Digital control method and system
By performing multimodal analysis and consistency calculation on the real-time dynamic demand information flow of the target object, and coordinating the adjustment of the control network topology and node parameters, the problem of intention element acquisition deviation in the existing technology is solved, and efficient and reliable digital control is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HENGYANG QICHENG INFORMATION TECHNOLOGY CO LTD
- Filing Date
- 2026-01-14
- Publication Date
- 2026-04-21
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing digital control methods struggle to efficiently and accurately deconstruct the real-time dynamic demand information flow of target objects online. They lack deep integration processing of time-series data streams, event signal streams, and natural language text streams, resulting in biased or missing acquisition of intent elements, weak adaptive capabilities of control networks, and difficulty in accurately matching the actual needs of target objects.
By acquiring the real-time dynamic demand information flow of the target object, performing multimodal parsing and intent deconstruction, constructing an initial control network, and combining it with full-dimensional state transition data for consistency calculation, coordinating the adjustment of topology connections and node parameters, generating an adaptive control network, and finally encapsulating and encoding it to obtain the final control command.
It achieves precise adaptation to the control network, improves the adaptability of the control network and the accuracy of the control logic, and enhances the overall efficiency and reliability of digital control.
Smart Images
Figure CN121900261A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of digital control technology, and in particular to a digital control method and system. Background Technology
[0002] In the field of digital control, existing technologies often struggle to efficiently and accurately deconstruct the real-time dynamic demand information flow of target objects online. Traditional methods are mostly limited to the parsing of single-modal data, lacking in-depth fusion processing of time-series data streams, event signal streams, and natural language text streams. At the same time, they fail to effectively verify the information integrity and logical consistency of the extracted potential intent fragments, resulting in deviations or omissions in the obtained intent elements. This, in turn, affects the rationality and reliability of subsequent control network construction, making it difficult to accurately match the actual needs of the target object.
[0003] Existing digital control methods have significant shortcomings in the initial control network optimization and adjustment stage. They fail to fully integrate the comprehensive state transition data of the target object for systematic analysis. Traditional optimization methods lack effective purification and normalization of state data, as well as spatiotemporal alignment with intent elements. The scientific rigor of consistency calculations is insufficient, resulting in limited accuracy in generating logical structure optimization vectors. The adjustment of the topology connections and node parameters of the initial control network is mostly an independent operation, lacking coordination. This makes the optimized control network less adaptive, and the final output control commands are difficult to adapt to the dynamic state changes of the target object. Consequently, the control efficiency and accuracy fail to meet expectations. Therefore, improving the accuracy of digital control has become an urgent problem to be solved. Summary of the Invention
[0004] This invention provides a digital control method and system to solve the problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides a digital control method, comprising: S1. Obtain the real-time dynamic demand information flow of the target object, and perform online intent deconstruction on the real-time dynamic demand information flow to obtain the intent elements of the target object; S2. Perform topology construction on the intent elements to obtain the initial control network of the target object; S3. Apply the initial control network and simultaneously collect the full-dimensional state transition data of the target object; S4. Perform consistency calculation on the full-dimensional state transition data and the intent elements to obtain the logical structure optimization vector of the target object; S5. Based on the logical structure optimization vector, the topology connections and node parameters in the initial control network are coordinated and adjusted to obtain the adaptive control network of the target object. S6. Encapsulate and encode the adaptive control network to obtain the final control command for the target object.
[0006] In a preferred embodiment, the step of acquiring the real-time dynamic demand information flow of the target object and performing online intent deconstruction on the real-time dynamic demand information flow to obtain the intent elements of the target object includes: Collect real-time dynamic demand information flow of the target object; Multimodal parsing is performed on the real-time dynamic demand information stream to obtain the time-series data stream, event signal stream, and natural language text stream of the target object; Syntactic dependency analysis is performed on the natural language text stream to obtain a preliminary semantic framework of the target object; By context binding the time-series data stream and the event signal stream, composite information of the target object is obtained; Cluster analysis is performed on the composite information to obtain the potential intent fragments of the target object; By removing incomplete information segments and logically contradictory segments from the potential intent segments, the intent elements of the target object are obtained.
[0007] In a preferred embodiment, the step of topology construction of the intent elements to obtain the initial control network of the target object includes: Logical correlation analysis is performed on the intent elements to obtain the logical correlation relationships of the target object; The logical relationships are quantitatively evaluated to obtain the logical relationship strength of the target object; Using the intent elements as nodes and the logical relationships as edges, and assigning confidence weights to the edges based on the strength of the logical relationships, an initial logical topology structure for the target object is constructed. The initial logical topology is subjected to stability analysis, and the analyzed logical topology is formatted and encapsulated to obtain the initial control network of the target object.
[0008] In a preferred embodiment, applying the initial control network and simultaneously collecting full-dimensional state transition data of the target object includes: The initial control network is compiled into executable logic to obtain the control instruction stream of the target object; Based on the control instruction flow, determine the requirement descriptor of the target object; Based on the demand descriptor, the original state data stream of the target object is parsed to obtain the standard state data stream of the target object; The standard state data stream is time-aligned to obtain a time-series data segment of the target object; Pattern matching is performed on the time-series data segments to obtain the abnormal events of the target object; The abnormal events are fused and encapsulated to obtain full-dimensional state transition data of the target object.
