource evaluation parameter is determined to be 1.

[0159] Based on the resource evaluation parameters corresponding to the number of personnel types, the number of historical processing nodes, the number of spare parts types, and the number of tool types, the resource consumption evaluation value is calculated according to the following formula: , ;

[0160] In the formula: This represents the resource consumption evaluation value corresponding to the i-th candidate maintenance cognitive node. The larger the value, the less processing resources are required to execute the corresponding processing measures;

[0161] This indicates the resource evaluation parameters corresponding to the number of personnel types being processed;

[0162] This represents the resource evaluation parameter corresponding to the number of historical processing nodes;

[0163] This indicates the resource evaluation parameters corresponding to the type and quantity of spare parts;

[0164] This indicates the resource evaluation parameters corresponding to the number of tool types.

[0165] IV. Processing Timeliness Evaluation Value

[0166] Extract the processing time from the historical after-sales work orders associated with the i-th candidate maintenance cognitive node, and determine the number of work order transfers based on the corresponding work order transfer records.

[0167] in: This represents the processing time required to execute the corresponding processing measure for the i-th candidate maintenance cognitive node;

[0168] This represents the number of work order transfers corresponding to the i-th candidate maintenance awareness node.

[0169] The processing time evaluation parameter is calculated using the following formula: ;

[0170] In the formula: This represents the processing time evaluation parameter corresponding to the i-th candidate maintenance awareness node;

[0171] This represents the maximum processing time among all candidate maintenance awareness nodes;

[0172] This represents the minimum processing time among all candidate maintenance awareness nodes.

[0173] Calculate the work order flow evaluation parameters using the following formula: ;

[0174] In the formula: This represents the work order flow evaluation parameter corresponding to the i-th candidate maintenance cognition node;

[0175] This represents the maximum number of work order transfers for each candidate maintenance awareness node;

[0176] This represents the minimum number of work order transfers for each candidate maintenance awareness node.

[0177] The shorter the processing time, The larger the value, the fewer the number of work order processing steps. The larger.

[0178] Based on the processing time evaluation parameters and work order flow evaluation parameters, the processing time evaluation value is calculated according to the following formula: , ;

[0179] In the formula: This represents the processing timeliness evaluation value corresponding to the i-th candidate maintenance awareness node. The larger the value, the shorter the processing time required to implement the corresponding processing measures and the fewer times the work order needs to be processed.

[0180] V. Strategy Evaluation Value

[0181] Based on the verification evaluation value, path evaluation value, resource consumption evaluation value, and processing timeliness evaluation value, the strategy evaluation value corresponding to the i-th candidate maintenance cognition node is calculated according to the following formula: , ;

[0182] In the formula: This represents the strategy evaluation value corresponding to the i-th candidate maintenance cognitive node. The larger the value, the higher the degree to which the corresponding candidate maintenance cognitive node meets the verification conditions, the higher the degree of execution of the cognitive verification path, the less processing resources are required to execute the corresponding processing measures, and the higher the processing time.

[0183] VI. Determining the Cognitive Nodes of Target Maintenance

[0184] Compare the strategy evaluation values ​​corresponding to each candidate maintenance knowledge node, and determine the sequence number corresponding to the target maintenance knowledge node according to the following formula: ;

[0185] In the formula: This indicates the sequence number corresponding to the target maintenance awareness node;

[0186] This indicates that the candidate maintenance cognitive node number corresponding to the largest strategy evaluation value is determined from the strategy evaluation values ​​corresponding to each candidate maintenance cognitive node.

[0187] The serial number is Candidate maintenance cognitive nodes are identified as target maintenance cognitive nodes, and work order processing strategies are generated based on the target maintenance cognitive nodes to be processed after-sales work orders.

[0188] When at least two candidate maintenance cognitive nodes have the same strategy evaluation value, the candidate maintenance cognitive node with the largest verification evaluation value is determined as the target maintenance cognitive node; when the corresponding verification evaluation values ​​are still the same, the candidate maintenance cognitive node with the largest processing time evaluation value is determined as the target maintenance cognitive node.

[0189] In this embodiment of the invention, by converting the verification node satisfaction status, cognitive verification path execution status, processing resource quantity, processing time, and work order circulation number into evaluation values ​​with the same range and the larger the value, the better the evaluation result, the impact of the difference in the dimensions of different evaluation information on the strategy evaluation result can be reduced, and the target maintenance cognitive node matching the after-sales work order to be processed can be determined based on the strategy evaluation value.

[0190] In a preferred embodiment of the present invention, the processing procedure of the work order allocation module includes:

[0191] Obtain the work order processing strategy generated by the processing strategy generation module, and extract the maintenance method, work order flow method and processing priority from the work order processing strategy;

[0192] The type of personnel required to perform the corresponding repair work is determined based on the repair method, and the target processing nodes that the after-sales work orders to be processed need to go through are determined based on the work order flow method.

