Intelligent session state management method and system

Abstract

The invention discloses an intelligent session state management method and system, and the method comprises the following steps: constructing a rule graph structure based on a Rete algorithm, and achieving the efficient matching of context conditions; collecting real-time operation data of communication nodes, generating an initial state matching result, and constructing a multi-dimensional state vector set; inputting the vector into a Hopfield neural network, establishing an energy function model, and executing iterative calculation to realize state convergence; comparing the initial state with the convergence state, calculating the structure consistency, and feeding back a matching threshold value in the adjustment rule graph; and when the consistency meets a set condition, outputting a final stable session state set. According to the method, an energy optimization mechanism of rule driving and a neural network is fused, and intelligent session state management in a complex environment is supported.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a method and system for intelligent session state management. Background Technology

[0002] With the rapid development of complex communication scenarios such as edge computing, smart terminal collaboration, and vehicle-to-everything (V2X) communication, multi-node session state management in communication systems has become a core technical requirement for ensuring connection stability and service continuity. Currently, session state management generally relies on logical judgment methods based on static rule tables or finite state machines, identifying and transitioning node states through preset conditions, and combining this with a centralized controller for state synchronization and coordination.

[0003] Existing session state management technologies have significant shortcomings when dealing with complex and dynamically changing communication environments. On the one hand, traditional rule-based state judgment lacks context-adaptive capabilities and cannot dynamically adjust matching logic based on real-time node data and environmental characteristics, easily leading to state recognition delays, mismatches, or improper resource scheduling. On the other hand, state synchronization relies on centralized servers or master nodes for state coordination, making it difficult to effectively support autonomous collaboration between nodes in distributed networks, and prone to synchronization lag, state drift, or consistency failures. Existing methods lack effective intelligent convergence mechanisms, making it difficult to automatically repair and reconstruct states in abnormal situations such as missing state information, communication anomalies, or node disconnection, resulting in poor robustness of the overall communication system.

[0004] Therefore, how to provide a method and system for intelligent session state management is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a method and system for intelligent session state management. This invention combines the Rete algorithm to construct an efficient rule matching graph structure and uses the Hopfield neural network to establish a state energy function model. Through context data matching, state vector construction, energy convergence evolution and feedback regulation, it realizes dynamic identification, adaptive adjustment and consistency maintenance of multi-node session states in communication systems. It has the advantages of stable structure, strong self-repair capability and high state synchronization accuracy.

[0006] According to an embodiment of the present invention, a method and system for intelligent session state management are provided, the method comprising the following steps:

[0007] Construct a rule graph structure, define the mapping relationship between rule nodes and condition nodes based on the Rete algorithm, and form an executable rule matching graph;

[0008] The system acquires real-time context data from each node in the communication system, inputs the real-time context data into the rule matching graph, performs condition traversal and matching operations, and outputs the initial state matching result set.

[0009] Extract matching state labels and corresponding context feature parameters from the initial state matching result set, and construct a multi-dimensional state vector set by combining the operation indicators of the communication nodes;

[0010] Initialize the Hopfield neural network by taking a multidimensional set of state vectors as input, configuring the connection weight matrix and bias vector, and establishing an energy function model based on the Hopfield neural network;

[0011] Perform iterative calculations of the energy function model, update the multidimensional state vector set in each iteration, and record the energy descent trajectory until the state converges;

[0012] Obtain the converged state output vector, perform a corresponding mapping comparison with the initial state matching result set, and calculate the structural consistency identifier and difference index;

[0013] Based on the difference index, the rule matching rule thresholds in the graph are adjusted, and the final stable session state set is output after the structural consistency identifier meets the conditions.

[0014] Optionally, forming an executable rule matching graph includes:

[0015] Parse all session rules to be processed, parse the antecedent of each session rule into an ordered set of atomic conditions, define each atomic condition in the set of atomic conditions as a condition node, and define all session rules as a rule node;

[0016] Define the mapping relationship between rule nodes and condition nodes. The mapping relationship is represented by establishing directed dependency edges from the condition node to the corresponding rule node.

[0017] Based on the above mapping relationship, a rule graph structure is constructed. The rule graph structure is a directed graph, which includes a condition node layer, a rule node layer and a set of directed edges between them. Each path in the graph represents the logical condition flow of a rule.

[0018] Based on the rule graph structure, an α network is generated, where each α network node corresponds to a condition node. A condition matching function is configured to receive context data input from the communication system and perform matching judgments, outputting matching data objects that meet the conditions. The outputs of all α network nodes are aggregated to form a condition matching result set.

[0019] Using the condition matching result set as input, conflict prediction pre-stratification is performed, which specifically includes constructing a trigger frequency matrix between condition nodes, performing cluster analysis on the trigger frequency matrix to obtain a cluster set of condition nodes, setting node hierarchical indexes based on the cluster structure, and generating a hierarchical connection table.

[0020] A β network is constructed based on a hierarchical connection table. Each β network node performs a condition combination operation. The left input of a β network node is the partial matching path result output by the previous layer β network node, and the right input is the condition matching result output by the current layer α network node. When the combination condition is satisfied, the β network node outputs a new intermediate matching path result.

[0021] Configure the α network node with the α network's memory unit to store the original conditions of the current successful match, and configure the β network node with the β network's memory unit to cache intermediate matching path results, thereby enabling incremental processing of changing facts;

[0022] At the end of each path in the rule graph structure, a terminal rule node is established. The terminal node receives the complete matching path from the output of the last layer of the β network, identifies the corresponding rule node as activated, and outputs the final executable rule matching graph.

[0023] Optionally, generating the initial state matching result set includes:

[0024] Collect real-time context data of each node in the communication system. The real-time context data includes communication latency, packet loss rate, bandwidth utilization, buffer occupancy rate, node connection status, data packet timestamps, and round-trip latency.

[0025] The collected real-time context data is preprocessed according to the node dimension to establish a context data structure;

[0026] The context data structure is input into the α network in the rule matching graph. Each α network node corresponds to a condition node. The α network node receives the context data structure and performs condition traversal operation according to the bound condition matching function. The condition traversal operation includes reading the field values ​​in the context data structure, performing range comparison and status flag matching on the field values.