[0009] In a preferred embodiment, the step of performing consistency calculations on the full-dimensional state transition data and the intent elements to obtain the logical structure optimization vector of the target object includes: Identify outlier data points in the full-dimensional state transition data to obtain clean state transition data for the target object; The clean state transition data is parsed using a multi-source format, and the parsed state transition data is normalized to obtain the regularized state vector of the target object. The intent elements are vectorized and encoded to obtain the intent feature vector of the target object; By temporally binding the regularized state vector and the intention feature vector, the temporal mapping relationship of the target object is obtained; Based on the temporal mapping relationship, linear interpolation is performed on the regularized state vector and the intention feature vector to obtain the spatiotemporal aligned dataset of the target object; The intention-state data pairs in the spatiotemporal aligned dataset are compared for discrepancies to obtain a consistency deviation measure for the intention-state data pairs. Tensor synthesis is performed on the consistency deviation metric to obtain the logical structure optimization vector of the target object.
[0010] In a preferred embodiment, the formula for calculating the consistency deviation metric is: ; in, This represents the consistency deviation measure. Indicates the first The stated intention feature vector Indicates the first The regularized state vectors mentioned above. The Euclidean norm of a vector. Represents absolute value. This represents the preset smallest positive number. This represents the preset time decay coefficient. Indicates the first The absolute time difference of the stated intent-state data pairs This represents the preset sensitivity coefficient to state fluctuations. Indicates the first The variance of each of the stated intent-state data pairs This represents an exponential function.
[0011] In a preferred embodiment, the step of coordinating the adjustment of the topology connections and node parameters in the initial control network based on the logical structure optimization vector to obtain the adaptive control network of the target object includes: Based on the initial control network, the logical structure optimization vector is reconstructed to obtain the topology adjustment instruction and node parameter adjustment instruction of the target object; Based on the topology adjustment command, the graph structure of the topology connection in the initial control network is edited to obtain the optimized topology connection of the target object; Based on the node parameter adjustment instructions, the node parameters in the initial control network are collaboratively optimized to obtain the optimized node parameters of the target object. Based on the optimized topology and the optimized node parameters, construct the optimized control network for the target object; The performance of the optimized control network is evaluated to obtain the adaptive control network of the target object.
[0012] In a preferred embodiment, the step of reconstructing the logical structure optimization vector based on the initial control network to obtain the topology adjustment instructions and node parameter adjustment instructions for the target object includes: The optimized logical structure vector is mapped to the initial control network to obtain the associated components of the target object; The associated components are classified by intent to obtain the structural class components and parametric class components of the target object; The structural components and parameter components are normalized and decoded to obtain the initial topology adjustment instruction and the initial node parameter adjustment instruction of the target object; The logical correlation between the initial topology adjustment instruction and the initial node parameter adjustment instruction is verified to obtain the topology adjustment instruction and node parameter adjustment instruction of the target object.
[0013] In a preferred embodiment, the step of encapsulating and encoding the adaptive control network to obtain the final control command for the target object includes: The adaptive control network is traversed topologically to obtain the core topological connections and active node parameters of the target object; The core topology connection is jointly bound with the active node parameters to obtain the structured network object of the target object; Convert the structured network object into an intermediate instruction representation of the target object; The intermediate instruction representation is logically reconstructed to obtain the final control instruction for the target object.
[0014] To address the above problems, the present invention also provides a digital control system, the system comprising: The intent deconstruction module is used to acquire the real-time dynamic demand information flow of the target object and perform online intent deconstruction on the real-time dynamic demand information flow to obtain the intent elements of the target object; An initial network construction module is used to construct the topology of the intent elements to obtain the initial control network of the target object; The state transition data collection module is used to apply the initial control network and synchronously collect full-dimensional state transition data of the target object; The optimized vector generation module is used to perform consistency calculations on the full-dimensional state transition data and the intent elements to obtain the logical structure optimized vector of the target object. The initial network optimization module is used to coordinately adjust the topology connections and node parameters in the initial control network based on the logical structure optimization vector to obtain the adaptive control network of the target object. The final instruction generation module is used to encapsulate and encode the adaptive control network to obtain the final control instructions for the target object.
[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention performs online intent deconstruction on the real-time dynamic demand information flow of the target object, accurately extracts intent elements and constructs an initial control network. By combining the consistency calculation of full-dimensional state transition data and intent elements, it realizes the coordinated adjustment of the control network topology connection and node parameters, effectively improving the adaptability of the control network to the needs of the target object and the accuracy of the control logic, making the control process more in line with the actual operating state of the target object.
[0016] 2. Through multimodal analysis, timing alignment, and deviation measurement, this invention can comprehensively capture the state transition information of the target object. By optimizing the vector through logical structure, the control network can be adaptively optimized. The subsequent encapsulation and encoding process further ensures the integrity and executability of the final control command, significantly improving the overall efficiency of digital control and providing more efficient and reliable digital control support for the target object. Attached Figure Description
[0017] Figure 1 This is a flowchart illustrating a digital control method according to an embodiment of the present invention. Figure 2 A functional block diagram of a digital control system provided in an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation
[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
[0019] This application provides a digital control method. The executing entity of this digital control method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the digital control method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.
