A running state reconstruction method and device, electronic equipment and storage medium

Abstract

The application provides a running state reconstruction method and device, electronic equipment and storage medium. The method comprises: obtaining historical running information and real-time observation information corresponding to a running object, and generating multi-dimensional state representation information; constructing a running state evolution memory structure based on a historical state change process, including a running state chain, and generating a candidate running state set based on the multi-dimensional state representation information and the running state evolution memory structure; screening the candidate running state set based on state constraint information, and determining a target running state that meets a running evolution condition; updating a state node and state evolution association information in the running state evolution memory structure according to state evolution information corresponding to the target running state, forming an updated running state chain, and maintaining running state evolution continuity. The application improves the relevance and credibility of state determination in the running state reconstruction process based on the association between the historical running process of the running object and the real-time observation information.

Description

Technical Field

[0001] This application relates to the field of operational status analysis technology, and in particular to an operational status reconstruction method, apparatus, electronic device and storage medium. Background Technology

[0002] With the continuous development of industrial equipment, intelligent systems, and complex operating platforms, operating objects are typically in dynamically changing operating environments, and their operation is affected by various factors such as their own operating conditions, changes in the external environment, and changes in related factors. In the above application scenarios, it is necessary to analyze the current operating status of the operating objects to support operations such as operation process monitoring, anomaly analysis, and subsequent operation management.

[0003] In existing technologies, a common approach to operational status analysis is to extract features from the state parameters of the operational object based on real-time collected operational data, and then match the extracted state features with preset state conditions to determine the current operational status of the operational object based on the matching results. This method typically uses currently collected operational information to establish a correspondence between currently observed data and operational status, thereby achieving the identification of the operational object's status.

[0004] However, in actual operation, the current state changes of the operating object usually exhibit a certain degree of continuity. When relying solely on currently collected operational information for state analysis, the real-time observation information may not fully reflect the overall operational process of the operating object. This makes it difficult for the operational state determined based on the current observation information to adequately represent the state change relationships of the operating object during continuous operation, resulting in insufficient correlation between the operational state analysis and the actual operational process. Furthermore, when real-time observation information is incomplete, there are differences between state characteristics, or multiple operational states are all reasonable to a certain extent, existing technologies typically lack a technical mechanism to express the reliability of the state determination results. It is difficult to distinguish the reliability of different state determination results and to provide an evaluation basis for the continuous updating and maintenance of the operational state. Summary of the Invention

[0005] This application provides a method, apparatus, electronic device, and storage medium for reconstructing operational status, which addresses the problem in the prior art that when performing status analysis based on currently collected operational information, the real-time operational information may not fully reflect the continuous operation process of the operating object, resulting in insufficient correlation between the determined operational status and the operational process, and difficulty in expressing the credibility of the operational status reconstruction results.

[0006] To address the aforementioned technical problems, this application provides a method for reconstructing operational states, comprising: The historical operation information and real-time observation information corresponding to the running object are obtained, and the state feature is parsed based on the real-time observation information to generate multi-dimensional state representation information describing the current operation characteristics of the running object. An operational state evolution memory structure is constructed based on the historical state change process in the historical operational information. The operational state evolution memory structure includes an operational state chain formed by state nodes and state evolution association information. A candidate operational state set is generated based on the multidimensional state representation information and the operational state evolution memory structure. The candidate operating state set is filtered based on state constraint information, and the target operating state that meets the operating evolution conditions is determined based on state evaluation information. The state evaluation information includes state confidence, which characterizes the credibility of the candidate operating state as the target operating state. The state nodes and state evolution association information in the running state evolution memory structure are updated according to the state evolution information corresponding to the target running state to form an updated running state chain, so as to maintain the continuity of running state evolution.

[0007] Optionally, the step of performing state feature parsing based on the real-time observation information to generate the multi-dimensional state representation information includes: The real-time observation information is parsed to determine the corresponding observation information category; State features are extracted from the real-time observation information according to the observation information category to obtain multiple state features of different dimensions, and a state feature set is formed based on the state features. Perform correlation analysis on the state features in the state feature set to determine the correlation relationships between the state features; Based on the aforementioned correlation, multiple state features are fused to generate the multidimensional state representation information.

[0008] Optionally, the association analysis of the state features in the state feature set includes: Obtain the feature attribute information corresponding to different state features in the state feature set; The association relationships between different state features are determined based on the aforementioned feature attribute information; Based on the aforementioned relationship, feature association information is generated to describe the degree of association between state features, and state feature fusion processing is performed based on the feature association information.

[0009] Optionally, constructing the operational state evolution memory structure based on the historical state change process in the historical operational information includes: Multiple historical state information is extracted from the historical operation information, and the state change process corresponding to the multiple historical state information is analyzed; Based on the temporal correlation, state change relationship, and state change path among multiple historical state information, historical state association information is generated; Based on the historical state association information, determine the state nodes corresponding to different historical states; State evolution association information is established based on the state change relationships between the state nodes, forming the running state evolution memory structure.

[0010] Optionally, the generation of historical state association information includes: The temporal correlation between historical states is determined based on the occurrence time of the states corresponding to multiple historical state information. The relationship between states is determined based on the direction and degree of state change between adjacent historical states; Based on the time correlation and the state change relationship, the evolutionary correlation characteristics between historical states are determined, and the historical state correlation information is generated.

[0011] Optionally, establishing state evolution association information includes: Determine the direction of state change between state nodes based on the historical state characteristics corresponding to different state nodes; The state transition relationship between state nodes is determined based on the direction of state change. Based on the state transition relationship, state evolution association information is generated to represent the state evolution path between different state nodes.

[0012] Optionally, generating a candidate running state set based on the multidimensional state representation information and the running state evolution memory structure includes: Extract the state change features corresponding to the current operating state based on the multidimensional state representation information; Determine state matching features based on the state change characteristics; Based on the state matching features, associated state nodes are determined in the running state evolution memory structure; A candidate running state set is generated based on the state evolution association information corresponding to the associated state nodes.

[0013] Optionally, determining the associated state node includes: Obtain the node state features corresponding to each state node in the running state evolution memory structure; Candidate associated nodes are determined based on the matching relationship between the node state features and the state matching features; The associated state node is determined based on the continuity of the state evolution corresponding to the candidate associated node.

[0014] Optionally, generating the candidate running state set includes: Multiple candidate state change processes are determined based on the state evolution association information corresponding to the associated state nodes; Predict the corresponding state change result based on the candidate state change process; The multiple states change results are determined as the candidate running states set.

[0015] Optionally, the step of filtering the candidate operating state set based on state constraint information and determining the target operating state based on state evaluation information includes: Obtain the state change process corresponding to each candidate running state in the candidate running state set; Based on the state constraint information, a constraint analysis is performed on the state change process to determine the constraint satisfaction result corresponding to the candidate operating state. Based on the constraint satisfaction results and the matching relationship between the candidate running state and the multidimensional state representation information, a state confidence level is determined to characterize the degree to which the candidate running state satisfies the state constraints and matches the current running state, and state evaluation information is generated based on the state confidence level. The target operating state that meets the operating evolution conditions is determined based on the state evaluation information.