[0193] Obtain the available processing personnel corresponding to each target processing node, and match the type of processing personnel required for the maintenance method with the personnel type and processing capabilities of each available processing personnel. Determine the target processing personnel from the processing personnel that meet the preset matching conditions.

[0194] Based on the target processing nodes and the processing order between them recorded in the work order flow method, generate the work order flow path corresponding to the after-sales work order to be processed.

[0195] Send the pending after-sales work orders, repair methods, and processing priorities to the corresponding target personnel and target processing nodes, and control the pending after-sales work orders to flow to each target processing node in sequence according to the work order flow path;

[0196] Each target processing node determines the processing order of the after-sales work orders based on their processing priority, and the corresponding target processing personnel perform the corresponding repair operations according to the repair method.

[0197] After each target processing node completes its corresponding maintenance task, the processing content, processing time, actual spare parts and tools used, and work order flow records of each target processing node are obtained to form the actual processing process corresponding to the after-sales work order to be processed.

[0198] After the work order is processed, the status of the equipment and the status of fault elimination are obtained to form the actual processing result corresponding to the after-sales work order to be processed.

[0199] The actual processing process and results are associated with the pending after-sales work orders and sent to the graph update module, so that the graph update module can update the fault fact graph, maintenance cognition graph and cognition mapping relationship based on the data generated by this work order processing.

[0200] In this embodiment of the invention, by determining the target processing personnel, target processing nodes, and work order flow paths according to the work order processing strategy, and controlling the after-sales work orders to be processed to be completed according to the work order flow paths, the processing measures corresponding to the target maintenance cognition nodes can be actually executed, and at the same time, the actual processing process and actual processing results for dual-map updates can be formed.

[0201] In a preferred embodiment of the present invention, the processing procedure of the map update module includes:

[0202] Obtain completed after-sales work orders and extract equipment operation facts, actual processing process, actual processing results, and applicable characteristic information from the after-sales work orders;

[0203] The actual processing includes the actual processing measures adopted, the types of personnel involved, the processing nodes, the spare parts and tools actually used, the processing time, and the work order flow records; the actual processing results include the equipment status and fault handling results after processing; and the applicable characteristic information includes the equipment model, software version, and operating environment.

[0204] Based on the equipment operation facts corresponding to the after-sales work order, the corresponding fault fact nodes are determined in the fault fact map, and the target maintenance cognition nodes determined by the processing strategy generation module are obtained.

[0205] Based on the confirmed fault causes, actual handling measures, and actual handling results during the actual handling process, the maintenance knowledge information generated by this after-sales work order is determined, and the maintenance knowledge information is compared with the maintenance knowledge information stored in the target maintenance knowledge node.

[0206] When the maintenance knowledge information generated by this after-sales work order is consistent with the maintenance knowledge information stored in the target maintenance knowledge node, the actual processing process, actual processing result and applicable characteristic information corresponding to this after-sales work order will be associated with the target maintenance knowledge node, and the cognitive mapping relationship between the fault fact node and the target maintenance knowledge node will be retained.

[0207] When the maintenance knowledge information generated by this after-sales work order is inconsistent with the maintenance knowledge information stored in the target maintenance knowledge node, the maintenance knowledge graph is queried to see if there is a maintenance knowledge node corresponding to this maintenance knowledge information.

[0208] When a corresponding maintenance cognitive node exists, the actual processing process, actual processing result and applicable characteristic information corresponding to this after-sales work order are associated with the maintenance cognitive node, and a cognitive mapping relationship between the maintenance cognitive node and the corresponding fault fact node is established or updated.

[0209] When there is no corresponding maintenance cognitive node, a new maintenance cognitive node is established based on the fault cause confirmed by this after-sales work order, the actual handling measures adopted, and the actual handling results. The equipment model, software version, and operating environment are used as applicable characteristic information of the new maintenance cognitive node, and a cognitive mapping relationship is established between the new maintenance cognitive node and the corresponding fault fact node.

[0210] When the equipment operation facts corresponding to this after-sales work order include fault fact items that have not yet been recorded in the existing fault fact nodes, the new fault fact items and corresponding fact values ​​will be added to the corresponding fault fact nodes.

[0211] When the current after-sales work order confirms that there is an error in the fault fact item or fact value recorded in the existing fault fact node, the corresponding fault fact node shall be corrected according to the equipment operation facts confirmed in this actual processing.

[0212] When the equipment operation facts corresponding to this after-sales work order cannot jointly represent the same equipment fault with the existing fault fact nodes, a new fault fact node is established based on the equipment operation facts, and a cognitive mapping relationship is established between the new fault fact node and the maintenance cognitive node corresponding to this after-sales work order.