[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a digital control method according to an embodiment of the present invention. In this embodiment, the digital control method includes: S1. Obtain the real-time dynamic demand information flow of the target object, and perform online intent deconstruction on the real-time dynamic demand information flow to obtain the intent elements of the target object; In this embodiment of the invention, the step of acquiring the real-time dynamic demand information flow of the target object and performing online intent deconstruction on the real-time dynamic demand information flow to obtain the intent elements of the target object includes: Collect real-time dynamic demand information flow of the target object; Multimodal parsing is performed on the real-time dynamic demand information stream to obtain the time-series data stream, event signal stream, and natural language text stream of the target object; Syntactic dependency analysis is performed on the natural language text stream to obtain a preliminary semantic framework of the target object; By context binding the time-series data stream and the event signal stream, composite information of the target object is obtained; Cluster analysis is performed on the composite information to obtain the potential intent fragments of the target object; By removing incomplete information segments and logically contradictory segments from the potential intent segments, the intent elements of the target object are obtained.
[0021] By using sensors, data interfaces, and interactive terminals related to the target object, various demand-related data generated by the target object during operation are continuously captured. These data comprehensively cover the demand-oriented information transmitted by the target object in different application scenarios, and are ultimately integrated to form a complete and continuous real-time dynamic demand information flow.
[0022] For the collected real-time dynamic demand information stream, the data is classified and analyzed according to its presentation form and source attributes. The continuously generated quantitative data sequence over time is extracted into a time-series data stream, the discrete signals with specific triggering significance are extracted into an event signal stream, and the demand description text expressed in natural language is extracted into a natural language text stream. This ensures that the three data streams accurately correspond to the relevant content in the original information stream without any omissions.
[0023] Each sentence in the natural language text stream is analyzed sentence by sentence to identify the grammatical function of each word in the sentence and the dependency relationship between words. The logical relationship of core grammatical components such as subject, predicate, and object is clarified, and the core semantic orientation expressed by the sentence is sorted out. Based on these semantic relationships, a preliminary semantic framework that can fully reflect the core meaning of the text is constructed.
[0024] Using a unified timeline as a benchmark, the quantized data at different time nodes in the time-series data stream are correlated and matched with the trigger signals occurring within the same time interval in the event signal stream. At the same time, the specific background information of the two scenarios is combined to organically integrate the interrelated time-series data and event signals, forming a composite information that includes both continuous quantized information and discrete trigger information.
[0025] Based on the demand tendencies and inherent connections reflected in composite information, composite information with related content and similar logical orientations is grouped into the same category. By systematically sorting out the core content and common demand orientations of each type of composite information, information fragments that can centrally reflect a certain aspect of the target audience's demand tendencies are extracted. These information fragments are potential intention fragments.
[0026] Each potential intent fragment is checked for completeness to determine if any key information is missing or data elements are incomplete. Such fragments with incomplete information are directly eliminated. At the same time, the remaining fragments are checked for logical consistency by comparing the logical relationships within the fragments and between different fragments. Fragments with logical contradictions or causal conflicts are eliminated. The complete and logically consistent fragments that remain after screening are the intent elements.
[0027] The beneficial effect is that the implementation process, through precise processing in multiple stages, gradually extracts and filters complete and consistent intent elements from the real-time dynamic demand information flow, ensuring that the intent elements can accurately reflect the real needs of the target object, providing reliable and accurate basic data support for the subsequent construction of the initial control network, and ensuring the pertinence and effectiveness of the digital control process.
[0028] S2. Perform topology construction on the intent elements to obtain the initial control network of the target object; In this embodiment of the invention, the step of performing topology construction on the intent elements to obtain the initial control network of the target object includes: Logical correlation analysis is performed on the intent elements to obtain the logical correlation relationships of the target object; The logical relationships are quantitatively evaluated to obtain the logical relationship strength of the target object; Using the intent elements as nodes and the logical relationships as edges, and assigning confidence weights to the edges based on the strength of the logical relationships, an initial logical topology structure for the target object is constructed. The initial logical topology is subjected to stability analysis, and the analyzed logical topology is formatted and encapsulated to obtain the initial control network of the target object.
[0029] The core content and functional orientation of each intent element are systematically analyzed, and the interaction between different intent elements in supporting the realization of the target object's needs is deeply analyzed. The direct or indirect relationships between the intent elements, such as causality, progression, and parallelism, are clarified. After systematically organizing these clear relationships, the logical relationships of the target object are formed.
[0030] Based on the actual impact of logical relationships on the implementation of target object requirements, and considering the frequency of occurrence, scope of effect, and necessity of the relationships for fulfilling overall requirements, a specific strength assessment is made for each logical relationship. By comprehensively considering these actual performances, the strength level of each logical relationship is determined, ultimately yielding the logical relationship strength of the target object.
[0031] Each independent intent element is individually identified as an independent node in the topology. The logical relationships identified are used as edges connecting the corresponding nodes. Based on the previously determined logical relationship strength, a corresponding confidence weight is assigned to each edge connecting the nodes. The higher the relationship strength, the larger the corresponding confidence weight value. According to the actual logical relationship order between intent elements, all nodes, edges and corresponding confidence weights are organically integrated to construct the initial logical topology of the target object.
[0032] The system simulates the operation of the initial logical topology under different operating scenarios of the target object, continuously monitors the stability of connections between nodes and whether the confidence weights of edges remain stable, and determines whether the topology can continuously and accurately reflect the logical relationships between intent elements in different environments. If there are potential instabilities, the system promptly investigates and confirms that the structure is stable. Then, the logical topology is standardized according to the preset network format specifications, clarifying the storage format, data interface, and calling rules of nodes, edges, and confidence weights. After formatting and encapsulation, the initial control network of the target object is obtained.