[0027] The output of the α network node that meets the matching rule is marked as a successful match, and the condition node of the successful match and its corresponding data record are formed into a condition-data pair;

[0028] All condition-data pairs are categorized according to the mapping relationship between condition nodes and rule nodes defined in the rule graph structure, and then aggregated into their respective rule paths;

[0029] Count the number of successfully matched condition nodes on each rule path, and mark the rule path with at least one successfully matched condition node as the initial matching path;

[0030] Based on the matching results of all initial matching paths, an initial state matching result set is generated, which includes rule identifiers, condition hit status, and corresponding context data.

[0031] Optionally, constructing a multidimensional set of state vectors includes:

[0032] Read the matching records of each rule path from the initial state matching result set, identify the successfully matched condition nodes, and extract the context data fields associated with the condition nodes. The context data fields include communication delay, packet loss rate, bandwidth utilization, cache utilization, connection stability, data packet transmission interval, and round-trip delay.

[0033] Based on the matching records of the condition nodes, a matching status label is generated. Each matching status label contains the condition node identifier, the matching condition type, and the matching time index.

[0034] For each matching state label, extract the parameter values ​​from the corresponding context data field, perform normalization and interval encoding operations on the extracted parameter values ​​to obtain the parameter standardization results, and combine them in the order of the fields to form a single-condition node state feature vector;

[0035] The state feature vectors of all condition nodes under the same rule path are concatenated in the matching order to generate the initial state vector at the rule path level.

[0036] The system collects operational metrics for each node in the communication system, including CPU utilization, memory usage, cache hit rate, interface call frequency, and connection persistence time. These metrics are then uniformly quantified and standardized to obtain a node operational metric vector.

[0037] Based on the mapping relationship between rule nodes and condition nodes, the node's operation indicators are matched and concatenated with the initial state vector of the corresponding rule path level to form a node-level composite state vector.

[0038] Arrange all node-level composite state vectors in order of communication priority to construct a multi-node state vector matrix;

[0039] The multi-node state vector matrix is ​​formatted and encoded to output a set of multi-dimensional state vectors with a unified structure.

[0040] Optionally, establishing an energy function model includes:

[0041] The constructed multidimensional state vector set is input into the Hopfield neural network. Each dimension of the multidimensional state vector set corresponds to a neuron node, which is used to represent the structured features of the current session state.

[0042] Construct a Hopfield neural network structure, create a neuron node for each multidimensional state vector dimension in the Hopfield neural network, and establish bidirectional symmetric connections between all nodes to form a symmetric fully connected topology;

[0043] Based on the historical joint distribution of each dimension in the multidimensional state vector, calculate the connection weight value between each pair of neurons and generate a connection weight matrix.

[0044] By combining the activation distribution of each neuron node in the historical state evolution, the bias value of each node is calculated, and a bias vector is constructed.

[0045] Each value in the input multidimensional state vector is assigned to the corresponding neuron node as the initial state, and normalization is performed on all state values ​​to generate a standardized initial network state.

[0046] Based on the established connection weight matrix, bias vector, and standardized initial network state, an energy function model is constructed.

[0047] Optionally, performing iterative calculations of the energy function model until state convergence includes:

[0048] Load the constructed energy function model, take the standardized initial network state as input, and initialize the iteration parameters of the Hopfield neural network, including the maximum number of iterations, energy change threshold, and state stability determination condition;

[0049] Extract the current state vector of all neurons from the standardized initial network state, establish an iterative loop structure, and select neurons sequentially according to the asynchronous update strategy in each iteration.

[0050] For the selected neuron node, its local input signal is calculated. The local input signal is determined by the current state of other neurons connected to the corresponding neuron node, the corresponding connection weights, and the bias value of the corresponding neuron node.

[0051] The state values ​​of neuron nodes are updated based on the calculated local input signals, and the updated state is written into the current multidimensional state vector set. Unselected neuron nodes retain their original state, forming a network state vector.

[0052] Once all neurons have completed their state updates in an iteration, the energy function model is called to calculate the overall energy value of the network. The difference between the new overall network energy value and the previous overall network energy value is compared, and the energy change results are recorded.

[0053] Store the overall network energy value, the overall network energy difference, and the corresponding multidimensional state vector after each iteration into the energy descent trajectory table;

[0054] Repeat the process of updating neuron states and calculating energy values ​​until the energy change is lower than a preset threshold, at which point the network is considered to have reached convergence.

[0055] When the convergence condition is met, the network state vector of the last iteration in the energy descent trajectory table is extracted to generate the converged state output vector.

[0056] Optionally, the calculation of structural consistency indicators and difference metrics includes:

[0057] Extract the rule path records from the initial state matching result set generated by the rule matching graph, obtain the successfully matched condition nodes and their corresponding context data fields in each rule path, and form a rule path field set;

[0058] Extract the final state output vector from the state output vector of the energy function model, analyze the index order and field meaning of each dimension in the state output vector, and establish a correspondence table between dimension index and field label;

[0059] Based on the correspondence table between dimension index and field label, establish the field mapping relationship between the initial state matching result set and the state output vector, and establish a one-to-one mapping pair between each rule path field and the corresponding dimension in the state output vector;

[0060] After the mapping relationship is established, the corresponding data value of each mapping pair is read in sequence. Numerical difference calculation is performed on continuous fields, state consistency judgment is performed on logical fields, and encoding matching verification is performed on symbolic fields. The comparison result of each field is recorded as a matching mark.

[0061] Statistically analyze all matching tags in each rule path, calculate the ratio of the number of successfully matched fields to the total number of fields, and form a structural consistency identifier for the rule path.

[0062] During the comparison process, all fields that are judged to be mismatched are extracted, their numerical offsets are calculated, and the results are aggregated to form a difference index.

[0063] Optional, the generation of the final stable set of session states includes:

[0064] Receive the structural consistency identifier and determine whether the stability threshold condition is met. The stability threshold condition is that the proportion of the number of matching fields in the structural consistency identifier is not less than a preset proportion threshold.

[0065] When the structural consistency identifier is less than the preset ratio threshold, all mismatched fields recorded in the difference index are extracted, and the corresponding condition node and rule path in the rule matching graph are located in combination with the field type.