[0213] After updating the maintenance cognition nodes, fault fact nodes, and cognition mapping relationships, the after-sales work orders that have been processed this time are stored as historical after-sales work orders, so that subsequent after-sales work orders to be processed can generate corresponding work order processing strategies based on the updated fault fact map and maintenance cognition map.

[0214] In this embodiment of the invention, by updating the maintenance cognition nodes and cognition mapping relationships according to the actual processing process and actual processing results, and updating the fault fact nodes when new or corrected equipment operation facts occur, the fault fact map and maintenance cognition map can continuously reflect the actual after-sales work order processing status, providing a data foundation for subsequent after-sales work order fact location, maintenance cognition acquisition and work order processing strategy generation.

[0215] The present invention will be described below with reference to a specific example:

[0216] Taking an industrial refrigeration equipment of model A1 experiencing a decrease in cooling capacity as an example, the operation process of the intelligent allocation and processing system for after-sales work orders based on knowledge graphs of this invention will be explained.

[0217] The dual-map construction module acquires multiple historical after-sales work orders that have been processed, and extracts equipment operation facts, maintenance knowledge information, and applicable feature information from the historical after-sales work orders.

[0218] Among these, some historical after-sales work orders record equipment operation facts including decreased cooling capacity, increased compressor operating frequency, and decreased evaporation pressure. The dual-map construction module establishes a fault fact node F1 based on the equipment operation facts.

[0219] Three different maintenance perceptions were formed in the historical after-sales work orders associated with the fault fact node F1. The first maintenance perception node C1 recorded the fault cause as evaporator frosting, the handling measure as performing evaporator defrosting, and the result as the equipment returning to normal after defrosting; its applicable characteristic information includes equipment model A1, software version V1.2, and high humidity operating environment.

[0220] The fault recorded in the second maintenance awareness node C2 is due to abnormal electronic expansion valve control parameters. The solution is to update the electronic expansion valve control parameters. The result is that the opening of the electronic expansion valve returns to normal and the cooling capacity of the equipment is restored after the parameter update. Its applicable characteristic information includes equipment model A1, software version V1.3, and an operating environment with an ambient temperature of 30℃ to 40℃.

[0221] The fault recorded in the third maintenance awareness node C3 is mechanical jamming of the electronic expansion valve. The solution is to replace the electronic expansion valve. The result is that the cooling capacity of the equipment is restored after the electronic expansion valve is replaced. Its applicable characteristic information includes equipment model A1, software version V1.2 to V1.4, and operating environment with an ambient temperature of 25℃ to 40℃.

[0222] The dual-map construction module establishes cognitive mapping relationships between the fault fact node F1 and the first maintenance cognition node C1, the second maintenance cognition node C2, and the third maintenance cognition node C3, respectively, to indicate that the same equipment operation fact corresponds to different maintenance cognitions in the historical processing process.

[0223] The system received a pending after-sales service work order W1. Work order W1 states: Equipment model A1, software version V1.3, operating ambient temperature 36℃; equipment cooling capacity decreased, compressor operating frequency increased from 45Hz to 62Hz, evaporator pressure decreased from 0.42MPa to 0.25MPa; electronic expansion valve control command opening was 45%, actual feedback opening was 15%; fault log recorded electronic expansion valve control deviation error code E37; no frost appeared on the evaporator surface.

[0224] The work order fact location module extracts equipment status and fault logs from the pending after-sales work order W1 and standardizes the equipment operation facts, matching the decreased cooling capacity, increased compressor operating frequency, decreased evaporator pressure, and abnormal electronic expansion valve opening with fault fact nodes in the fault fact graph. After matching, all necessary fault fact items stored in fault fact node F1 match the pending after-sales work order W1, and the matching ratio of fault fact items reaches the preset location threshold. Therefore, fault fact node F1 is identified as the target fault fact node.

[0225] The cognitive conflict identification module obtains the first maintenance cognitive node C1, the second maintenance cognitive node C2, and the third maintenance cognitive node C3 based on the cognitive mapping relationship corresponding to the fault fact node F1.

[0226] Since the three maintenance awareness nodes recorded the causes of the fault as evaporator frosting, abnormal electronic expansion valve control parameters, and mechanical jamming of the electronic expansion valve, it was determined that there was a conflict in the causes of the fault among the three maintenance awareness nodes. Since the three maintenance awareness nodes respectively adopted evaporator defrosting, updating electronic expansion valve control parameters, and replacing electronic expansion valve, it was determined that there was also a conflict in the handling measures.

[0227] The cognitive conflict identification module associates the fault fact node F1, the three maintenance cognitive nodes, the fault cause conflict, the handling measure conflict, and the equipment model, software version and operating environment corresponding to the three maintenance cognitive nodes to generate cognitive conflict events.