[0033] The beneficial effect is that the implementation process transforms the intent elements into a clear, well-connected, and stable initial control network through progressive processing. This ensures that the initial control network can accurately map the logical relationships between the intent elements, providing a structured and executable framework for the subsequent application of the initial control network and the collection of full-dimensional state transition data, thus guaranteeing the orderly advancement of the digital control process.
[0034] S3. Apply the initial control network and simultaneously collect the full-dimensional state transition data of the target object; In this embodiment of the invention, the application of the initial control network and the synchronous acquisition of full-dimensional state transition data of the target object includes: The initial control network is compiled into executable logic to obtain the control instruction stream of the target object; Based on the control instruction flow, determine the requirement descriptor of the target object; Based on the demand descriptor, the original state data stream of the target object is parsed to obtain the standard state data stream of the target object; The standard state data stream is time-aligned to obtain a time-series data segment of the target object; Pattern matching is performed on the time-series data segments to obtain the abnormal events of the target object; The abnormal events are fused and encapsulated to obtain full-dimensional state transition data of the target object.
[0035] A comprehensive analysis is conducted on the topology, node relationships, and edge confidence weights of the initial control network. The control logic and execution rules contained therein are sorted out, and these abstract logical rules are transformed into a continuous sequence of instructions that the machine can directly recognize and execute. Each instruction corresponds to a specific control action in the initial control network, ultimately forming a coherent and implementable control instruction flow.
[0036] By deeply analyzing the core control objectives and execution directions of each instruction in the control instruction flow, key information that can summarize the core needs of the target object is extracted, including control direction, expected effect, execution boundary, etc. This key information is then integrated according to standardized description specifications to form a requirement descriptor that can accurately define the needs of the target object.
[0037] Based on the requirement descriptor, identify the key data fields that need to be extracted and retained in the original state data stream of the target object. Decompose and reorganize the original state data stream according to the unified format requirements, transform the data storage format, encoding method and field arrangement order, eliminate the format differences of data from different sources, and ensure that all data conforms to the unified specification standard, thereby obtaining the standard state data stream.
[0038] A precise timestamp is added to each data point in the standard state data stream. Based on a fixed time interval, the standard state data acquired from different acquisition periods and different acquisition devices are calibrated in the time dimension so that all data points form a continuous and consistent data sequence in chronological order. Then, the complete data set in different time segments is divided according to the actual application requirements to obtain time-series data segments.
[0039] Based on historical data of the target object under normal operating conditions, a standard data model covering various conventional operating scenarios is constructed. This model includes the data fluctuation range, change trend and correlation characteristics during normal operation. Each time series data segment is compared and analyzed point by point with the standard data model to identify abnormal data sequences in the data segments that exceed the normal fluctuation range and do not conform to the conventional change trend. The changes in the target object's operating status corresponding to these abnormal data sequences are abnormal events.
[0040] All identified abnormal events are summarized, and the occurrence time, specific data manifestations, involved state dimensions, and degree of impact on the operation of the target object are recorded in detail for each abnormal event. This information is classified and organized, and duplicate and redundant content is removed. The information is then integrated and encapsulated in a unified structured format to form full-dimensional state transition data that can comprehensively and completely reflect all state changes of the target object during operation.
[0041] The beneficial effects are that the implementation process, through the progressive processing from control command flow to full-dimensional state transition data, ensures the standardization, temporal consistency and integrity of state data, accurately captures abnormal changes in the target object's operation process, and provides comprehensive and reliable data support for the subsequent consistency calculation of full-dimensional state transition data and intent elements, thus ensuring the accuracy and effectiveness of the digital control optimization process.
[0042] S4. Perform consistency calculation on the full-dimensional state transition data and the intent elements to obtain the logical structure optimization vector of the target object; In this embodiment of the invention, the step of performing consistency calculations on the full-dimensional state transition data and the intent elements to obtain the logical structure optimization vector of the target object includes: Identify outlier data points in the full-dimensional state transition data to obtain clean state transition data for the target object; The clean state transition data is parsed using a multi-source format, and the parsed state transition data is normalized to obtain the regularized state vector of the target object. The intent elements are vectorized and encoded to obtain the intent feature vector of the target object; By temporally binding the regularized state vector and the intention feature vector, the temporal mapping relationship of the target object is obtained; Based on the temporal mapping relationship, linear interpolation is performed on the regularized state vector and the intention feature vector to obtain the spatiotemporal aligned dataset of the target object; The intention-state data pairs in the spatiotemporal aligned dataset are compared for discrepancies to obtain a consistency deviation measure for the intention-state data pairs. Tensor synthesis is performed on the consistency deviation metric to obtain the logical structure optimization vector of the target object.
[0043] The formula for calculating the consistency deviation metric is as follows: ; in, This represents the consistency deviation measure. Indicates the first The stated intention feature vector Indicates the first The regularized state vectors mentioned above. The Euclidean norm of a vector. Represents absolute value. This represents the preset smallest positive number. This represents the preset time decay coefficient. Indicates the first The absolute time difference of the stated intent-state data pairs This represents the preset sensitivity coefficient to state fluctuations. Indicates the first The variance of each of the stated intent-state data pairs This represents an exponential function.