[0066] Based on the field difference type, the rule threshold adjustment operation is performed, specifically including: continuous fields are adjusted to narrow the numerical tolerance range of condition matching, logical fields are adjusted to improve the sensitivity of state judgment, and symbolic fields are adjusted to update the symbol matching function in the condition node;

[0067] Write the adjusted rule threshold into the corresponding condition node to complete the local update of the target rule path in the rule matching graph;

[0068] Re-execute the matching process based on the updated rule matching graph, generate a new initial state matching result set, and recalculate the structural consistency identifier;

[0069] When the recalculated consistency identifier is greater than or equal to the stability threshold condition, the rule path, context data field and final state output vector corresponding to the rule matching structure are extracted and combined to form a session state structure item.

[0070] Summarize all session state structure items that meet the consistency threshold condition, generate the final stable session state set, and output it.

[0071] Optionally, the system includes the following modules:

[0072] The rule matching graph construction module is used to parse session rules and generate a rule graph structure containing the mapping relationship between rule nodes and condition nodes, forming an executable rule matching graph;

[0073] The context acquisition and preprocessing module is used to acquire real-time context data such as communication latency, packet loss rate, bandwidth utilization, and buffer occupancy rate from the communication system, and to establish a context data structure by node dimension.

[0074] The rule matching execution module is used to input the context data structure into the rule matching graph, perform condition matching and path construction through α network nodes and β network nodes, and generate an initial state matching result set;

[0075] The state vector construction module is used to extract matching state labels and context feature parameters based on the initial state matching result set, and generate a multi-dimensional state vector set by combining node operation indicators.

[0076] The energy model construction module is used to establish the connection weight matrix and bias vector based on the Hopfield neural network, construct the energy function model, and initialize the neural network structure.

[0077] The energy convergence calculation module is used to perform iterative calculations of the energy function model, update the neuron state using an asynchronous update strategy, and generate the final convergence state vector and energy descent trajectory.

[0078] The consistency comparison module is used to perform field-level mapping and comparison between the converged state vector and the initial state matching result set, and output structural consistency identifiers and difference indicators.

[0079] The rule feedback adjustment module is used to adjust the matching threshold parameters of relevant rule paths in the rule matching graph based on the difference indicators.

[0080] The session state output module is used to output the final stable set of session states when the structural consistency identifier meets the preset conditions.

[0081] The beneficial effects of this invention are:

[0082] (1) By constructing a rule matching graph structure based on the Rete algorithm and combining conditional traversal of context data with hierarchical clustering optimization, this invention achieves efficient session state recognition under complex communication conditions, significantly improving the accuracy and execution efficiency of state recognition in multi-node environments.

[0083] (2) This invention introduces the Hopfield neural network into the session state management process. By constructing the connection weight matrix and bias vector, an energy function model is established, which supports state convergence and recovery in the case of communication node abnormality or state loss, thereby enhancing the system's self-repair capability and state stability.

[0084] (3) This invention combines the structural consistency comparison mechanism between the initial state and the converged state with the difference index feedback adjustment of the matching threshold in the matching rule graph, realizing the bidirectional collaboration between rule logic and state evolution, and has the ability to dynamically adjust and adaptively optimize, effectively improving the robustness and scalability of the communication system under high concurrency and weak connection conditions. Attached Figure Description

[0085] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0086] Figure 1 The flowchart below shows a method and system for intelligent session state management proposed in this invention.

[0087] Figure 2This is a flowchart illustrating the process of constructing a rule matching graph structure based on the Rete algorithm. Detailed Implementation

[0088] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0089] refer to Figures 1-2 A method and system for intelligent session state management, the method comprising the following steps:

[0090] Construct a rule graph structure, define the mapping relationship between rule nodes and condition nodes based on the Rete algorithm, and form an executable rule matching graph;

[0091] The system acquires real-time context data from each node in the communication system, inputs the real-time context data into the rule matching graph, performs condition traversal and matching operations, and outputs the initial state matching result set.

[0092] Extract matching state labels and corresponding context feature parameters from the initial state matching result set, and construct a multi-dimensional state vector set by combining the operation indicators of the communication nodes;

[0093] Initialize the Hopfield neural network by taking a multidimensional set of state vectors as input, configuring the connection weight matrix and bias vector, and establishing an energy function model based on the Hopfield neural network;

[0094] Perform iterative calculations of the energy function model, update the multidimensional state vector set in each iteration, and record the energy descent trajectory until the state converges;

[0095] Obtain the converged state output vector, perform a corresponding mapping comparison with the initial state matching result set, and calculate the structural consistency identifier and difference index;

[0096] Based on the difference index, the rule matching rule thresholds in the graph are adjusted, and the final stable session state set is output after the structural consistency identifier meets the conditions.

[0097] In this embodiment, forming an executable rule matching graph includes:

[0098] Parse all session rules to be processed, parse the antecedent of each session rule into an ordered set of atomic conditions, define each atomic condition in the set of atomic conditions as a condition node, and define all session rules as a rule node;

[0099] Define the mapping relationship between rule nodes and condition nodes. The mapping relationship is represented by a directed dependency edge established from the condition node to the corresponding rule node, indicating that the rule is valid only if the set of conditions is met simultaneously.

[0100] Based on the above mapping relationship, a rule graph structure is constructed. The rule graph structure is a directed graph, which includes a condition node layer, a rule node layer and a set of directed edges between them. Each path in the graph represents the logical condition flow of a rule.

[0101] Based on the rule graph structure, an α network is generated, where each α network node corresponds to a condition node. A condition matching function is configured to receive context data input from the communication system and perform matching judgments, outputting matching data objects that meet the conditions. The outputs of all α network nodes are aggregated to form a condition matching result set.

[0102] Using the condition matching result set as input, conflict prediction pre-stratification is performed, which specifically includes constructing a trigger frequency matrix between condition nodes, performing cluster analysis on the trigger frequency matrix to obtain a cluster set of condition nodes, setting node level indexes based on the cluster structure, and generating a hierarchical connection table to guide the legal path range of node combinations in the construction of the β network.

[0103] A β network is constructed based on a hierarchical connection table. Each β network node performs a condition combination operation. The left input of a β network node is the partial matching path result output by the previous layer β network node, and the right input is the condition matching result output by the current layer α network node. When the combination condition is satisfied, the β network node outputs a new intermediate matching path result and passes it to the next layer β network node.