[0228] The cognitive verification path generation module establishes a first verification node N1, a second verification node N2, and a third verification node N3 for the first maintenance cognitive node C1, the second maintenance cognitive node C2, and the third maintenance cognitive node C3, respectively, with each verification node corresponding to a maintenance cognitive node.

[0229] For the first verification node N1, three verification conditions are generated based on the conflicts in fault causes, handling measures, and handling results corresponding to the first maintenance cognition node C1. The first verification condition is used to verify whether there is evaporator frosting on the current equipment, and the verification fields include the evaporator surface condition and evaporator temperature; the second verification condition is used to verify whether the defrosting treatment is applicable to the current equipment, and the verification fields include the equipment model, software version, and operating environment; the third verification condition is used to verify the historical processing results of the defrosting treatment, and the verification fields include the historical defrosting process, the equipment status after defrosting, and historical fault recurrence records.

[0230] For the second verification node N2, three verification conditions are also generated. The first verification condition is used to verify whether there is an abnormality in the electronic expansion valve control parameters of the current equipment. The verification fields include the opening degree of the electronic expansion valve control command, the actual feedback opening degree, and the fault code. The second verification condition is used to verify whether the updated electronic expansion valve control parameters are applicable to the current equipment. The verification fields include the equipment model, software version, and operating environment. The third verification condition is used to verify the historical processing results of the updated control parameters. The verification fields include the historical parameter update process, the equipment status after the parameter update, and historical fault recurrence records.

[0231] For the third verification node N3, the first verification condition is used to verify whether there is mechanical jamming of the electronic expansion valve in the current equipment. The verification fields include the opening degree of the electronic expansion valve control command, the actual feedback opening degree, and the valve body action status. The second verification condition is used to verify whether replacing the electronic expansion valve is suitable for the current equipment. The verification fields include the equipment model, software version, and operating environment. The third verification condition is used to verify the historical processing results of replacing the electronic expansion valve. The verification fields include the historical replacement process, the status of the equipment after replacement, and historical fault recurrence records.

[0232] In each verification node, the fault cause verification condition serves as a prerequisite for the handling measure verification condition, and the handling measure verification condition serves as a prerequisite for the handling result verification condition. The cognitive verification path generation module determines the execution order of each verification condition based on the above verification dependencies and generates the cognitive verification path.

[0233] The processing strategy generation module executes the verification conditions in each verification node according to the cognitive verification path. For the first verification node N1, the after-sales work order W1 to be processed records that no frost has appeared on the evaporator surface. Therefore, the verification condition for the fault cause corresponding to evaporator frost does not meet the verification requirements. Since the fault cause verification condition is a necessary verification condition, the first maintenance cognitive node C1 has not met the preset verification pass condition, is not determined as a candidate maintenance cognitive node, and does not participate in the calculation of the strategy evaluation value.

[0234] For the second verification node N2, the electronic expansion valve control command opening is 45%, while the actual feedback opening is 15%, resulting in a 30 percentage point deviation. Furthermore, the fault log records control deviation anomaly code E37. Therefore, the verification conditions for the fault cause corresponding to the abnormal electronic expansion valve control parameters meet the verification requirements. The equipment model corresponding to the pending after-sales work order W1 is A1, the software version is V1.3, and the operating environment temperature is 36℃, all within the applicable characteristic range of the second maintenance cognitive node C2. Therefore, the verification conditions for the processing measures corresponding to updating the electronic expansion valve control parameters meet the verification requirements. Further investigation of the ten historical after-sales work orders associated with the second maintenance cognitive node C2 reveals that all ten historical after-sales work orders fully executed the control parameter update operation. Nine of these work orders returned to normal after processing, and the same fault did not recur within the preset time range. Therefore, the verification conditions for the processing results corresponding to the second maintenance cognitive node C2 meet the verification requirements. All three verification conditions recorded by the second verification node N2 were actually executed and all met the verification requirements. Therefore, the second maintenance cognitive node C2 meets the preset verification pass conditions and is identified as a candidate maintenance cognitive node.

[0235] For the third verification node N3, there is a significant deviation between the control command opening degree and the actual feedback opening degree of the electronic expansion valve, and the equipment detection information shows that the valve body resistance of the electronic expansion valve is greater than the preset resistance threshold. Therefore, the verification conditions for the fault cause corresponding to the mechanical jamming of the electronic expansion valve meet the verification requirements. The equipment model, software version, and operating environment corresponding to the pending after-sales work order W1 are all within the applicable characteristic range of the third maintenance cognition node C3. Therefore, the verification conditions for the handling measures corresponding to replacing the electronic expansion valve meet the verification requirements. Further investigation of the eight historical after-sales work orders associated with the third maintenance cognition node C3 shows that all eight of them performed the electronic expansion valve replacement operation. Five of them remained normal for a long time after the handling, and three of them experienced abnormal electronic expansion valve opening again within the preset time range. Since the historical fault recurrence did not meet the preset handling result verification requirements, the handling result verification conditions corresponding to the third maintenance cognition node C3 do not meet the verification requirements. All three verification conditions recorded by the third verification node N3 were actually executed, and two of them met the verification requirements. The preset verification pass condition is that both the fault cause verification condition and the handling measure verification condition meet the verification requirements, and the proportion of verification conditions that meet the verification requirements is not less than two-thirds. Therefore, the third maintenance cognition node C3 is also determined as a candidate maintenance cognition node.