[0044] Referring to the range of state data and data change patterns during normal operation of the target object, each data point in the full-dimensional state transition data is checked one by one to determine whether it exceeds the normal data range, whether there are sudden changes with adjacent data points without reasonable basis, or whether there is a logical contradiction with the overall data trend. All data points that meet the above abnormal conditions are marked and removed. The remaining data that meets the normal operation rules and is complete is integrated to form clean state transition data.
[0045] For clean state transition data from different collection sources with different format specifications, the data is classified and decomposed according to the source attributes and format types. All types of data are parsed into a standardized basic data form. Then, all parsed data are scaled and adjusted according to a fixed ratio so that the numerical range of all data is uniformly mapped to the same preset interval, eliminating the magnitude difference of data from different sources. Finally, these processed data are arranged in a fixed order to form a regular state vector.
[0046] The core semantic information, functional orientation features, and demand priorities of each intent element are extracted in depth. This abstract element information is transformed into a quantifiable numerical form. The numerical values are arranged in order according to the feature categories and logical relationships of the elements. All the quantifiable values corresponding to each intent element are combined in sequence to form an intent feature vector that can completely and accurately reflect the core attributes of the intent element.
[0047] Using a unified timeline as a benchmark, we sort out the time nodes or time intervals corresponding to the regularized state vectors and the intention feature vectors, and establish a one-to-one correspondence between the regularized state vectors and the intention feature vectors within the same time node or time interval. We clarify the temporal correspondence between the actual state reflected by each regularized state vector and the expected demand reflected by the corresponding intention feature vector. This relationship established on the time dimension is called the temporal mapping relationship.
[0048] Based on the temporal distribution of the regularized state vector and the intention feature vector in the temporal mapping relationship, intermediate vector data that conforms to the changing trend of the two vector pairs corresponding to two adjacent time nodes is generated according to the numerical characteristics and time interval of the vectors before and after. This makes the regularized state vector and the intention feature vector form a continuous and unbroken distribution on the time axis, ensuring that there is a corresponding regularized state vector and intention feature vector matching each other at each time point. These complete and time-aligned vector pairs together constitute the spatiotemporally aligned dataset.
[0049] Each intent-state data pair in the spatiotemporal aligned dataset is extracted one by one. The expected demand state carried by the intent feature vector in the data pair is compared with the actual operating state presented by the regular state vector. The deviation between the two in terms of core feature orientation, key information expression and state value range is analyzed. The degree and range of this deviation are comprehensively described to form a consistency deviation metric that can accurately reflect the difference between the two.
[0050] Collect all intent-state data pairs and their corresponding consistency deviation measurement results. Classify and integrate these deviation measurements according to the time dimension, feature dimension, and state association dimension. Construct a holistic data structure that covers multi-dimensional information from the scattered deviation measurement results. Then, systematically integrate and refine this multi-dimensional data structure to transform it into a vector form that can directly guide the optimization and adjustment of the initial control network, and finally form a logical structure optimization vector.
[0051] Intent elements are transformed into a computable vector form through vectorization encoding, ultimately yielding the intent feature vector; clean state transition data is first decomposed into different source format differences through multi-source format parsing, and then normalized to adjust the data to a uniform scale, ultimately yielding a regularized state vector; the absolute time difference of the intent-state data pair is obtained by directly acquiring the absolute value of the difference in the time dimension of the data pair; the variance of the intent-state data pair is obtained by calculating the average of the squares of the differences between all data points in the data pair and the data mean.
[0052] The preset minimum normal number is a fixed value set in advance before the calculation begins; the preset time decay coefficient is a fixed value set in advance before the calculation begins to adjust the degree of time influence; the preset state fluctuation sensitivity coefficient is a fixed value set in advance before the calculation begins to adjust the degree of state fluctuation influence.
[0053] The method comprehensively reflects the degree of consistency deviation between the intention feature vector and the regular state vector by calculating the ratio of the absolute value of the dot product of the intention feature vector and the regular state vector to the product of their Euclidean norms plus a preset minimum normal number, multiplying it by the result of an exponential function determined by the time decay coefficient and the absolute time difference of the intention-state data pair, and finally adding the product of the state fluctuation sensitivity coefficient and the variance of the intention-state data pair. The preset minimum normal number can avoid the case where the denominator is zero. The exponential function part reflects the decay effect of time difference on consistency, and the product of the state fluctuation sensitivity coefficient and the variance reflects the additional effect of state fluctuation on consistency deviation.
[0054] When the absolute value of the dot product of the intention feature vector and the regularized state vector increases while the product of their Euclidean norms remains unchanged, the aforementioned ratio will increase, thus increasing the overall result; when the product of the Euclidean norms of the intention feature vector and the regularized state vector increases while the absolute value of their dot product remains unchanged, the aforementioned ratio will decrease, thus decreasing the overall result.
[0055] When the absolute time difference of the intent-state data pair increases, the result of the exponential function decreases, which in turn reduces the product of the aforementioned ratio and the result of the exponential function, resulting in a decrease in the overall result. When the variance of the intent-state data pair increases while the state fluctuation sensitivity coefficient remains unchanged, the product of the two increases, resulting in a increase in the overall result. When the time decay coefficient increases while the absolute time difference of the intent-state data pair remains unchanged, the result of the exponential function decreases, which in turn reduces the product of the aforementioned ratio and the result of the exponential function, resulting in a decrease in the overall result.