[0104] Configure the α network node with the α network's memory unit to store the original conditions of the current successful match, and configure the β network node with the β network's memory unit to cache intermediate matching path results, thereby enabling incremental processing of changing facts;

[0105] At the end of each path in the rule graph structure, a terminal rule node is established. The terminal node receives the complete matching path from the output of the last layer of the β network, identifies the corresponding rule node as activated, and outputs the final executable rule matching graph.

[0106] The α network node layer, hierarchical index structure, β network structure, terminal rule node layer and their connection relationships together constitute the final executable rule matching graph. This matching graph takes communication context data as input, completes state path parsing, rule determination and outputs structured state matching results.

[0107] In this embodiment, generating the initial state matching result set includes:

[0108] Collect real-time context data of each node in the communication system. The real-time context data includes communication latency, packet loss rate, bandwidth utilization, buffer occupancy rate, node connection status, data packet timestamps, and round-trip latency.

[0109] The collected real-time context data is preprocessed according to the node dimension to establish a context data structure. Each context data structure corresponds to the state description of a node in the current communication cycle.

[0110] The context data structure is input into the α network in the rule matching graph. Each α network node corresponds to a condition node. The α network node receives the context data structure and performs condition traversal operation according to the bound condition matching function. The condition traversal operation includes reading the field values ​​in the context data structure, performing range comparison and status flag matching on the field values, and determining whether the data meets the matching rules of the corresponding condition node.

[0111] Reading field values ​​from the context data structure and performing condition matching operations includes the following implementation methods:

[0112] The system first maps the fields of the real-time context data uploaded by each node according to a preset communication structure template. Each context data structure contains multiple fields, such as communication delay, packet loss rate, bandwidth utilization, connection status identifier, cache utilization, and packet arrival timestamp. The α network node calls the condition field bound to itself from this structured data. For example, the field "delay" for α network node A1 is defined as "delay between 10ms and 100ms". The system extracts the current value of this field and performs an interval comparison. If the value falls within the specified range, the match is successful and marked as "satisfied"; otherwise, it is marked as "not satisfied". For logical fields (such as connection status "established"), the matching function is a status flag comparison. The system determines whether the field value is consistent with the preset status. All matching judgments are executed in parallel on the local node. After execution, the matching result of each condition node is output.

[0113] The output of the α network node that meets the matching rule is marked as a successful match, and the condition node of the successful match and its corresponding data record are formed into a condition-data pair;

[0114] All condition-data pairs are categorized according to the mapping relationship between condition nodes and rule nodes defined in the rule graph structure, and then aggregated into their respective rule paths;

[0115] Count the number of successfully matched condition nodes on each rule path, and mark the rule path with at least one successfully matched condition node as the initial matching path;

[0116] Based on the matching results of all initial matching paths, an initial state matching result set is generated, which includes rule identifiers, condition hit status, and corresponding context data.

[0117] In this embodiment, constructing a multidimensional state vector set includes:

[0118] Read the matching records of each rule path from the initial state matching result set, identify the successfully matched condition nodes, and extract the context data fields associated with the condition nodes. The context data fields include communication delay, packet loss rate, bandwidth utilization, cache utilization, connection stability, data packet transmission interval, and round-trip delay.

[0119] A matching status label is generated based on the matching records of the condition nodes. Each matching status label contains the condition node identifier, the matching condition type, and the matching time index, which are used to uniquely identify the current matching status.

[0120] For each matching state label, extract the parameter values ​​from the corresponding context data field, perform normalization and interval encoding operations on the extracted parameter values ​​to obtain the parameter standardization results, and combine them in the order of the fields to form a single-condition node state feature vector;

[0121] The normalization and interval coding operations involve performing linear normalization on the extracted contextual feature parameters, such as communication latency, packet loss rate, and bandwidth utilization, based on a preset maximum and minimum value range, converting the original values ​​into small values ​​within the range of [0,1]. Subsequently, a fixed interval is set according to application requirements (such as 0–0.2, 0.2–0.4, etc.), and the normalization result is mapped to the corresponding interval label. One-hot coding is then used to convert it into a fixed-dimensional vector representation.

[0122] The state feature vectors of all condition nodes under the same rule path are concatenated in the order of matching to generate an initial state vector at the rule path level. This initial state vector records the parameter state of all matching conditions within the rule path.

[0123] The specific implementation of sequential concatenation of matching order is as follows: For multiple condition nodes that have been successfully matched in the same rule path, they are numbered according to their original arrangement order in the rule setting, for example, arranged from left to right or from top to bottom as C1, C2, C3. The system reads the state feature vector corresponding to each condition node in this order and concatenates them in the order of numbering, that is, taking the vector of C1 as the starting point, and connecting the vectors of C2 and C3 in sequence to form a continuous high-dimensional vector sequence, which serves as the complete initial state vector under the rule path.

[0124] The system collects operational metrics for each node in the communication system, including CPU utilization, memory usage, cache hit rate, interface call frequency, and connection persistence time. These metrics are then uniformly quantified and standardized to obtain a node operational metric vector.

[0125] The specific implementation of unified quantization and standardization is as follows: For each numerical data in the communication node operation index, linear scaling is performed according to the preset business range threshold, and the original values ​​are uniformly mapped to the [0,1] interval; for classification or status indicators, a fixed discrete encoding method is used for quantization, for example, normal, overload, and abnormal are encoded as [1,0,0], [0,1,0], and [0,0,1], respectively; all standardized and quantized results are arranged in a fixed order of fields to form a consistent operation index vector.

[0126] Based on the mapping relationship between rule nodes and condition nodes, the node's operation indicators are matched and concatenated with the initial state vector of the corresponding rule path level to form a node-level composite state vector.

[0127] Arrange all node-level composite state vectors in order of communication priority to construct a multi-node state vector matrix;

[0128] The specific implementation of the communication priority order is as follows: The system assigns communication priority identifiers according to the role weight of each communication node in the session, data forwarding frequency, or historical communication importance. For example, the priority of the master node is set to P1, the key relay node to P2, and the edge node to P3. When constructing a multi-node state vector matrix, the composite state vectors corresponding to the nodes are sorted according to the priority identifiers. The vectors of nodes with higher priority are placed in the first position, and the vectors of nodes with lower priority are arranged in order after them, forming an ordered vector matrix.