[0016] Optionally, the state constraint information includes at least one of the following: Time constraint information used to describe the temporal continuity of state changes; State transition constraint information used to describe the rationality of the state transition process; Association constraint information used to describe the state coordination relationship between different associated objects.

[0017] Optionally, the generation of state evaluation information includes: The degree of constraint satisfaction corresponding to the candidate running state is determined based on the state constraint results. The degree of state matching is determined based on the correlation between the multidimensional state representation information and the state evolution process corresponding to the candidate operating state; The state evaluation information is generated based on the degree of constraint satisfaction and the degree of state matching.

[0018] Optionally, updating the state nodes and state evolution association information in the running state evolution memory structure includes: The new state node is determined based on the state evolution information corresponding to the target operating state; The evolution association information of the new state is determined based on the state change relationship between the new state node and the existing state node; The running state evolution memory structure is updated based on the newly added state nodes and the newly added state evolution association information to form an updated running state chain.

[0019] Optionally, determining the newly added state evolution association information includes: Obtain the state change characteristics between the target running state and existing state nodes; Determine the direction of state change and the state transition relationship based on the aforementioned state change characteristics; The newly added state evolution association information is generated based on the state change direction and state transition relationship.

[0020] Optionally, when the real-time observation information cannot fully reflect the current state of the running object, it includes: Based on the operational state evolution memory structure, historical state nodes associated with the current state representation information are determined; The continuous state evolution process is determined based on the state evolution association information corresponding to the historical state nodes; Based on the continuous state evolution process, the target operating state is determined in combination with the state confidence level to maintain the continuity of the operating state evolution.

[0021] This application also provides an operating state reconfiguration device, comprising: The state representation construction module is used to obtain historical operation information and real-time observation information corresponding to the running object, and perform state feature parsing based on the real-time observation information to generate multi-dimensional state representation information describing the current operation characteristics of the running object; The state evolution memory construction module is used to construct an operating state evolution memory structure based on the historical state change process in the historical operating information. The operating state evolution memory structure includes an operating state chain formed by state nodes and state evolution association information, and generates a candidate operating state set based on the multi-dimensional state representation information and the operating state evolution memory structure. The state determination module is used to filter the candidate operating state set based on state constraint information, and determine the target operating state that meets the operating evolution conditions based on state evaluation information. The state evaluation information includes state confidence, which characterizes the credibility of the candidate operating state as the target operating state. The state chain update module is used to update the state nodes and state evolution association information in the running state evolution memory structure according to the state evolution information corresponding to the target running state, so as to form an updated running state chain and maintain the continuity of running state evolution.

[0022] This application also provides an electronic device, the electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the above-described operation state reconstruction method.

[0023] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described running state reconstruction method.

[0024] Compared with the existing technology that directly performs state analysis based on current operating data, this application obtains historical operating information and real-time observation information corresponding to the operating object, and generates multi-dimensional state representation information based on the real-time observation information, so that the operating state analysis process can simultaneously associate current observation information and historical operating processes.

[0025] Furthermore, this application constructs an operational state evolution memory structure based on the historical state change process in historical operational information, and generates a candidate operational state set based on the correlation between multidimensional state representation information and the operational state evolution memory structure. This enables the operational state determination process to be analyzed in conjunction with the historical state change relationship of the operational object. Compared with the method of determining the state based solely on current observation information, this improves the correlation between the state determination process and the operational process.

[0026] Furthermore, this application performs constraint screening on the state change process corresponding to the candidate operating state based on state constraint information, and determines the target operating state based on state evaluation information, so that the determination process of the target operating state can be evaluated in conjunction with the state change process, reducing the possibility of inconsistency between the state determination and the actual operating process due to a single state matching.

[0027] Furthermore, this application updates the state nodes and state evolution association information in the running state evolution memory structure according to the state change information corresponding to the target running state, forming an updated running state chain, and maintains the continuity of running state evolution based on the updated running state chain, so that the running state reconstruction process can be adjusted as the running process changes.

[0028] Furthermore, this application characterizes the evaluation results of candidate operating states through state evaluation information and characterizes the credibility of the target operating state as the result of operating state reconstruction through state confidence information. This enables the system to express the credibility of the state reconstruction result when multiple candidate operating states have a certain degree of rationality or when real-time observation information is incomplete, thus providing an evaluation basis for determining the target operating state, updating the state chain, and maintaining the continuity of operating state evolution. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the runtime state reconstruction method according to an embodiment of this application; Figure 2 This is a schematic diagram of the overall processing architecture for reconstructing the running state in an embodiment of this application; Figure 3 This is a schematic diagram illustrating the process of acquiring historical running information and real-time observation information of the running object in an embodiment of this application; Figure 4 This is a schematic diagram illustrating the process of generating multidimensional state representation information based on real-time observation information in an embodiment of this application; Figure 5 This is a schematic diagram illustrating the process of parsing historical state changes based on historical operational information, according to an embodiment of this application. Figure 6 This is a schematic diagram illustrating the process of constructing a runtime state evolution memory structure based on historical state change processes in an embodiment of this application. Figure 7 A schematic diagram illustrating the structure of the running state chain formed by the running state nodes and state evolution association information in this application embodiment; Figure 8 This is a schematic diagram illustrating the process of generating a candidate running state set based on multidimensional state representation information and running state evolution memory structure in an embodiment of this application. Figure 9 This is a schematic diagram illustrating the process of filtering candidate operating states based on state constraint information according to an embodiment of this application; Figure 10 This is a schematic diagram illustrating the process of determining the target operating state based on state evaluation information in an embodiment of this application; Figure 11 This is a schematic diagram illustrating the process of updating the running state evolution memory structure according to the target running state in an embodiment of this application; Figure 12 This is a schematic diagram illustrating the process of maintaining the continuity of the evolution of the running state based on the updated running state chain, as described in an embodiment of this application.

[0030] Figure 13 This is a schematic diagram of the operating state reconstruction device according to an embodiment of this application; Figure 14 This is a schematic diagram of the electronic device structure according to an embodiment of this application; Figure 15 This is a schematic diagram of a readable storage medium structure according to an embodiment of this application. Detailed Implementation

[0031] The following detailed description, in conjunction with the accompanying drawings, illustrates a method, apparatus, electronic device, and storage medium for reconstructing operational states according to embodiments of this application. It should be noted that the following embodiments are merely illustrative of the technical solutions of this application and are not intended to limit the scope of protection of this application. Other embodiments obtained by those skilled in the art based on the disclosed technical content of this application without inventive effort should also fall within the scope of protection of this application.

[0032] Figure 1 This is a schematic diagram of a runtime state reconstruction method provided in an embodiment of this application. Figure 2 This is a schematic diagram of the overall architecture for runtime state reconstruction provided in an embodiment of this application. Figure 1 and Figure 2 As shown, the operational state reconstruction method provided in this embodiment can be applied to application scenarios involving state analysis, state prediction, and continuous state updates of operational objects with dynamic changes. The operational objects may include industrial equipment objects, production process objects, intelligent aquaculture objects, and other objects with continuous state change characteristics.