[0056] When the state fluctuation sensitivity coefficient increases and the variance of the intention-state data pair remains unchanged, the product of the two will increase, thus increasing the overall result; when the preset minimum normal number increases, the aforementioned ratio will decrease, thus decreasing the overall result.
[0057] The beneficial effect is that the implementation process, through refined processing in multiple stages, achieves deep integration and accurate comparison of all-dimensional state change data and intent elements. The resulting logical structure optimization vector can comprehensively and accurately reflect the direction and degree of optimization required by the control network, providing a precise and reliable basis for the subsequent topology connection and node parameter adjustment of the initial control network, and ensuring the pertinence and effectiveness of the adaptive control network construction.
[0058] S5. Based on the logical structure optimization vector, the topology connections and node parameters in the initial control network are coordinated and adjusted to obtain the adaptive control network of the target object. In this embodiment of the invention, the step of coordinating the adjustment of the topology connections and node parameters in the initial control network based on the logical structure optimization vector to obtain the adaptive control network of the target object includes: Based on the initial control network, the logical structure optimization vector is reconstructed to obtain the topology adjustment instruction and node parameter adjustment instruction of the target object; Based on the topology adjustment command, the graph structure of the topology connection in the initial control network is edited to obtain the optimized topology connection of the target object; Based on the node parameter adjustment instructions, the node parameters in the initial control network are collaboratively optimized to obtain the optimized node parameters of the target object. Based on the optimized topology and the optimized node parameters, construct the optimized control network for the target object; The performance of the optimized control network is evaluated to obtain the adaptive control network of the target object.
[0059] The step of reconstructing the logical structure optimization vector based on the initial control network to obtain the topology adjustment instructions and node parameter adjustment instructions for the target object includes: The optimized logical structure vector is mapped to the initial control network to obtain the associated components of the target object; The associated components are classified by intent to obtain the structural class components and parametric class components of the target object; The structural components and parameter components are normalized and decoded to obtain the initial topology adjustment instruction and the initial node parameter adjustment instruction of the target object; The logical correlation between the initial topology adjustment instruction and the initial node parameter adjustment instruction is verified to obtain the topology adjustment instruction and node parameter adjustment instruction of the target object.
[0060] The logical structure optimization vector is fully mapped to the topology, node distribution, and edge relationships of the initial control network. The specific components of the initial control network pointed to by each dimension in the optimization vector are clarified, including the topological connection positions that need to be adjusted and the types of node parameters to be optimized. These optimization vector parts that correspond to the initial control network are called the correlation components.
[0061] Based on the nature of the adjustment requirements carried by the associated components, all associated components are classified and divided. Components that point to the adjustment of the connection relationship between nodes in the initial control network, the change of connection path, or the increase or decrease of the number of connections are classified as structural components, while components that point to the correction of the node's own attribute values, functional thresholds, or operating parameters are classified as parameter components.
[0062] According to the preset instruction encoding standard, the structural components are decoded to transform the topology adjustment requirements contained therein into specific, clear and executable operation instructions, forming the initial topology adjustment instructions; at the same time, the parameter components are decoded according to the same encoding standard to transform the node parameter optimization requirements into specific parameter adjustment operation instructions, forming the initial node parameter adjustment instructions.
[0063] A comprehensive examination is conducted on the internal logic of the initial topology adjustment instructions, the initial node parameter adjustment instructions, and the logical relationships between the two types of instructions. This is to determine whether there are any instruction conflicts, operational contradictions, or contradictory execution orders. Conflicting instructions are eliminated, and ambiguous instructions are corrected. This ensures that all instructions are logically compatible and can work synergistically on the initial control network. After verification, the final topology adjustment instructions and node parameter adjustment instructions are determined.
[0064] Strictly following the specific requirements of the topology adjustment instructions, the graph structure of the initial control network is edited. New edges are established between specified nodes according to the "add connection" requirement in the instructions, invalid or redundant edges are removed according to the "delete connection" requirement, and the association between nodes is reconstructed according to the "adjust connection path" requirement. This ensures that all adjustments to the topology connections accurately match the requirements of the instructions, ultimately forming an optimized topology connection.
[0065] Based on the node parameter adjustment instructions, and combined with the functional positioning of each node in the initial control network and its association logic with other nodes, the node parameters are adjusted and optimized one by one. When adjusting the parameters of a single node, its impact on the surrounding nodes and the operation of the entire network is fully considered to ensure that the adjustments of multiple node parameters cooperate and adapt to each other, so that all node parameters reach the optimal adaptation state and the optimized node parameters are obtained.
[0066] The optimized topology connections are used as new network edges. The optimized node parameters are assigned to the corresponding network nodes one by one. According to the logical relationship between nodes and edges, all nodes and edge systems are integrated into a complete network structure. This network structure fully inherits the core functional framework of the initial control network and improves its adaptability through optimization and adjustment. This is the optimized control network.
[0067] The simulated optimization control network performs control tasks in the actual operating scenario of the target object, continuously monitors the network's response speed, control command execution accuracy, stability during operation, and adaptability to the dynamic needs of the target object, and checks for problems such as logical conflicts, operational lag, or control failure. If all monitoring indicators meet the preset performance standards, the optimized control network is an adaptive control network.
[0068] The beneficial effects are that the implementation process achieves the precise conversion of the logical structure optimization vector into specific adjustment instructions through instruction reconstruction. Then, through the coordinated adjustment of topology connections and node parameters and performance evaluation, a highly adaptable and stable adaptive control network is constructed, ensuring that the control network can dynamically respond to changes in the needs of the target object, thereby improving the flexibility and reliability of digital control.