[0129] The multi-node state vector matrix is ​​formatted and encoded to output a set of multi-dimensional state vectors with a unified structure.

[0130] In this embodiment, establishing the energy function model includes:

[0131] The constructed multidimensional state vector set is input into the Hopfield neural network. Each dimension of the multidimensional state vector set corresponds to a neuron node, which is used to represent the structured features of the current session state.

[0132] Construct a Hopfield neural network structure, create a neuron node for each multidimensional state vector dimension in the Hopfield neural network, and establish bidirectional symmetric connections between all nodes to form a symmetric fully connected topology;

[0133] Based on the historical joint distribution of each dimension in the multidimensional state vector, the connection weight value between each pair of neurons is calculated to generate a connection weight matrix. The connection weight matrix is ​​then symmetrically processed to ensure that it satisfies the requirements of symmetry and no self-connection.

[0134] The historical joint distribution among dimensions in a multidimensional state vector refers to the probability relationship of any two state dimensions appearing simultaneously in the same state across multiple historical state samples. Its specific implementation includes: the system counts the number of times each pair of dimensions appears simultaneously in the same state (e.g., high or low value) in the historical convergent state vector set, calculates their co-occurrence frequency, and compares it with their respective independent activation frequencies to form a joint probability matrix; this joint distribution is used to measure the state correlation between dimensions, and thus serves as the basis for calculation when constructing the connection weight matrix, reflecting the intensity of cooperative changes between neurons.

[0135] By combining the activation distribution of each neuron node in the historical state evolution, the bias value of each node is calculated, and a bias vector is constructed.

[0136] The activation distribution in historical state evolution refers to the frequency distribution of a neuron node being activated (i.e., taking a high value) across multiple historical state vectors. Its specific implementation includes: during training or inference, the system records the number of activations of each neuron's corresponding dimension in converged states and calculates the proportion of the neuron in an activated state (e.g., greater than a set threshold) across all input samples, forming an activation probability distribution; this distribution is used to determine the bias value of the node, with a higher bias value indicating a greater tendency for activation, and is used to adjust the local stability reference in the energy function during network structure initialization;

[0137] Each value in the input multidimensional state vector is assigned to the corresponding neuron node as the initial state, and normalization is performed on all state values ​​to generate a standardized initial network state.

[0138] Based on the established connection weight matrix, bias vector, and standardized initial network state, an energy function model is constructed. The energy function is used to measure the energy level of the current neural network structure under the combination of states and serves as the basic evaluation function for subsequent state evolution and convergence determination.

[0139] In this embodiment, performing iterative calculations of the energy function model until state convergence includes:

[0140] Load the constructed energy function model, take the standardized initial network state as input, and initialize the iteration parameters of the Hopfield neural network, including the maximum number of iterations, energy change threshold, and state stability determination condition;

[0141] Extract the current state vector of all neurons from the standardized initial network state, establish an iterative loop structure, and select neurons sequentially according to the asynchronous update strategy in each iteration.

[0142] The asynchronous update strategy refers to selecting only one neuron node for state calculation and update in each iteration of the Hopfield neural network, rather than updating all nodes simultaneously. Its specific implementation includes: at the beginning of each iteration, selecting one node from all neurons in a preset order or using a pseudo-random algorithm; calculating the local input signal currently received by that node, and determining whether the node needs to flip its state based on the signal and a set threshold; immediately writing the updated state vector, and then selecting the next node to repeat the process until all nodes in the current iteration have been updated sequentially.

[0143] For the selected neuron node, its local input signal is calculated. The local input signal is determined by the current state of other neurons connected to the corresponding neuron node, the corresponding connection weights, and the bias value of the corresponding neuron node.

[0144] The state values ​​of neuron nodes are updated based on the calculated local input signals, and the updated state is written into the current multidimensional state vector set. Unselected neuron nodes retain their original state, forming a network state vector.

[0145] Once all neurons have completed their state updates in an iteration, the energy function model is called to calculate the overall energy value of the network. The difference between the new overall network energy value and the previous overall network energy value is compared, and the energy change results are recorded.

[0146] The overall network energy value, the overall network energy difference, and the corresponding multidimensional state vector after each iteration are stored in the energy descent trajectory table, which is used to record the energy evolution process and state change sequence of each iteration.

[0147] Repeat the process of updating neuron states and calculating energy values ​​until the energy change is lower than a preset threshold, at which point the network is considered to have reached convergence.

[0148] When the convergence condition is met, the network state vector of the last iteration in the energy descent trajectory table is extracted to generate the converged state output vector.

[0149] In this embodiment, the calculation of structural consistency identification and difference index includes:

[0150] Extract the rule path records from the initial state matching result set generated by the rule matching graph, obtain the successfully matched condition nodes and their corresponding context data fields in each rule path, and form a rule path field set;

[0151] Extract the final state output vector from the state output vector of the energy function model, analyze the index order and field meaning of each dimension in the state output vector, and establish a correspondence table between dimension index and field label;

[0152] Based on the correspondence table between dimension index and field label, establish the field mapping relationship between the initial state matching result set and the state output vector, and establish a one-to-one mapping pair between each rule path field and the corresponding dimension in the state output vector;

[0153] After the mapping relationship is established, the corresponding data value of each mapping pair is read in sequence. Numerical difference calculation is performed on continuous fields, state consistency judgment is performed on logical fields, and encoding matching verification is performed on symbolic fields. The comparison result of each field is recorded as a matching mark.

[0154] The specific method for performing numerical difference calculation on continuous fields is as follows: extract the numerical values ​​of the corresponding fields in the rule matching results and convergence states, calculate the absolute value of the difference between the two, and compare this value with the set tolerance threshold to determine whether there is an offset. The method for performing state consistency judgment on logical fields is as follows: compare the Boolean values ​​or multi-value states of the two fields one by one, and consider them consistent only when the states are completely identical. The method for performing encoding matching verification on symbolic fields is as follows: convert the field values ​​into discrete codes of a unified format using a preset encoding table, and compare whether the codes are equal to determine the matching status.

[0155] Statistically analyze all matching tags of each rule path, calculate the ratio of the number of successfully matched fields to the total number of fields, and form a structural consistency identifier for the rule path, which is used to reflect the degree of consistency between the convergence state and the rule matching structure.