[0033] In actual operation, the state of an object is usually not directly determined by data at a single point in time, but rather gradually formed under the combined influence of historical operating processes, current operating conditions, and external environmental factors. For example, for industrial equipment, whether the equipment is currently in normal operating condition depends not only on the currently collected operating parameters such as temperature, vibration, and load, but also on the load change trends, abnormal development processes, and maintenance status during the equipment's previous operation. For intelligent aquaculture objects, the current health status is not only related to current vital signs data, but also influenced by historical growth stages, behavioral changes, and changes in the aquaculture environment. Therefore, if state identification is based solely on data collected at the current moment, it is easy to encounter situations where current behaviors are similar but actual evolutionary stages differ, resulting in state judgment results that cannot accurately reflect the true state of the object.

[0034] Preferably, such as Figure 2As shown, this embodiment addresses the issue of historical evolution characteristics in the state of a running object by constructing a running state reconstruction architecture that combines real-time state representation and historical state evolution memory. This architecture does not directly establish a one-to-one mapping between real-time observation information and the target state. Instead, it continuously evaluates candidate running states using historical running information, current state representation information, and state constraint information to determine the current running state. First, multi-dimensional state representation information describing the current running characteristics is generated using real-time observation information. Simultaneously, a running state evolution memory structure representing the state change pattern is constructed based on historical running information. Further, a set of candidate running states is generated based on the correlation between the current state representation information and the historical state evolution pattern. Constraint analysis and evaluation of the candidate states are then performed in conjunction with the current actual running conditions of the running object to determine the target running state. Finally, the running state evolution memory structure is updated based on the determined target running state, allowing the historical state evolution pattern to be continuously improved as the running object continues to operate.

[0035] Through the above architecture, this embodiment combines the independent real-time state judgment process and historical data analysis process in traditional state identification, transforming historical operational data from merely a basis for querying or statistics into structured state evolution information that can participate in the reconstruction of the current state. Compared to methods that rely solely on current data for state classification, this embodiment can fully utilize the state change patterns formed during the long-term operation of the running object, improving the accuracy and stability of state reconstruction results in complex operating environments.

[0036] Preferably, in this embodiment, historical operation information and real-time observation information can be acquired based on the actual operational characteristics of different operating objects. Historical operation information describes the state changes of the operating object during its past operation and may include historical operation parameters, historical state records, historical anomaly records, historical environmental change information, and other information reflecting the historical state of the operating object. Real-time observation information describes the state performance of the operating object in its current operational phase and may include current operation parameters, current environmental information, current behavior information, and other information collected in real time. By simultaneously acquiring historical operation information and real-time observation information, the state reconstruction process can simultaneously possess the ability to analyze temporal continuity and the ability to perceive the current state.

[0037] For example, in industrial equipment operation scenarios, sensors can collect information on temperature, vibration, current, pressure, and load during equipment operation, and combine this data with historical operation logs, maintenance records, and fault records to form historical operation information. In smart animal husbandry scenarios, image acquisition devices, environmental monitoring devices, and physiological state monitoring devices can acquire information on animal behavior, environment, and vital signs, and combine this data with historical growth records, health records, and environmental change records to form historical operation information. The fusion of information from different sources provides a more complete data foundation for subsequent state characterization construction and state evolution analysis.

[0038] S1. Obtain historical operation information and real-time observation information corresponding to the running object, and perform state feature parsing based on the real-time observation information to generate multi-dimensional state representation information describing the current operation characteristics of the running object.

[0039] Figure 3 This is a schematic diagram illustrating the process of obtaining runtime object information provided in an embodiment of this application. Figure 4 This is a schematic diagram illustrating the process of generating multidimensional state representation information provided in an embodiment of this application.

[0040] Preferably, such as Figure 3 As shown, during execution S1, this embodiment first acquires the historical operation information and real-time observation information corresponding to the running object, and then processes them uniformly according to the data characteristics of different information sources. Since operation information in real-world applications typically originates from multiple acquisition devices or multiple information systems, information from different sources may have different data formats, sampling times, and state meanings. Directly fusing the raw data could lead to some state information not being effectively correlated. Therefore, this embodiment parses and organizes the acquired multi-source operation information, enabling information from different sources to be processed within a unified state analysis framework.

[0041] Specifically, real-time observation information can be classified and analyzed based on its data source, acquisition time, data change patterns, and corresponding state meanings, and state features reflecting changes in the state of the operating object can be extracted. Through this processing, the raw data can be converted into information expressions with state semantics, enabling subsequent state analysis processes to no longer directly rely on the original parameters but to perform reasoning based on feature information with clear state meanings.

[0042] Preferably, after completing the parsing and processing of real-time observation information, this embodiment further extracts state features that reflect the current state of the running object based on the parsing results, and generates multi-dimensional state representation information based on the correlation between multiple state features. For example... Figure 4As shown, the state representation generation process in this embodiment is not simply a matter of splicing data from different sources. Instead, it addresses the issue that the state of a running object is usually influenced by multiple factors by comprehensively analyzing the feature information corresponding to different state factors, so that the generated state representation information can more comprehensively describe the current state of the running object.

[0043] Specifically, when extracting state features, feature information reflecting changes in the state of the operating object can be extracted from real-time observation information based on the actual operational characteristics of the object. For example, in industrial equipment operation scenarios, features such as equipment operating status, load change, abnormal change, and environmental impact can be extracted based on changes in equipment operating parameters; in smart farming scenarios, features such as activity change, health change, and environmental adaptation can be extracted based on collected behavioral, physiological, and environmental information. By converting raw observation data into state features, the expression differences between different data sources can be reduced, enabling subsequent state analysis processes to be based on information with clear state meanings.

[0044] Preferably, since the actual state of a running object is usually not determined by a single state factor, but rather by the combined effect of multiple state factors, this embodiment, after obtaining multiple state features, further analyzes the correlation between different state features and generates multidimensional state representation information based on the correlation. Specifically, state feature correlation information can be determined based on the time change relationship, change trend relationship, and mutual influence relationship between different state features, so that different state features form a correlation expression with state meaning.

[0045] For example, during the operation of industrial equipment, increased load may lead to temperature rise, and changes in vibration may indicate changes in the internal structure of the equipment. Analyzing only load, temperature, and vibration parameters separately cannot accurately reflect the process of equipment state changes. In intelligent animal husbandry, changes in ambient temperature may affect animal activity, and changes in activity may reflect trends in animal health. Analyzing only a single behavioral indicator is also insufficient to accurately determine the animal's current state. Therefore, by establishing correlations between state characteristics, multidimensional state representation information can not only reflect current state parameters but also demonstrate the interrelationships between different state factors.

[0046] Preferably, when generating multidimensional state representation information, this embodiment can also adjust the association method between state features according to the state formation rules corresponding to different operating objects, so as to improve the adaptability of the state representation process to different application scenarios. Since the state change mechanisms of different operating objects are different, for example, industrial equipment usually exhibits a process of changing operating parameters and performance degradation, while intelligent aquaculture objects usually exhibit a process of coordinated changes in behavior, physiology, and environment. Therefore, if a fixed state expression method is used, it may not be able to fully reflect the state characteristics of different objects.