[0236] Therefore, the second maintenance cognitive node C2 and the third maintenance cognitive node C3 jointly participate in the subsequent strategy evaluation. Based on the number of verification conditions satisfying the verification requirements and the total number of verification conditions in the second verification node N2, the verification evaluation value of the second maintenance cognitive node C2 is calculated to be 1.000; based on the number of verification conditions actually executed and the total number of verification conditions, the path evaluation value of the second maintenance cognitive node C2 is calculated to be 1.000. Based on the number of verification conditions satisfying the verification requirements and the total number of verification conditions in the third verification node N3, the verification evaluation value of the third maintenance cognitive node C3 is calculated to be 0.667; based on the number of verification conditions actually executed and the total number of verification conditions, the path evaluation value of the third maintenance cognitive node C3 is calculated to be 1.000.

[0237] The above evaluation results show that the verification conditions in both verification nodes are actually executed according to the verification dependency relationship, but the fault cause, handling measures and handling results recorded in the second maintenance cognition node C2 have a higher degree of verification matching with the current after-sales work order W1 to be processed.

[0238] The processing strategy generation module further extracts processing resource and processing efficiency data from the historical after-sales work orders associated with the second maintenance cognitive node C2 and the third maintenance cognitive node C3.

[0239] The updated control parameter processing measures corresponding to the second maintenance cognitive node C2 require a type of personnel, namely, control system technicians; require a processing node, namely, a remote technical processing node; do not require replacement of spare parts; require a tool, namely, a parameter configuration tool; its historical average processing time is 40 minutes, and its historical average work order circulation is 1 time.

[0240] The replacement of the electronic expansion valve at the third maintenance knowledge node C3 requires two types of personnel: refrigeration maintenance personnel and electrical testing personnel; it requires two processing nodes: on-site testing node and on-site maintenance node; it requires one electronic expansion valve spare part and two types of tools: refrigeration maintenance tools and electrical testing tools; its historical average processing time is 150 minutes, and its historical average work order circulation is 3 times.

[0241] The resource consumption evaluation value of the second maintenance cognitive node C2 is calculated to be 1.000 and the resource consumption evaluation value of the third maintenance cognitive node C3 is 0 based on the normalization results, according to the number of personnel types, number of processing nodes, number of spare parts types and number of tool types corresponding to the two candidate maintenance cognitive nodes.

[0242] The evaluation results indicate that the types of personnel, processing nodes, spare parts, and tools required to update the control parameters of the electronic expansion valve are generally less than the processing resources required to replace the electronic expansion valve, and its resource consumption is relatively low.

[0243] Based on the historical average processing time and historical average work order turnover of the two candidate maintenance cognitive nodes, the processing time evaluation value of the second maintenance cognitive node C2 is calculated to be 1.000, and the processing time evaluation value of the third maintenance cognitive node C3 is 0.

[0244] The evaluation results indicate that updating the electronic expansion valve control parameters requires less processing time and fewer work order processing steps compared to replacing the electronic expansion valve, thus enabling faster completion of current after-sales work order processing.

[0245] Based on the verification evaluation value of 1.000, path evaluation value of 1.000, resource consumption evaluation value of 1.000, and processing time evaluation value of 1.000 corresponding to the second maintenance cognitive node C2, the strategy evaluation value of the second maintenance cognitive node C2 is calculated to be 1.000.

[0246] Based on the verification evaluation value of 0.667, path evaluation value of 1.000, resource consumption evaluation value of 0, and processing timeliness evaluation value of 0 corresponding to the third maintenance cognitive node C3, the strategy evaluation value of the third maintenance cognitive node C3 is calculated to be approximately 0.417.

[0247] Since the strategy evaluation value of the second maintenance cognitive node C2 (1.000) is greater than the strategy evaluation value of the third maintenance cognitive node C3 (0.417), the second maintenance cognitive node C2 is determined as the target maintenance cognitive node.

[0248] Although the replacement of the electronic expansion valve corresponding to the third maintenance awareness node C3 passed the preset verification conditions and is feasible to resolve the current fault, its historical handling results show that the fault recurred, and it requires more personnel, processing nodes, spare parts and tools, as well as a relatively longer processing time and more work order transfers. The second maintenance awareness node C2 has a better verification matching degree, resource consumption and processing time, and is therefore more suitable as the priority handling solution for the current pending after-sales work orders.