[0156] During the comparison process, all fields that are judged to be mismatched are extracted, their numerical offsets are calculated, and the results are aggregated to form a difference index. The difference index records the degree of deviation of each field.

[0157] In this embodiment, the generation of the final stable session state set includes:

[0158] Receive the structural consistency identifier and determine whether the stability threshold condition is met. The stability threshold condition is that the proportion of the number of matching fields in the structural consistency identifier is not less than a preset proportion threshold.

[0159] When the structural consistency identifier is less than the preset ratio threshold, all mismatched fields recorded in the difference index are extracted, and the corresponding condition node and rule path in the rule matching graph are located in combination with the field type.

[0160] Based on the field difference type, the rule threshold adjustment operation is performed, specifically including: continuous fields are adjusted to narrow the numerical tolerance range of condition matching, logical fields are adjusted to improve the sensitivity of state judgment, and symbolic fields are adjusted to update the symbol matching function in the condition node;

[0161] Write the adjusted rule threshold into the corresponding condition node to complete the local update of the target rule path in the rule matching graph;

[0162] Re-execute the matching process based on the updated rule matching graph, generate a new initial state matching result set, and recalculate the structural consistency identifier;

[0163] When the recalculated consistency identifier is greater than or equal to the stability threshold condition, the rule path, context data field and final state output vector corresponding to the rule matching structure are extracted and combined to form a session state structure item.

[0164] Summarize all session state structure items that meet the consistency threshold condition, generate the final stable session state set, and output it.

[0165] In this embodiment, the system includes the following modules:

[0166] The rule matching graph construction module is used to parse session rules and generate a rule graph structure containing the mapping relationship between rule nodes and condition nodes, forming an executable rule matching graph;

[0167] The context acquisition and preprocessing module is used to acquire real-time context data such as communication latency, packet loss rate, bandwidth utilization, and cache occupancy rate from the communication system, and to establish a context data structure by node dimension.

[0168] The rule matching execution module is used to input the context data structure into the rule matching graph, perform condition matching and path construction through α network nodes and β network nodes, and generate an initial state matching result set;

[0169] The state vector construction module is used to extract matching state labels and context feature parameters based on the initial state matching result set, and generate a multi-dimensional state vector set by combining node operation indicators.

[0170] The energy model construction module is used to establish the connection weight matrix and bias vector based on the Hopfield neural network, construct the energy function model, and initialize the neural network structure.

[0171] The energy convergence calculation module is used to perform iterative calculations of the energy function model, update the neuron state using an asynchronous update strategy, and generate the final convergence state vector and energy descent trajectory.

[0172] The consistency comparison module is used to perform field-level mapping and comparison between the converged state vector and the initial state matching result set, and output structural consistency identifiers and difference indicators.

[0173] The rule feedback adjustment module is used to adjust the matching threshold parameters of relevant rule paths in the rule matching graph based on the difference indicators.

[0174] The session state output module is used to output the final stable set of session states when the structural consistency identifier meets the preset conditions.

[0175] Example 1:

[0176] To verify the feasibility of this invention in practice, this invention is applied to a multi-node collaborative communication network, demonstrating the specific implementation and technical features of the intelligent session state management method described in this invention in a complex communication environment.

[0177] In this application scenario, the communication system consists of several physical communication nodes, each responsible for tasks such as device access, state synchronization, and command scheduling. With the increase in the number of connected devices and the instability of the communication environment, frequent synchronization issues, slow anomaly recovery, and low state reconstruction efficiency arise between nodes. To address these issues, the intelligent session state management mechanism proposed in this invention is introduced to improve the robustness and adaptability of the overall communication system.

[0178] A rule-matching graph structure based on the Rete algorithm is constructed. Specifically, the state management rules in the communication system configuration are parsed, rule antecedents are decomposed into atomic condition sets, mapped to condition nodes, and the entire rule is treated as a rule node. Directed dependency edges are established to form a mapping relationship between condition nodes and rule nodes. Subsequently, α-networks and β-networks are constructed. α-network nodes correspond to condition nodes, and matching functions are configured to process real-time context data. β-network nodes are used for condition combination and path construction, ultimately forming an executable rule-matching graph.

[0179] Context data of communication nodes is collected in real time, including metrics such as communication latency, bandwidth utilization, packet loss rate, buffer occupancy, connection status, and data transmission interval, and organized into a context data structure by node dimension. This data is input into the rule matching graph, triggering α network node condition matching. Successfully matched conditions and context parameters form condition-data pairs, which are then aggregated according to rule paths to generate an initial state matching result set.

[0180] Based on the matching results, the matching state labels and their associated context fields are extracted, and standardized feature vectors are constructed through normalization and interval encoding. Multiple feature vectors under the same rule path are concatenated sequentially to form an initial state vector at the rule path level, and then combined with node operation indicators to form a composite state vector at the node level, thus constructing a multi-node state vector matrix.

[0181] The state vector matrix is ​​input into the Hopfield neural network model, with each dimension corresponding to a neuron, constructing a symmetric fully connected structure. The connection weight matrix and bias vector are calculated using the historical joint distribution to establish an energy function model. After initializing the neuron states, an asynchronous update strategy is used to iteratively update the state node by node, recording the energy change trajectory and state set after each iteration until convergence.

[0182] After the energy function converges, the system performs a field mapping comparison between the converged state output vector and the initial state matching result set. Continuous fields are judged by numerical difference, logical fields are judged by state consistency, and symbolic fields are verified by encoding matching. The proportion of matching fields in each rule path is counted to generate a structural consistency identifier, and the differences between inconsistent fields are aggregated to form a difference index.

[0183] If structural consistency is insufficient, the matching thresholds of relevant condition nodes in the rule matching graph are adjusted based on the feedback of the difference index, the rule matching conditions are updated, and the matching calculation is retried. If the consistency index meets the set threshold, the system maps the current converged state vector to a stable set of session states and synchronizes it to each node. The following data is recorded during system operation:

[0184] Table 1 Comparison of State Synchronization Efficiency

[0185] Table 2 Number of times automatic recovery from abnormal states

[0186] Table 3 Duration of State Inconsistency

[0187] Table 1 shows that the average state synchronization time of nodes using the method of this invention is 43ms, while the average time of nodes using the traditional rule table mechanism is as high as 117ms. This indicates that the combination of the Rete rule graph structure and the Hopfield energy convergence model greatly improves the efficiency of session state synchronization, resulting in faster response speed and stronger adaptability to high-concurrency scenarios. Table 2 shows the number of automatic recovery times of abnormal states. Within the same period, the number of node recovery times under the method of this invention is significantly more than that under the traditional mechanism (e.g., 7 times for node A03 and 2 times for node A03), indicating that the system has stronger fault self-recovery capability and can achieve rapid detection and repair of session anomalies.