[0047] To address the aforementioned issues, this embodiment selects an appropriate state feature organization method based on the state change patterns of the operating object, enabling multi-dimensional state representation information to form state descriptions that better reflect the actual operating process for different operating objects. For example, for industrial equipment, the focus can be on the correlation between changes in operating parameters and the formation process of abnormal states; for intelligent aquaculture objects, the focus can be on the correlation between behavioral changes, physiological changes, and environmental changes. This approach improves the adaptability of state representation information to complex operating scenarios, providing a more accurate basis for subsequent state reconstruction based on historical state evolution patterns.

[0048] Preferably, after generating the multidimensional state representation information, this embodiment can further structure and organize the state representation information so that it can be associated with the subsequently constructed runtime state evolution memory structure. Since the subsequent state reconstruction process needs to utilize both current state information and historical state change patterns, if the current state information exists only as independent parameters, it is difficult to establish an effective connection with historical state nodes. Therefore, this embodiment unifies the expression of current state features, feature association information, and state change information, enabling the generated multidimensional state representation information to serve as the connection basis between the current state and the historical state evolution process. Through the above processing, this embodiment achieves the transformation from raw real-time observation information to multidimensional state representation information with state semantics, allowing the runtime state reconstruction process to simultaneously utilize the current state performance and the correlation between state factors, improving the accuracy of subsequent state matching, candidate state generation, and target state determination processes.

[0049] S2, construct an operation state evolution memory structure based on the historical state change process in the historical operation information. The operation state evolution memory structure includes an operation state chain formed by state nodes and state evolution association information, and generate a candidate operation state set based on the multi-dimensional state representation information and the operation state evolution memory structure.

[0050] Figure 5 This is a schematic diagram illustrating the historical state change analysis process provided in an embodiment of this application. Figure 6This is a schematic diagram illustrating the process of constructing the runtime state evolution memory structure provided in an embodiment of this application. Figure 7 This is a schematic diagram of the state chain generation process provided in an embodiment of this application.

[0051] Preferably, such as Figure 5 and Figure 6 As shown, after obtaining the historical operation information corresponding to the running object, this embodiment does not directly store the historical operation information as independent data. Instead, it further analyzes the state changes at different time stages during the historical operation to extract historical state information that reflects the evolutionary pattern of the running object's state. Since the current state of the running object is usually the result of continuous changes across multiple historical stages, if only the historical state results are saved without recording the relationships between different states, the historical operation process cannot be used to assist in judging the current state. Therefore, this embodiment determines the historical states corresponding to different time stages based on the historical operation information and analyzes the direction, degree, and state transition relationships between different historical states to establish structured information that can describe the state change process of the running object. The historical state information represents the state performance of the running object at different stages, and the relationships between the states represent the evolutionary pattern of the running object changing from one state to another. Through this method, historical operation data that traditionally exists in time series form can be converted into evolutionary information with state semantics, providing a data foundation for subsequent state reconstruction based on historical patterns.

[0052] Preferably, such as Figure 6 As shown, after determining the historical state information, this embodiment further constructs an operational state evolution memory structure based on the correlation between historical states. Specifically, this embodiment uses different historical states as state nodes and establishes state evolution correlation information based on the changing relationships between state nodes. This allows the operational state evolution memory structure to not only record which states the operating object has been in, but also to represent how different states change and the possible state stages after the change. Compared with the prior art, which only uses historical data to statistically analyze the state range or determine whether the current data deviates from the historical range, the operational state evolution memory structure constructed in this embodiment can retain the continuous information in the process of state change of the operating object, transforming historical operational information from static data recording into a dynamic evolution structure that can participate in state reasoning. For example, in the operation of industrial equipment, when the equipment changes from a normal operating state to an abnormal operating state, it usually undergoes a process of performance change, parameter deviation, and gradual formation of abnormal characteristics. The state evolution correlation information can record the above-mentioned change process. In intelligent breeding, when animals change from a normal state to a healthy abnormal state, they also exhibit continuous changes in behavior, physiology, and environmental adaptation. The state evolution correlation information can describe the above-mentioned change process.

[0053] Preferably, such as Figure 7 As shown, after constructing state nodes and state evolution association information, this embodiment further organizes and forms an operational state chain based on the degree of association between state nodes, enabling multiple related state nodes to form a directional state change path according to the actual operation process. The operational state chain represents the state evolution process of the operating object between different operational stages, allowing the subsequent state reconstruction process to match the corresponding historical evolution path based on the current state performance, rather than making static judgments based solely on the state parameters at the current time point. Compared to traditional methods based on historical data statistical analysis, this embodiment, by constructing a state chain, enables the system not only to acquire historical state information but also to predict possible development directions using the evolutionary patterns between historical states. For example, in the operation of industrial equipment, a state chain can represent the continuous change process of the equipment from a normal operating state, a performance change state, to an abnormal state; in intelligent aquaculture, a state chain can represent the changes in behavioral and physiological characteristics of the aquatic object as it changes from a normal state to an abnormal state. Furthermore, during the continuous operation of the running object, when new running state information is obtained, the state nodes and state evolution association information can be updated according to the newly added state information, so that the running state chain can be continuously supplemented with new state change paths, thereby improving the adaptability of the running state evolution memory structure to the long-term running process.

[0054] S3, based on state constraint information, filter the set of candidate operating states, and determine the target operating state that meets the operating evolution conditions according to state evaluation information. The state evaluation information includes state confidence, which characterizes the credibility of the candidate operating state as the target operating state.

[0055] Figure 8 This is a schematic diagram of the candidate running state generation process provided in an embodiment of this application.

[0056] Preferably, such as Figure 8As shown, after generating multi-dimensional state representation information and constructing the running state evolution memory structure, this embodiment further combines the multi-dimensional state representation information corresponding to the current running state with historical state evolution patterns to generate a candidate running state set. Unlike existing technologies that directly output a single state result based on currently collected data, this embodiment considers that the running object may have multiple state evolution possibilities during actual operation. Even if the current state is the same, different historical running processes may correspond to different development directions. Therefore, this embodiment does not directly map the current state representation information to a unique running state. Instead, it first utilizes the correlation between the current state representation information and the running state evolution memory structure to determine multiple possible state evolution paths, and generates a candidate running state set based on these paths. This candidate running state set is used to subsequently combine state constraint information and state evaluation information to determine the current running state. By retaining multiple possible state outcomes, the problem of biased state judgments due to insufficient current observation information or similar state behavior can be avoided, allowing for further selection of target states that better reflect the actual operation process by combining operational constraints.