[0249] The processing strategy generation module reads the updated electronic expansion valve control parameter processing measures recorded in the second maintenance cognition node C2 and identifies them as the maintenance method for the pending after-sales work order W1. Based on the historical processing resources corresponding to the second maintenance cognition node C2, it determines that the required personnel type for executing this maintenance method is a control system technician, the processing node is a remote technical processing node, the required tool is a parameter configuration tool, and no equipment spare parts replacement is required. Based on the business flow relationship between the remote technical processing node and the work order acceptance node, it generates a work order flow method from the work order acceptance node to the remote technical processing node; based on the processing timeliness information corresponding to the second maintenance cognition node C2, it determines the processing priority of the pending after-sales work order W1 as high priority.

[0250] The processing strategy generation module will update the maintenance method of the electronic expansion valve control parameters, the work order flow method from the work order acceptance node to the remote technical processing node, and associate them with high priority to generate the work order processing strategy corresponding to the after-sales work order W1 to be processed, and send it to the work order allocation module.

[0251] The work order allocation module extracts the maintenance method, work order flow method, and processing priority from the work order processing strategy. Based on the maintenance method of updating the electronic expansion valve control parameters, it determines the required personnel type as a control system technician and identifies the remote technical processing node as the target processing node. Currently, there are two available control system technicians at the remote technical processing node. The first technician is currently handling three work orders, and the second technician is currently handling one work order. Based on the technician type and current availability, the work order allocation module identifies the second technician as the target processing personnel.

[0252] The work order allocation module generates a work order flow path from the work order acceptance node to the remote technical processing node based on the work order flow method, and controls the pending after-sales work order W1 to enter the remote technical processing node according to the work order flow path. The target processing personnel, according to the work order processing strategy, adjust the electronic expansion valve control gain parameter from the original value of 1.20 to 1.05, and the valve response compensation parameter from the original value of 0.80 to 0.92. After the parameter update, the electronic expansion valve control command opening is 46%, and the actual feedback opening is 44%, with the deviation reduced to 2 percentage points; the compressor operating frequency decreases from 62Hz to 47Hz; the evaporation pressure recovers from 0.25MPa to 0.40MPa; the equipment cooling capacity recovers to the preset range, and fault code E37 is eliminated.

[0253] The work order allocation module records the target personnel, remote technical processing nodes, parameter adjustments, actual processing time (38 minutes), and work order transfer count (1 time) for this work order, forming the actual processing process. Based on the restoration of equipment cooling capacity, evaporator pressure, and fault code elimination, it generates the actual processing result of successful processing. The work order allocation module associates the actual processing process and result with the pending after-sales work order W1 and sends it to the graph update module.

[0254] The graph update module obtains the actual processing process and actual processing results corresponding to the after-sales work order W1 to be processed, and extracts the fault cause confirmed by actual processing, the actual processing measures adopted, and the actual processing results.

[0255] The fault confirmed by actual handling of this work order was that the electronic expansion valve control parameters were abnormal. The actual handling measure was to update the electronic expansion valve control parameters. The actual handling result was that the opening of the electronic expansion valve returned to normal, the cooling capacity of the equipment was restored, and the fault code was eliminated, which is consistent with the maintenance knowledge information recorded in the second maintenance knowledge node C2.

[0256] Therefore, the graph update module retains the cognitive mapping relationship between the fault fact node F1 and the second maintenance cognitive node C2, and associates the parameter adjustment content, target personnel type, remote technical processing node, parameter configuration tool, 38-minute processing time, 1 work order transfer record, and successful processing result corresponding to this work order as new historical processing records with the second maintenance cognitive node C2. The graph update module further associates the equipment model A1, software version V1.3, and ambient temperature 36℃ corresponding to the after-sales work order W1 to be processed with the applicable characteristic information of the second maintenance cognitive node C2.

[0257] Because of the discrepancy between the electronic expansion valve control command opening degree and the actual feedback opening degree in this after-sales work order, and because fault code E37 has not yet been recorded in fault fact node F1, the graph update module supplements the equipment operation facts to fault fact node F1, enabling fault fact node F1 to more completely represent this equipment fault. After completing the graph update, the pending after-sales work order W1 is stored as a historical after-sales work order, allowing the actual processing process, actual processing results, and resource timeliness data to participate in the cognitive verification and strategy evaluation of subsequent after-sales work orders.