[0188] Table 3 shows the duration of state inconsistency, which further verifies the robustness of the system of the present invention: the duration of state inconsistency of nodes using this method is controlled within 8 seconds, while most traditional mechanisms exceed 30 seconds, with the longest being 32.8 seconds. This indicates that through the dynamic adjustment and rule feedback mechanism guided by the energy function, the system can complete state reconstruction and consistency verification more quickly.

[0189] In summary, these data fully demonstrate the effectiveness of the technical solution of this invention in communication systems, especially in multi-node environments, where it significantly improves state identification, fault handling, and consistency maintenance.

[0190] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method of intelligent session state management, characterized by, Comprise: Construct a rule graph structure, define the mapping relationship between rule nodes and condition nodes based on the Rete algorithm, and form an executable rule matching graph; Obtain the real-time context data of each node in the communication system, input the real-time context data into the rule matching graph, perform condition traversal and matching operations, and output an initial state matching result set; Extract the matching state label and corresponding context feature parameters from the initial state matching result set, and construct a multi-dimensional state vector set in combination with the running indicators of the communication nodes; Initialize the Hopfield neural network, input the multi-dimensional state vector set, configure the connection weight matrix and bias vector, and establish an energy function model based on the Hopfield neural network; Perform iterative calculation of the energy function model, update the multi-dimensional state vector set in each iteration, and record the energy descent trajectory until the state converges; Obtain the converged state output vector, and perform corresponding mapping comparison with the initial state matching result set to calculate the structure consistency identifier and difference index; Adjust the related rule threshold in the rule matching graph based on the difference index, and output the final stable conversation state set when the structure consistency identifier meets the conditions.

2. The method of intelligent session state management of claim 1, wherein, Forming an executable rule matching graph includes: Parse all the conversation rules to be processed, parse the antecedent part of each conversation rule into an ordered atomic condition set, and define each atomic condition in the atomic condition set as a condition node, and define all conversation rules as a rule node; Define the mapping relationship between rule nodes and condition nodes, which is represented as a directed dependency edge from a condition node to a corresponding rule node; According to the above mapping relationship, a rule graph structure is constructed, which is a directed graph containing a condition node layer, a rule node layer and a directed edge set therebetween, and each path in the graph represents the logical condition flow of a rule; Based on the rule graph structure, generate an alpha network, where each alpha network node corresponds to a condition node, configure a condition matching function to receive context data input by the communication system and perform matching judgment, output matching data objects that meet the conditions, and the outputs of all alpha network nodes are collected to form a condition matching result set; Take the condition matching result set as input, perform conflict prediction and pre-layering, which specifically includes constructing a trigger frequency matrix between condition nodes, clustering the trigger frequency matrix to obtain a condition node clustering set, setting a node level index according to the clustering structure, and generating a hierarchical connection table; Based on the hierarchical connection table, build a beta network, each beta network node of the beta network performs condition combination operation, the left input of the beta network node is the partial matching path result output by the previous layer beta network node, and the right input is the condition matching result output by the current layer alpha network node. When the combined conditions are met, the beta network node outputs a new intermediate matching path result; Establish a terminal rule node at the end of each path of the rule graph structure, the terminal node receives the complete matching path output from the last layer of the beta network, identifies that the corresponding rule node is activated, and outputs the final executable rule matching graph.

3. The method of intelligent session state management of claim 1, wherein, Generating an initial state matching result set includes: Collecting real-time context data of each node in the communication system, the real-time context data including communication delay, packet loss rate, bandwidth utilization, cache occupancy, node connection state, data packet timestamp and round-trip delay; Pretreating the collected real-time context data according to node dimension, and establishing a context data structure; Inputting the context data structure into an α network in a rule matching graph, each α network node corresponding to a condition node, the α network node receiving the context data structure and performing a condition traversal operation according to a bound condition matching function, the condition traversal operation including reading a field value in the context data structure, and performing interval comparison and state flag matching on the field value; Marking the α network node satisfying the matching rule as a matching success result, and forming a condition-data pair of the condition node and a corresponding data record of the matching success result; Classifying all condition-data pairs according to a mapping relationship between condition nodes and rule nodes defined in the rule graph structure, and converging into respective rule paths; Counting the number of condition nodes matching successfully on each rule path, and identifying a rule path with at least one condition node matching successfully as an initial matching path; Generating an initial state matching result set including a rule identifier, a condition hit state and corresponding context data according to the matching conditions of all initial matching paths.

4. The method of intelligent session state management of claim 1, wherein, Building a multi-dimensional state vector set includes: Reading the matching records of each rule path from the initial state matching result set, identifying the condition nodes matching successfully therein, and extracting context data fields associated with the condition nodes, the context data fields including communication delay, packet loss rate, bandwidth utilization, cache occupancy, connection stability, data packet transmission interval and round-trip delay; Generating a matching state label according to the matching records of the condition nodes, each matching state label including a condition node identifier, a matching condition type and a matching time index; For each matching state label, extracting parameter values from the corresponding context data fields, performing normalization and interval coding operations on the extracted parameter values to obtain parameter standardization results, and combining the parameter standardization results in field order to form a single-condition node state feature vector; Concatenating all condition node state feature vectors under the same rule path in matching order to generate a rule path-level initial state vector; Collecting running indexes of each node in the communication system, the running indexes including CPU usage, memory occupancy, cache hit rate, interface call frequency and connection holding time, and performing unified quantization and standardization processing on the running indexes to obtain a node running index vector; According to the mapping relationship between the rule nodes and the condition nodes, matching and concatenating the node running index and the rule path-level initial state vector corresponding to the node to form a node-level composite state vector; Arranging all node-level composite state vectors in order of communication priority to build a multi-node state vector matrix; Performing format coding on the multi-node state vector matrix to output a multi-dimensional state vector set with a unified structure.