[0057] Preferably, such as Figure 8 As shown, when generating a candidate set of operating states, this embodiment first matches the state features in the multidimensional state representation information with the historical state nodes in the operating state evolution memory structure to determine the associated state nodes that have an evolutionary relationship with the current operating state. The matching process does not solely rely on the parameter similarity between the current state features and historical state nodes; it further combines the trend of state feature changes and the historical state evolution relationship to determine the degree of association. Since the operating object may exhibit similar state behaviors at different evolutionary stages—for example, industrial equipment may experience temperature increases during normal load changes, and similar temperature changes may occur during abnormal equipment evolution—matching based on only a single parameter could easily lead to misidentifying different evolutionary stages as the same state. Therefore, this embodiment determines associated state nodes by combining state feature matching and historical evolution relationships, enabling the current state to match a historical evolution path that more closely reflects the actual change process.

[0058] Preferably, after determining the associated state nodes, this embodiment further generates a candidate running state set based on the state evolution association information corresponding to the associated state nodes. For example... Figure 8As shown, the candidate running state generation process can determine multiple possible state outcomes based on the subsequent state change paths of associated state nodes, and use these multiple state outcomes as candidate running states. These candidate running states are not simply predicted based on current data, but are deduced based on the historical state change patterns of the running object, thus reflecting the state development trend during the actual operation of the running object.

[0059] For example, in industrial equipment operation scenarios, when the current operating parameters of the equipment exhibit certain abnormal trends, this embodiment can determine whether the equipment is in a performance change stage, an anomaly formation stage, or a stable recovery stage based on historical equipment state evolution paths, and use these possible states as candidate operating states. In intelligent farming scenarios, when animals exhibit changes in activity or feeding, this embodiment can determine whether they are in a normal environmental adaptation state, a staged change state, or a health abnormality state based on historical behavior change paths. Through the above methods, the system can obtain possible state results from multiple historical evolution paths, providing a foundation for subsequent state constraint screening.

[0060] Preferably, to improve the matching degree between the candidate running state set and the actual running process, this embodiment further considers the continuity of state evolution when generating candidate running states. Since the state changes of a running object usually have a certain continuous pattern, the current state usually does not unconditionally jump to any historical state, but is constrained by the evolutionary relationship between previous and subsequent states. Therefore, in determining candidate running states, this embodiment judges whether the candidate state conforms to the historical evolutionary pattern based on the state transition relationship between associated state nodes, and prioritizes retaining state results with continuous evolutionary relationships. Through the above processing, the candidate running state set generated in this embodiment combines the current state performance and historical evolutionary patterns, transforming the state reconstruction process from the traditional "current data matching state" to "current state combined with historical evolutionary path reasoning state". This method can more accurately reflect the actual state change process of the running object and provide a reliable foundation for subsequently determining the target running state based on state constraint information and state evaluation information.

[0061] Figure 9 This is a schematic diagram of the candidate running state constraint screening process provided in an embodiment of this application. Figure 10 This is a schematic diagram illustrating the process of determining the target operating state provided in an embodiment of this application.

[0062] Preferably, such as Figure 9As shown, after generating the candidate running state set, this embodiment further performs constraint analysis on the candidate running states based on the current running conditions and state change patterns of the running object to filter candidate states that conform to the actual running process. Since the candidate running states are possible state results derived from historical state evolution patterns, and the actual running process may also be affected by current environmental conditions, real-time running conditions, and changes in the state of related objects, not all candidate states obtained based on historical paths are valid at the current moment. Directly selecting a state as the target state from the candidate running state set may lead to inconsistencies between the determined result and the current actual running situation. Therefore, this embodiment introduces state constraint information to perform a rationality analysis on the state change process corresponding to the candidate running states, determining whether the candidate states meet the current state change conditions of the running object. In this way, the "possible states" provided by historical evolution patterns can be further converted into "effective states" that conform to the current running conditions, improving the accuracy of the state reconstruction results.

[0063] Preferably, in performing state constraint analysis, this embodiment does not merely determine whether the state parameters corresponding to the candidate operating state meet the preset range, but rather makes a comprehensive judgment based on the correlations in the state evolution process of the operating object. The state constraint information can be used to describe the temporal continuity, state transition, and state correlation relationships that need to be satisfied during state changes. For example, in the operation of industrial equipment, the change from a normal state to an abnormal state usually requires a parameter change process within a certain time range. If the current operating conditions of the equipment do not conform to this change pattern, the corresponding abnormal state should not be considered a priority candidate result. In intelligent farming scenarios, changes in animal behavior may be affected by environmental changes. If the current environmental changes can explain the behavioral changes, the behavioral changes should not be directly judged as health abnormalities. Through the above constraint analysis method, this embodiment avoids the problem of mechanical inference based solely on historical state paths, ensuring that candidate operating states not only meet historical evolution patterns but also conform to the current actual operating environment.

[0064] Preferably, such as Figure 10As shown, after completing the state constraint analysis, this embodiment further determines the target operating state based on the evaluation information corresponding to the candidate operating states. Specifically, this embodiment does not simply screen candidates based on whether they meet the constraints, but further comprehensively considers the degree of matching between the candidate operating state and the current multi-dimensional state representation information, the continuity of the historical evolution path corresponding to the candidate state, and the degree to which the candidate state meets the state constraints, generating corresponding state evaluation information, and determining the target operating state based on the state evaluation results. The state evaluation information includes state confidence information, which characterizes the credibility of the target operating state as the result of operating state reconstruction. Since multiple states may simultaneously meet some conditions during the actual operation of the object, for example, industrial equipment may simultaneously exhibit normal load changes and abnormal trend characteristics, and intelligent aquaculture objects may simultaneously exhibit environmental adaptation behavior and health change behavior, using only a single judgment condition may not accurately distinguish different candidate states. This embodiment comprehensively evaluates candidate operating states from multiple dimensions by integrating multiple evaluation factors, so that the finally determined target operating state can better conform to the actual change process of the object.

[0065] Preferably, when generating state evaluation information, this embodiment can determine the candidate state evaluation result based on the degree of state constraint satisfaction, the degree of state feature matching, and the degree of state evolution continuity corresponding to the candidate operating state. Specifically, the degree of state constraint satisfaction indicates whether the candidate state meets the current operating conditions, the degree of state feature matching indicates the consistency between the candidate state and the current real-time observation information, and the degree of state evolution continuity indicates whether the candidate state conforms to the historical state change path.

[0066] In one implementation, when determining the state evaluation information corresponding to a candidate operating state, the candidate operating state can be comprehensively evaluated based on the degree to which the state constraints are satisfied, the degree of matching with multidimensional state representation information, and the rationality of the corresponding state evolution process, to generate the state evaluation information. Further, a state confidence level can be determined from the state evaluation information to characterize the reliability of the candidate operating state as a result of operating state reconstruction. The state confidence level reflects the degree to which the candidate operating state satisfies state constraints and conforms to the characteristics of the current operating state. When the state evaluation information indicates a high degree of consistency between the candidate operating state and the historical operating process, the current state representation, and the state constraints, the corresponding state confidence level is high; when there is a significant deviation between the candidate operating state and the above information, the corresponding state confidence level is low. The state confidence level can be represented numerically, in terms of levels, intervals, vectors, or other forms that can characterize the degree of reliability, to distinguish the reliability of reconstruction results from different candidate operating states.