[0258] In this embodiment of the invention, by calculating strategy evaluation values ​​for multiple verified maintenance awareness nodes, it is possible to avoid selecting a maintenance solution based solely on a single fault cause, a single historical successful processing record, or a single resource indicator. The strategy evaluation value simultaneously reflects the degree of matching between maintenance awareness and the current equipment fault, the completeness of the verification process, the amount of resources required for the handling measures, and the time and workflow costs required to complete the work order processing. Therefore, it is possible to identify a target maintenance awareness node with a higher overall suitability among multiple feasible maintenance solutions, thereby improving the rationality of work order processing strategy selection.

[0259] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A knowledge graph-based intelligent allocation and processing system for after-sales work orders, characterized in that, The system includes: The dual-graph construction module is used to construct a fault fact graph and a maintenance cognition graph based on the equipment operation facts and maintenance cognition information in historical after-sales work orders, and to establish a one-to-one or one-to-many cognitive mapping relationship between fault fact nodes and maintenance cognition nodes in the two graphs. The work order fact location module is used to determine the corresponding target fault fact node in the fault fact map based on the equipment operation facts in the after-sales work order to be processed. The cognitive conflict identification module is used to obtain the corresponding maintenance cognitive node based on the target fault fact node through cognitive mapping relationship. When there are multiple corresponding maintenance cognitive nodes and there is a conflict among the maintenance cognitive nodes, a cognitive conflict event is generated. The cognitive verification path generation module is used to generate corresponding cognitive verification paths based on cognitive conflict events. The cognitive verification path includes at least one verification node, each verification node corresponds to a maintenance cognitive node, and the verification conditions and verification dependencies of the corresponding maintenance cognitive node are recorded. The processing strategy generation module is used to verify each verification node according to the cognitive verification path, determine candidate maintenance cognitive nodes based on the verification results, and determine the target maintenance cognitive node and the work order processing strategy corresponding to the after-sales work order to be processed based on the strategy evaluation value of the corresponding processing measures of the candidate maintenance cognitive nodes. The work order allocation module is used to determine the target personnel, target processing nodes, and work order flow path according to the work order processing strategy, and to control the after-sales work orders to be processed to complete the work order processing according to the work order flow path and form the processing result.

2. The after-sales work order intelligent allocation and processing system based on knowledge graph as described in claim 1, characterized in that, The maintenance cognition node is used to store maintenance cognition information and applicable characteristic information of the corresponding fault fact node; the maintenance cognition information includes at least the fault cause, handling measures and handling results, and the applicable characteristic information includes at least the equipment model, software version and operating environment.

3. The after-sales work order intelligent allocation and processing system based on knowledge graph as described in claim 2, characterized in that, When there is a conflict in the maintenance cognitive information corresponding to a maintenance cognitive node, a corresponding cognitive conflict event is generated, including: Obtain each maintenance cognition node associated with the target fault fact node, and extract the maintenance cognition information corresponding to each maintenance cognition node; Compare the maintenance knowledge information corresponding to each maintenance knowledge node. When there is inconsistency between the fault cause, handling measures or historical handling results corresponding to at least two maintenance knowledge nodes, determine at least one of the conflict types among fault cause conflict, handling measure conflict and handling result conflict. Extract the applicable feature information corresponding to each maintenance cognitive node where a conflict occurs; The system associates the conflicting maintenance cognitive nodes, the identified conflict types, and the applicable characteristic information corresponding to each maintenance cognitive node to generate cognitive conflict events.

4. The after-sales work order intelligent allocation and processing system based on knowledge graph as described in claim 2, characterized in that, Based on the cognitive conflict event, a corresponding cognitive verification path is generated, including: Extract the maintenance cognitive nodes where the conflict occurred, the conflict type, and the corresponding applicable characteristic information from cognitive conflict events; Verification nodes are established for each maintenance cognition node. The maintenance cognition information corresponding to the conflict type is identified as the content to be verified, and the verification basis is determined according to the conflict type. Among them, the conflict of fault cause corresponds to the equipment status and fault log, the conflict of handling measures corresponds to the equipment model, software version and operating environment, and the conflict of handling result corresponds to the historical handling process, the equipment status after handling and the historical fault recurrence record. When there is a conflict in the cause of the fault, the fault fact item representing the cause of the fault is extracted by combining the fault fact node mapped by the maintenance cognition node, and the corresponding fact value is extracted from the after-sales work order to be processed; when there is a conflict in the handling measures, the applicable feature items with different values ​​are determined from the applicable feature information of each maintenance cognition node, and the corresponding fact value is extracted from the after-sales work order to be processed; when there is a conflict in the handling result, the result fact item representing the handling effect is extracted, and the corresponding historical handling process, post-handling equipment status and historical fault recurrence records are obtained from the historical after-sales work orders. The fault fact item, applicable feature item, or result fact item is determined as the feature item, and the corresponding fact value, historical processing process, post-processing equipment status, or historical fault recurrence record is determined as the verification benchmark. Verification conditions are generated based on the content to be verified, feature item, and verification benchmark of each verification node. Establish verification dependencies between verification nodes based on the data dependencies used by each verification condition, and connect each verification node according to the verification dependencies to generate a cognitive verification path.