[0069] S6. Encapsulate and encode the adaptive control network to obtain the final control command for the target object.
[0070] In this embodiment of the invention, the step of encapsulating and encoding the adaptive control network to obtain the final control command for the target object includes: The adaptive control network is traversed topologically to obtain the core topological connections and active node parameters of the target object; The core topology connection is jointly bound with the active node parameters to obtain the structured network object of the target object; Convert the structured network object into an intermediate instruction representation of the target object; The intermediate instruction representation is logically reconstructed to obtain the final control instruction for the target object.
[0071] By sequentially traversing the entire network topology according to the association hierarchy and connection path order of nodes in the adaptive control network, the key connection paths that play a crucial supporting role in the implementation of control functions are identified. These paths are the core topology connections. At the same time, nodes that continuously play a role in the current control scenario and affect the control effect, along with their corresponding parameters, are selected. These parameters are the active node parameters.
[0072] The core topology connections include the node association methods and connection logic, which are associated one-to-one with the corresponding active node parameters. Following the correspondence pattern of "connection relationship - node parameter", all relevant information is integrated into a set with a fixed data structure. This set fully contains the key information required for the adaptive control network to realize the core control function, which is the structured network object.
[0073] Based on the preset instruction conversion rules, the control logic corresponding to the core topology connection in the structured network object is converted into specific operation instruction descriptions, and the active node parameters are converted into parameter configuration content in the instructions. These scattered instruction descriptions and parameter configurations are integrated according to the internal logic of the structured network object to form an intermediate instruction expression that can initially reflect the control requirements.
[0074] A comprehensive review of each instruction fragment in the intermediate instruction representation is conducted to clarify the execution priority and logical dependencies of each fragment. Duplicate or redundant instruction content is eliminated, and necessary logical connection information is added during instruction execution. This ensures that all instruction fragments are arranged in a reasonable execution order, forming a logically rigorous, complete, and directly recognizable final control instruction that can be executed by the target object.
[0075] The beneficial effects are that the implementation process accurately extracts core control information through topology traversal, and gradually forms the final control command through joint binding, instruction conversion and logic reconstruction. This ensures that the final control command can fully carry the optimized control logic of the adaptive control network, has strong executability and accuracy, provides direct and effective digital control basis for the target object, and ensures the closed-loop implementation of the entire digital control process.
[0076] like Figure 2 The diagram shown is a functional block diagram of a digital control system provided in an embodiment of the present invention.
[0077] The digital control system 100 described in this invention can be installed in an electronic device. Depending on the functions implemented, the digital control system 100 may include an intent deconstruction module 101, an initial network construction module 102, a state transition data collection module 103, an optimization vector generation module 104, an initial network optimization module 105, and a final instruction generation module 106. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.
[0078] In this embodiment, the functions of each module / unit are as follows: The intent deconstruction module 101 is used to acquire the real-time dynamic demand information flow of the target object, and to perform online intent deconstruction on the real-time dynamic demand information flow to obtain the intent elements of the target object. The initial network construction module 102 is used to construct the topology of the intent elements to obtain the initial control network of the target object. The state transition data collection module 103 is used to apply the initial control network and synchronously collect the full-dimensional state transition data of the target object. The optimization vector generation module 104 is used to perform consistency calculations on the full-dimensional state transition data and the intent elements to obtain the logical structure optimization vector of the target object. The initial network optimization module 105 is used to coordinately adjust the topology connections and node parameters in the initial control network based on the logical structure optimization vector to obtain the adaptive control network of the target object. The final instruction generation module 106 is used to encapsulate and encode the adaptive control network to obtain the final control instruction of the target object.
[0079] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.
[0080] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.
[0081] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.
[0082] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0083] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.
[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims
1. A digital control method, characterized in that, The method includes: S1. Obtain the real-time dynamic demand information flow of the target object, and perform online intent deconstruction on the real-time dynamic demand information flow to obtain the intent elements of the target object; S2. Perform topology construction on the intent elements to obtain the initial control network of the target object; S3. Apply the initial control network and simultaneously collect the full-dimensional state transition data of the target object; S4. Perform consistency calculation on the full-dimensional state transition data and the intent elements to obtain the logical structure optimization vector of the target object; S5. Based on the logical structure optimization vector, the topology connections and node parameters in the initial control network are coordinated and adjusted to obtain the adaptive control network of the target object. S6. Encapsulate and encode the adaptive control network to obtain the final control command for the target object.
2. The digital control method as described in claim 1, characterized in that, The process of acquiring the real-time dynamic demand information flow of the target object and performing online intent deconstruction on the real-time dynamic demand information flow to obtain the intent elements of the target object includes: Collect real-time dynamic demand information flow of the target object; Multimodal parsing is performed on the real-time dynamic demand information stream to obtain the time-series data stream, event signal stream, and natural language text stream of the target object; Syntactic dependency analysis is performed on the natural language text stream to obtain a preliminary semantic framework of the target object; By context binding the time-series data stream and the event signal stream, composite information of the target object is obtained; Cluster analysis is performed on the composite information to obtain the potential intent fragments of the target object; By removing incomplete information segments and logically contradictory segments from the potential intent segments, the intent elements of the target object are obtained.