[0067] For example, for industrial equipment, a high evaluation result can be obtained when a candidate state matches the current temperature, vibration, and load change trends, and this state typically evolves from the current state stage during historical operation, while also meeting the current operating conditions of the equipment. For intelligent aquaculture, a more credible target state can be considered when a candidate state simultaneously conforms to the animal's current behavior, historical health change patterns, and current environmental conditions. Through the above-described state constraint screening and state evaluation process, this embodiment realizes a step-by-step determination process from a set of candidate operating states to a target operating state. The candidate operating state set is used to retain multiple possible results based on historical evolution patterns; state constraint information is used to exclude state results that do not meet the current operating conditions; and state evaluation information is used to determine the target state that best matches the current operating process from the state results that meet the conditions. This multi-stage state determination mechanism improves the reliability of the operating state reconstruction process in complex operating environments.

[0068] Figure 11 This is a schematic diagram of the runtime state chain update process provided in an embodiment of this application. Figure 12 This is a schematic diagram of the continuous reconstruction process of the operating state under abnormal observation conditions provided in the embodiments of this application.

[0069] Preferably, such as Figure 11 As shown, after determining the target operating state, this embodiment further updates the operating state evolution memory structure according to the target operating state, so that the operating state evolution memory structure can be continuously improved as the running object continues to run. Unlike the prior art that treats the state recognition result as a one-time output, this embodiment believes that the state change law of the running object will continuously change with the increase of running time. Therefore, the currently determined target operating state is not only used to represent the current state of the running object, but can also be used as new state evolution information to supplement the historical state evolution structure, so that the subsequent state reconstruction process can utilize a more complete state change law.

[0070] Specifically, such as Figure 11 As shown, this embodiment determines whether to add new state nodes and corresponding state evolution association information based on the association between the target running state and existing state nodes. When there is a correspondence between the current target running state and existing state nodes, the evolution relationship between existing state nodes can be improved according to the current running process; when the current target running state belongs to a new state stage not recorded in the historical state evolution structure, a new state node can be generated, and a state evolution association relationship can be established between this state node and existing state nodes. Through the above dynamic update method, the running state evolution memory structure can not only represent the historical running process, but also continuously absorb new running state change patterns.

[0071] Preferably, when updating the state evolution memory structure, this embodiment focuses on updating the evolutionary relationships between state nodes, rather than simply adding new historical data records. Since the change patterns between different state stages are more important than the outcome of a single state during the state change process of an object, recording the direction, trend, and state transition relationships between the target operating state and historical states can further improve the state chain structure. For example, in the operation of industrial equipment, when a new performance change pattern emerges after long-term operation, this change process can be used to form new state evolution association information, allowing subsequent equipment state judgments to reference the new change patterns. In intelligent aquaculture scenarios, when aquaculture objects exhibit new behavioral change patterns, updating the state evolution relationships can also allow the system to gradually adapt to the state change characteristics of different growth stages and environmental conditions.

[0072] Preferably, such as Figure 11 As shown, after updating the state nodes and state evolution association information, this embodiment further updates the running state chain, enabling new state change processes to be integrated into existing state evolution paths. Through state chain updates, the running state evolution memory structure can maintain continuity; that is, the current state is not only recorded as a new state result but also linked to past state change processes, thus forming a more complete running state evolution process. Compared to traditional methods that rely on fixed models or rules for state judgment, this embodiment continuously updates the state chain, giving the running state reconstruction process a certain degree of adaptability. As the running time of the object increases, the system can continuously accumulate new state change paths and utilize the newly added state information to optimize subsequent state reasoning processes, thereby improving the stability of state reconstruction in long-term running scenarios.

[0073] Preferably, such as Figure 12 As shown, to address the potential for missing, abnormal, or incomplete observation information during actual operation, this embodiment further utilizes an operational state evolution memory structure to achieve continuous state reconstruction under abnormal observation conditions. Since real-time observation information may be affected by factors such as abnormal acquisition equipment, communication anomalies, and environmental interference, the current acquisition results may not fully reflect the true state of the operating object. If state judgment is directly based on missing information, the state reconstruction results may be inaccurate.

[0074] Therefore, in this embodiment, when real-time observation information cannot fully describe the current state, the state analysis is not stopped directly or an uncertain result is output. Instead, the associated historical state nodes are determined based on the currently available state information, and the possible current state is inferred by combining the state evolution paths corresponding to the associated historical state nodes. The associated historical state nodes provide a reference for historical changes corresponding to the current state, and the state evolution path represents the possible process of the running object changing from a historical state to the current state.

[0075] For example, during the operation of industrial equipment, when abnormal sensor data leads to a lack of current status information, the current possible state can be inferred based on the equipment's previous operating status, historical trends, and existing state evolution paths. In intelligent farming scenarios, when some behavioral monitoring information cannot be obtained, the current state can be inferred by combining the animal's historical behavioral patterns with current environmental information. Through the above methods, even if some real-time observation information is abnormal, this embodiment can still maintain the continuity of the state reconstruction process by utilizing historical state evolution patterns.

[0076] Preferably, by combining the state reconstruction process with the runtime state evolution memory update process, this embodiment realizes the transformation of the runtime state analysis process from static judgment to dynamic evolution analysis. After a state reconstruction is completed, the determined target runtime state can be further fed back into the state evolution memory structure to improve historical state nodes and state evolution relationships. In subsequent state analysis processes, the updated state evolution memory structure can be used to generate candidate states that are more consistent with the actual runtime process. Through this cyclical update mechanism, the runtime state evolution memory structure can be continuously optimized as the running object runs for a long time, improving the accuracy and adaptability of runtime state reconstruction in complex dynamic environments.

[0077] Application Scenario 1: Industrial Equipment Operation Status Reconstruction Scenario Preferably, the operational status reconstruction method provided in this embodiment can be applied to scenarios such as industrial equipment operational status monitoring, equipment health management, and fault trend analysis. In industrial production processes, equipment operational status typically exhibits distinct evolutionary characteristics. Equipment anomalies are usually not directly caused by instantaneous parameter changes, but rather gradually form over long-term operation due to factors such as load variations, changes in the operating environment, equipment performance degradation, and historical anomaly processes. Therefore, the traditional method of comparing currently collected parameters with preset thresholds, while capable of identifying some obvious anomalies, struggles to accurately determine the current state stage of the equipment and cannot effectively distinguish between normal changes and abnormal evolution processes.

[0078] In this application scenario, real-time observation information during equipment operation can be acquired through sensors installed on industrial equipment. This includes information such as equipment temperature, vibration, current, pressure, load, and environmental conditions. This information is then combined with historical operation records, maintenance records, anomaly records, and state change records to form historical operation information. Based on the method provided in this application, the real-time observation information is first analyzed for state features, converting operating parameters from different sources into state features that reflect changes in the equipment's operating state. Furthermore, multi-dimensional state characterization information is generated based on the relationships between these different state features.