5. The after-sales work order intelligent allocation and processing system based on knowledge graph as described in claim 2, characterized in that, Each verification node is verified according to the cognitive verification path. Candidate maintenance cognitive nodes are determined based on the verification results. Then, based on the strategy evaluation value of the corresponding handling measures for the candidate maintenance cognitive nodes, the target maintenance cognitive node and the corresponding work order handling strategy for the pending after-sales work order are determined, including: According to the verification dependencies between verification nodes in the cognitive verification path, the verification conditions corresponding to each verification node are executed sequentially, and the candidate maintenance cognitive nodes that meet the verification conditions are determined based on the verification results of each verification node. Based on the handling measures corresponding to each candidate maintenance awareness node, extract the personnel type, handling node, spare parts type, tool type, handling time, and work order flow record corresponding to the execution handling measures from historical after-sales work orders; The processing resources corresponding to each candidate maintenance cognition node are determined based on the type of personnel, processing node, spare parts type, and tool type. The processing efficiency corresponding to each candidate maintenance cognition node is determined based on the processing time and work order flow record. Based on the verification results, verification dependencies, and corresponding processing resources and efficiency of the verification nodes in the cognitive verification path corresponding to each candidate maintenance cognitive node, calculate the strategy evaluation value corresponding to each candidate maintenance cognitive node. The candidate maintenance knowledge node with the highest strategy evaluation value is determined as the target maintenance knowledge node; Based on the processing measures, resources, and efficiency corresponding to the target maintenance awareness node, a work order processing strategy is generated, including maintenance methods, work order flow methods, and processing priorities.

6. The after-sales work order intelligent allocation and processing system based on knowledge graph as described in claim 4, characterized in that, Based on the content to be verified, applicable features, and corresponding verification benchmarks for each verification node, verification conditions are generated for each verification node, including: Extract the corresponding verification fields based on the content to be verified; Based on the applicable features, filter the key verification fields that correspond to the after-sales work orders to be processed from the verification fields; The device operating facts corresponding to the key verification fields are matched with the verification benchmark to obtain the verification results corresponding to each verification field. Construct a validation expression based on the key validation fields and their corresponding validation benchmarks; Use the validation expression as the validation condition for the corresponding validation node.

7. The after-sales work order intelligent allocation and processing system based on knowledge graph as described in claim 4, characterized in that, Based on the data dependencies between the equipment operating facts used for each verification condition, establish the verification dependencies between each verification node, including: Extract the equipment operation facts corresponding to each verification condition; Establish data reference relationships between actual equipment operation; Data dependencies between verification conditions are formed based on reference relationships; The execution order among verification nodes is determined based on data dependencies; The cognitive verification path is formed by connecting the verification nodes in the execution order.

8. The after-sales work order intelligent allocation and processing system based on knowledge graph as described in claim 5, characterized in that, The process of calculating the strategy evaluation value includes: Based on the verification results, the proportion of verification nodes that meet the verification conditions out of all verification nodes is statistically analyzed to determine the verification evaluation information. Based on the verification dependency statistics, the proportion of verification nodes actually executed in the cognitive verification path out of all verification nodes is used to determine path evaluation information; Resource consumption information is determined based on processing resources, and processing timeliness information is determined based on processing efficiency. Calculate the strategy evaluation value corresponding to each candidate maintenance cognition node based on the verification evaluation information, path evaluation information, resource consumption information, and processing time information.

9. The after-sales work order intelligent allocation and processing system based on knowledge graph as described in claim 5, characterized in that, Based on the processing measures, resources, and efficiency corresponding to the target maintenance awareness node, a work order processing strategy is generated, including maintenance methods, work order flow methods, and processing priorities, including: The maintenance method is determined based on the corresponding handling measures for the target maintenance awareness nodes; Match the corresponding processing resources according to the repair method; Determine the work order processing node based on processing resources; Generate the work order flow path based on the work order processing node; Adjust the processing priority of the work order flow path according to the processing efficiency. By associating maintenance methods, processing resources, work order flow paths, and processing priorities, a work order processing strategy corresponding to the target maintenance awareness node is formed.

10. The after-sales work order intelligent allocation and processing system based on knowledge graph as described in claim 1, characterized in that, The system also includes a graph update module, which is used to update the maintenance cognition nodes and cognition mapping relationships in the maintenance cognition graph based on the actual processing process and the actual processing results of the after-sales work orders to be processed. When it is confirmed that the equipment operation facts have been added or corrected, the fault fact nodes in the fault fact graph are updated simultaneously, so that the updated dual graphs can participate in the generation of work order processing strategies for subsequent after-sales work orders to be processed.

Images