5. The method of intelligent session state management of claim 1, wherein, Building an energy function model includes: The constructed multi-dimensional state vector set is input into the Hopfield neural network, each dimension of the multi-dimensional state vector set corresponding to a neuron node, for representing the structured features of the current session state; A Hopfield neural network structure is constructed, in which a neuron node is created for each dimension of the multi-dimensional state vector, and a bidirectional symmetric connection relationship is established between all nodes to form a symmetric fully connected topology structure; According to the historical joint distribution between the dimensions in the multi-dimensional state vector, the connection weight values between each pair of neurons are calculated to generate a connection weight matrix; The bias values of each node are calculated in combination with the activation distribution of each neuron node in the historical state evolution to construct a bias vector; Each value in the input multi-dimensional state vector is assigned to the corresponding neuron node as the initial state, and normalization processing is performed on all state values to generate a standardized initial network state; An energy function model is constructed based on the established connection weight matrix, bias vector and standardized initial network state.

6. The method of intelligent session state management of claim 1, wherein, The iterative calculation of the energy function model is performed until the state converges, including: The constructed energy function model is loaded, the standardized initial network state is taken as input, and the iterative parameters of the Hopfield neural network are initialized, including the maximum number of iterations, the energy change threshold and the state stability judgment condition; The current state vector of all neuron nodes is extracted from the standardized initial network state, and an iterative loop structure is established, in which the neuron nodes are selected one by one in each iteration according to the asynchronous update strategy; For the selected neuron node, its local input signal is calculated, which is determined by the current state of other neurons connected to the corresponding neuron node, the corresponding connection weight and the bias value of the corresponding neuron node; The state value of the neuron node is updated according to the calculated local input signal, and the updated state is written into the current multi-dimensional state vector set, while the states of the unselected neuron nodes remain unchanged to form a network state vector; After all neuron nodes complete state updating in one iteration, the network overall energy value is calculated by calling the energy function model, the new network overall energy value is compared with the previous network overall energy value, and the energy change result is recorded; The network overall energy value, network overall energy difference and corresponding multi-dimensional state vector after each iteration are stored in the energy descent trajectory table; The neuron state updating and energy value calculation are repeatedly performed until the energy change amplitude is lower than the preset threshold, and it is determined that the network reaches the convergence condition; When the convergence condition is met, the network state vector of the last iteration in the energy descent trajectory table is extracted to generate the converged state output vector.

7. The method of intelligent session state management of claim 1, wherein, The structure consistency identification and difference index calculation include: The rule path records in the initial state matching result set generated by the rule matching graph are extracted, the condition nodes and their corresponding context data fields that match successfully in each rule path are obtained, and a rule path field set is formed; Extract the final state output vector from the state output vector of the energy function model, analyze the index order and field meaning of each dimension in the state output vector, and establish a correspondence table between the dimension index and the field label; According to the correspondence table between the dimension index and the field label, the field mapping relationship between the initial state matching result set and the state output vector is established, and each rule path field and the corresponding dimension in the state output vector establishes a one-to-one mapping pair; After the mapping relationship is established, the corresponding data value of each mapping pair is read in turn, the numerical difference calculation is performed on the continuous type field, the state consistency judgment is performed on the logic type field, and the encoding matching verification is performed on the symbolic type field. The comparison result of each field is recorded as a matching mark; All matching marks of each rule path are counted, the proportion of matching successful field number and total field number is calculated, and the structural consistency identifier of the rule path is formed; In the comparison process, all the fields judged as not matching are extracted, the numerical value offset is calculated, and the result is aggregated to form a difference index.

8. The method of intelligent session state management of claim 1, wherein, The generation of the final stable session state set includes: Receiving the structural consistency identifier and judging whether it meets the stability threshold condition, wherein the stability threshold condition is that the proportion of the number of matching fields in the structural consistency identifier is not less than a preset proportion threshold; When the structural consistency identifier is less than the preset proportion threshold, all the unmatched fields recorded in the difference index are extracted, and the corresponding condition node and the rule path thereof are located in the rule matching graph combined with the field type; According to the field difference type, the rule threshold adjustment operation is performed, including: adjusting the continuous type field to narrow the numerical tolerance range of the condition matching, adjusting the logic type field to improve the state judgment sensitivity, and adjusting the symbolic type field to update the symbol matching function in the condition node; Write the adjusted rule threshold into the corresponding condition node to complete the local update of the target rule path in the rule matching graph; Re-execute the matching process based on the updated rule matching graph to generate a new initial state matching result set, and re-calculate the structural consistency identifier; When the recalculated consistency identifier is greater than or equal to the stability threshold condition, the rule path corresponding to the rule matching structure, the context data field and the final state output vector are extracted to form a session state structure item; All session state structure items that meet the consistency threshold condition are summarized to generate the final stable session state set and output.

9. A system for intelligent session state management, characterized by It includes the following modules: A rule matching graph construction module is used to parse the session rules and generate a rule graph structure containing the mapping relationship between rule nodes and condition nodes, forming an executable rule matching graph; A context collection and preprocessing module is used to collect real-time context data of communication delay, packet loss rate, bandwidth utilization rate and cache occupancy rate from a communication system, and establish a context data structure according to node dimensions; A rule matching execution module is used to input the context data structure into the rule matching graph, and perform condition matching and path construction through the α network node and the β network node to generate an initial state matching result set; A state vector construction module is configured to extract matching state labels and context feature parameters based on the initial state matching result set, and generate a multi-dimensional state vector set in combination with node operation indexes; An energy model construction module is configured to establish a connection weight matrix and a bias vector based on a Hopfield neural network, construct an energy function model, and initialize a neural network structure; An energy convergence calculation module is configured to perform iterative calculation of the energy function model, update neuron states using an asynchronous update strategy, and generate a final converged state vector and an energy descent trajectory; A consistency comparison module is configured to perform field-level mapping comparison between the converged state vector and the initial state matching result set, and output a structural consistency identifier and a difference index; A rule feedback adjustment module is configured to adjust matching threshold parameters of relevant rule paths in the rule matching graph according to the difference index; A session state output module is configured to output a final stable session state set when the structural consistency identifier meets a preset condition.

Images