[0079] Preferably, during the operation of industrial equipment, since different operating parameters often have interrelationships, this embodiment does not use a single parameter to determine the equipment status, but rather establishes an equipment status characterization by combining the correlations between multiple status factors. For example, changes in equipment load may cause temperature changes, changes in mechanical vibration may reflect changes in the internal structure of the equipment, and changes in the operating environment may affect the performance of equipment parameters. Therefore, by establishing the correlations between status characteristics, the generated equipment status characterization information can not only reflect the current parameter level, but also reflect the changing trends during the formation of the equipment's operating status.

[0080] Furthermore, based on the state change patterns formed during the equipment's historical operation, an evolutionary memory structure for the equipment's operating state is established. This structure uses the historical states corresponding to different operating stages as state nodes, and establishes state evolution correlation information based on the parameter change patterns during the equipment's transition from normal operating state to performance-changing state and abnormal state, thus creating a directional state evolution path for the equipment's historical operation. In this way, historical equipment data is no longer limited to fault statistics but can participate in the current equipment state analysis process.

[0081] When the current operating state of the equipment needs to be reconstructed, the current multi-dimensional state representation information can be matched with the equipment's operating state evolution memory structure to determine historical state nodes with evolutionary relationships to the current state, and multiple candidate operating states can be generated based on the corresponding state evolution paths. For example, when the equipment experiences a temperature rise, this embodiment does not directly determine that the equipment has malfunctioned, but rather combines the historical operating process to determine whether the temperature change conforms to the normal load change pattern, or whether it matches the abnormal evolution path of the equipment, thereby generating multiple possible state results.

[0082] Preferably, after generating candidate operating states, constraint analysis and evaluation can be further performed by combining the current operating conditions of the equipment, state transition patterns, and related state information of the equipment to determine the target state that better reflects the actual operating process. For example, for the same temperature rise phenomenon, if the equipment load increases synchronously and there is a similar change process in the past, it can be judged as a change in normal operating state; if the equipment load does not change but the vibration continues to increase, and the historical state evolution path shows that this change usually corresponds to the abnormal development stage of the equipment, the abnormal state evaluation result can be improved.

[0083] Through the above methods, this embodiment combines real-time equipment monitoring processes, historical operating patterns, and state evolution paths, transforming equipment state judgment from traditional parameter threshold judgments to comprehensive analysis based on the state evolution process. This improves the accuracy of equipment operating state identification, anomaly trend detection, and health management. Furthermore, during continuous equipment operation, once a new equipment operating state is determined, this state information can be fed back to the equipment operating state evolution memory structure to update existing state nodes and state evolution association information. This allows the equipment state model to continuously adapt to changes during long-term equipment operation, improving the state reconstruction capability under long-term operating conditions.

[0084] Application Scenario 2: Intelligent Aquaculture Operation Status Reconstruction Scenario Preferably, the operational state reconstruction method provided in this embodiment can also be applied to intelligent aquaculture scenarios to analyze the health, growth, and behavioral states of aquaculture subjects. In intelligent aquaculture, the state of aquaculture subjects is typically influenced by various factors such as individual characteristics, growth stage, environmental conditions, and historical changes, resulting in strong continuity and phased changes. If the state of aquaculture subjects is judged solely based on currently collected activity levels, feed intake, body temperature, or environmental parameters, the influence of historical changes on the current state can easily be overlooked.

[0085] In this application scenario, real-time observation information of the farmed organisms can be obtained through image acquisition devices, behavior monitoring devices, environmental acquisition devices, and physiological parameter detection devices. This information is then combined with historical growth data, historical behavior data, historical environmental data, and historical health status records to form historical operational information. Based on the method provided in this application, the real-time observation information can first be analyzed for state features to extract behavioral features, physiological features, and environmental impact features that reflect the current state of the farmed organisms. Furthermore, multi-dimensional state representation information can be generated by analyzing the correlations between different state features.

[0086] Preferably, since changes in the state of farmed animals are usually influenced by multiple factors, this embodiment establishes correlations between state characteristics, enabling the state analysis process to simultaneously consider the interactions between behavioral changes, physiological changes, and environmental changes. For example, changes in the temperature of the farming environment may lead to changes in animal activity levels, which in turn may reflect changes in health status. If only the single indicator of activity level is analyzed, it may be impossible to distinguish between normal environmental adaptation behavior and abnormal health behavior. Therefore, by combining the correlations between multiple state factors, the completeness of the description of the current state of the farmed animals can be improved.

[0087] Furthermore, based on the state change patterns formed during the historical growth process of the farmed organisms, an evolutionary memory structure of the operational state of the farmed organisms is established. This structure uses the states corresponding to different growth stages and different health stages as state nodes, and establishes state evolution correlation information based on the relationships between behavioral changes, physiological changes, and environmental changes. This enables the system to record the evolutionary patterns of the farmed organisms as they change from normal to abnormal states.

[0088] When changes in behavior, feed intake, or physiological indicators are detected in farmed animals during real-time monitoring, this embodiment does not directly determine the health status based on a single abnormal manifestation. Instead, it matches the current multidimensional state representation information with the historical state evolution memory structure to determine the possible corresponding historical state stage and generates multiple candidate states based on the historical evolution path. For example, when an animal's activity level decreases, it can be determined whether the change belongs to a normal environmental adaptation process or whether it is consistent with the historical evolutionary process of abnormal health states, in conjunction with the current environmental changes.

[0089] Preferably, after determining the candidate states, a comprehensive evaluation can be conducted by combining the current environmental conditions of the aquaculture object, historical state transition patterns, and state correlations to determine the target operating state. When some real-time observation information is missing or abnormal, continuous state reconstruction can be performed using historical state evolution paths to avoid the problem of being unable to determine the state of the aquaculture object due to anomalies in a single monitoring piece of information.

[0090] Through the above methods, this embodiment can integrate and analyze the behavioral, physiological, and environmental information of the farmed animals, as well as their historical change patterns. This enables the intelligent farming system not only to detect current abnormal behaviors but also to judge the trend of abnormality formation by combining the state evolution process, thereby improving the health monitoring and intelligent management capabilities of the farmed animals.

[0091] Preferably, as can be seen from the above embodiments, the operational state reconstruction method provided in this embodiment does not only identify the state based on the real-time observation information currently collected by the operational object, but combines the current state representation process with the historical state evolution process. By constructing an operational state evolution memory structure, historical operational information can participate in the current state reasoning process. Specifically, this embodiment first performs state feature analysis on the real-time observation information, and generates multi-dimensional state representation information that reflects the current state of the operational object by analyzing the correlation between different state features; at the same time, based on the state change patterns in the historical operational information, an operational state evolution memory structure including state nodes and state evolution correlation information is constructed, so that the historical operational process is transformed from the traditional data recording form into structured information that can describe the state change patterns.

[0092] Furthermore, this embodiment utilizes the correlation between the current multidimensional state representation information and the operational state evolution memory structure to generate a candidate operational state set. It then combines state constraint information and state evaluation information to screen and evaluate the candidate operational states, determining the target operational state that best matches the current operational process from multiple possible state outcomes. Compared to methods that rely solely on current parameter changes for state judgment, this embodiment introduces historical state evolution patterns, enabling the state reconstruction process to simultaneously consider current state performance, historical change trends, and actual operational constraints, thereby improving the consistency between the state judgment result and the actual state of the operational object.