3. The digital control method as described in claim 1, characterized in that, The step of constructing a topology for the intent elements to obtain the initial control network for the target object includes: Logical correlation analysis is performed on the intent elements to obtain the logical correlation relationships of the target object; The logical relationships are quantitatively evaluated to obtain the logical relationship strength of the target object; Using the intent elements as nodes and the logical relationships as edges, and assigning confidence weights to the edges based on the strength of the logical relationships, an initial logical topology structure for the target object is constructed. The initial logical topology is subjected to stability analysis, and the analyzed logical topology is formatted and encapsulated to obtain the initial control network of the target object.
4. The digital control method as described in claim 1, characterized in that, The application of the initial control network and the synchronous acquisition of full-dimensional state transition data of the target object include: The initial control network is compiled into executable logic to obtain the control instruction stream of the target object; Based on the control instruction flow, determine the requirement descriptor of the target object; Based on the demand descriptor, the original state data stream of the target object is parsed to obtain the standard state data stream of the target object; The standard state data stream is time-aligned to obtain a time-series data segment of the target object; Pattern matching is performed on the time-series data segments to obtain the abnormal events of the target object; The abnormal events are fused and encapsulated to obtain full-dimensional state transition data of the target object.
5. The digital control method as described in claim 1, characterized in that, The step of performing consistency calculations on the full-dimensional state transition data and the intent elements to obtain the logical structure optimization vector of the target object includes: Identify outlier data points in the full-dimensional state transition data to obtain clean state transition data for the target object; The clean state transition data is parsed using a multi-source format, and the parsed state transition data is normalized to obtain the regularized state vector of the target object. The intent elements are vectorized and encoded to obtain the intent feature vector of the target object; By temporally binding the regularized state vector and the intention feature vector, the temporal mapping relationship of the target object is obtained; Based on the temporal mapping relationship, linear interpolation is performed on the regularized state vector and the intention feature vector to obtain the spatiotemporal aligned dataset of the target object; The intention-state data pairs in the spatiotemporal aligned dataset are compared for discrepancies to obtain a consistency deviation measure for the intention-state data pairs. Tensor synthesis is performed on the consistency deviation metric to obtain the logical structure optimization vector of the target object.
6. The digital control method as described in claim 5, characterized in that, The formula for calculating the consistency deviation metric is as follows: ; in, This represents the consistency deviation measure. Indicates the first The stated intention feature vector Indicates the first The regularized state vectors mentioned above. The Euclidean norm of a vector. Represents absolute value. This represents the preset smallest positive number. This represents the preset time decay coefficient. Indicates the first The absolute time difference of the stated intent-state data pairs This represents the preset sensitivity coefficient to state fluctuations. Indicates the first The variance of each of the stated intent-state data pairs This represents an exponential function.
7. The digital control method as described in claim 1, characterized in that, The step of coordinating the adjustment of the topology connections and node parameters in the initial control network based on the logical structure optimization vector to obtain the adaptive control network of the target object includes: Based on the initial control network, the logical structure optimization vector is reconstructed to obtain the topology adjustment instruction and node parameter adjustment instruction of the target object; Based on the topology adjustment command, the graph structure of the topology connection in the initial control network is edited to obtain the optimized topology connection of the target object; Based on the node parameter adjustment instructions, the node parameters in the initial control network are collaboratively optimized to obtain the optimized node parameters of the target object. Based on the optimized topology and the optimized node parameters, construct the optimized control network for the target object; The performance of the optimized control network is evaluated to obtain the adaptive control network of the target object.
8. The digital control method as described in claim 7, characterized in that, The step of reconstructing the logical structure optimization vector based on the initial control network to obtain the topology adjustment instructions and node parameter adjustment instructions for the target object includes: The optimized logical structure vector is mapped to the initial control network to obtain the associated components of the target object; The associated components are classified by intent to obtain the structural class components and parametric class components of the target object; The structural components and parameter components are normalized and decoded to obtain the initial topology adjustment instruction and the initial node parameter adjustment instruction of the target object; The logical correlation between the initial topology adjustment instruction and the initial node parameter adjustment instruction is verified to obtain the topology adjustment instruction and node parameter adjustment instruction of the target object.
9. The digital control method as described in claim 1, characterized in that, The process of encapsulating and encoding the adaptive control network to obtain the final control command for the target object includes: The adaptive control network is traversed topologically to obtain the core topological connections and active node parameters of the target object; The core topology connection is jointly bound with the active node parameters to obtain the structured network object of the target object; Convert the structured network object into an intermediate instruction representation of the target object; The intermediate instruction representation is logically reconstructed to obtain the final control instruction for the target object.
10. A digital control system, characterized in that, The system for implementing the digital control method according to claim 1 includes: The intent deconstruction module is used to acquire the real-time dynamic demand information flow of the target object and perform online intent deconstruction on the real-time dynamic demand information flow to obtain the intent elements of the target object; An initial network construction module is used to construct the topology of the intent elements to obtain the initial control network of the target object; The state transition data collection module is used to apply the initial control network and synchronously collect full-dimensional state transition data of the target object; The optimized vector generation module is used to perform consistency calculations on the full-dimensional state transition data and the intent elements to obtain the logical structure optimized vector of the target object. An initial network optimization module is used to coordinately adjust the topology connections and node parameters in the initial control network based on the logical structure optimization vector to obtain the adaptive control network of the target object. The final instruction generation module is used to encapsulate and encode the adaptive control network to obtain the final control instructions for the target object.