[0093] Furthermore, after determining the target operating state, this embodiment further updates the operating state evolution memory structure using the target operating state, enabling the system to continuously absorb new state change patterns of the operating object and provide a more complete evolutionary basis in subsequent state analysis. Simultaneously, in the event of missing or abnormal real-time observation information, this embodiment can also perform continuous state reconstruction based on historical state evolution paths, reducing the impact of single observation information anomalies on the state judgment results.

[0094] Therefore, this embodiment combines multi-dimensional state representation, state evolution memory, candidate state reasoning, constraint evaluation, and dynamic update processes to form a continuous state reconstruction mechanism for dynamically operating objects. Through the aforementioned operating state evaluation mechanism, this embodiment can determine the current operating state when multiple candidate operating states all have a certain degree of rationality, and uses state confidence information to express the credibility of the operating state reconstruction result. This application transforms the operating state analysis process from a static judgment based on current data to a dynamic reasoning process that combines historical evolution patterns, thereby improving the accuracy, continuity, and adaptability of state reconstruction in complex operating environments.

[0095] like Figure 13As shown, this application also provides an operating state reconfiguration device, comprising: The state representation construction module is used to obtain historical operation information and real-time observation information corresponding to the running object, and perform state feature parsing based on the real-time observation information to generate multi-dimensional state representation information describing the current operation characteristics of the running object; The state evolution memory construction module is used to construct an operating state evolution memory structure based on the historical state change process in the historical operating information. The operating state evolution memory structure includes an operating state chain formed by state nodes and state evolution association information, and generates a candidate operating state set based on the multi-dimensional state representation information and the operating state evolution memory structure. The state determination module is used to filter the candidate operating state set based on state constraint information, and determine the target operating state that meets the operating evolution conditions based on state evaluation information. The state evaluation information includes state confidence, which characterizes the credibility of the candidate operating state as the target operating state. The state chain update module is used to update the state nodes and state evolution association information in the running state evolution memory structure according to the state evolution information corresponding to the target running state, so as to form an updated running state chain and maintain the continuity of running state evolution.

[0096] like Figure 14 As shown, an electronic device according to an embodiment of this application includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is configured to implement the steps of the running state reconstruction method as described above in this application.

[0097] like Figure 15 As shown, a computer-readable storage medium is provided in an embodiment of this application. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the running state reconstruction method described above in this application.

[0098] The above Figures 13-15 For an exemplary description, please refer to the above. Figure 1 This will not be elaborated upon here.

Claims

1. A method for reconstructing operational states, characterized in that, include: The historical operation information and real-time observation information corresponding to the running object are obtained, and the state feature is parsed based on the real-time observation information to generate multi-dimensional state representation information describing the current operation characteristics of the running object. An operational state evolution memory structure is constructed based on the historical state change process in the historical operational information. The operational state evolution memory structure includes an operational state chain formed by state nodes and state evolution association information. A candidate operational state set is generated based on the multidimensional state representation information and the operational state evolution memory structure. The candidate operating state set is filtered based on state constraint information, and the target operating state that meets the operating evolution conditions is determined based on state evaluation information. The state evaluation information includes state confidence, which characterizes the credibility of the candidate operating state as the target operating state. The state nodes and state evolution association information in the running state evolution memory structure are updated according to the state evolution information corresponding to the target running state to form an updated running state chain, so as to maintain the continuity of running state evolution.

2. The operational state reconstruction method according to claim 1, characterized in that, The process of parsing state features based on the real-time observation information to generate the multi-dimensional state representation information includes: The real-time observation information is parsed to determine the corresponding observation information category; State features are extracted from the real-time observation information according to the observation information category to obtain multiple state features of different dimensions, and a state feature set is formed based on the state features. Perform correlation analysis on the state features in the state feature set to determine the correlation relationships between the state features; Based on the aforementioned correlation, multiple state features are fused to generate the multidimensional state representation information.

3. The operational state reconstruction method according to claim 2, characterized in that, The association analysis of the state features in the state feature set includes: Obtain the feature attribute information corresponding to different state features in the state feature set; The association relationships between different state features are determined based on the aforementioned feature attribute information; Based on the aforementioned relationship, feature association information is generated to describe the degree of association between state features, and state feature fusion processing is performed based on the feature association information.

4. The operational state reconstruction method according to claim 1, characterized in that, The construction of the operational state evolution memory structure based on the historical state change process in the historical operational information includes: Multiple historical state information is extracted from the historical operation information, and the state change process corresponding to the multiple historical state information is analyzed; Based on the temporal correlation, state change relationship, and state change path among multiple historical state information, historical state association information is generated; Based on the historical state association information, determine the state nodes corresponding to different historical states; State evolution association information is established based on the state change relationships between the state nodes, forming the running state evolution memory structure.

5. The operational state reconstruction method according to claim 4, characterized in that, The generated historical state association information includes: The temporal correlation between historical states is determined based on the occurrence time of the states corresponding to multiple historical state information. The relationship between states is determined based on the direction and degree of state change between adjacent historical states; Based on the time correlation and the state change relationship, the evolutionary correlation characteristics between historical states are determined, and the historical state correlation information is generated.

6. The operational state reconstruction method according to claim 4, characterized in that, The establishment of state evolution association information includes: Determine the direction of state change between state nodes based on the historical state characteristics corresponding to different state nodes; The state transition relationship between state nodes is determined based on the direction of state change. Based on the state transition relationship, state evolution association information is generated to represent the state evolution path between different state nodes.

7. The operational state reconstruction method according to claim 1, characterized in that, The generation of a candidate running state set based on the multidimensional state representation information and the running state evolution memory structure includes: Extract the state change features corresponding to the current operating state based on the multidimensional state representation information; Determine state matching features based on the state change characteristics; Based on the state matching features, associated state nodes are determined in the running state evolution memory structure; A set of candidate running states is generated based on the state evolution association information corresponding to the associated state nodes.

8. An operating state reconfiguration device, characterized in that, include: The state representation construction module is used to obtain historical operation information and real-time observation information corresponding to the running object, and perform state feature parsing based on the real-time observation information to generate multi-dimensional state representation information describing the current operation characteristics of the running object; The state evolution memory construction module is used to construct an operating state evolution memory structure based on the historical state change process in the historical operating information. The operating state evolution memory structure includes an operating state chain formed by state nodes and state evolution association information, and generates a candidate operating state set based on the multi-dimensional state representation information and the operating state evolution memory structure. The state determination module is used to filter the candidate operating state set based on state constraint information, and determine the target operating state that meets the operating evolution conditions based on state evaluation information. The state evaluation information includes state confidence, which characterizes the credibility of the candidate operating state as the target operating state. The state chain update module is used to update the state nodes and state evolution association information in the running state evolution memory structure according to the state evolution information corresponding to the target running state, so as to form an updated running state chain and maintain the continuity of running state evolution.

9. An electronic device, characterized in that, The electronic device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the operating state reconstruction method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the runtime state reconstruction method as described in any one of claims 1 to 7.

Images