A method for collecting running state of a device supporting multi-network environment

Abstract

This invention discloses a method for collecting device operating status in multi-network environments, relating to the field of network communication technology. The method includes: initiating a data acquisition end to detect available communication networks in parallel, acquiring connectivity, bandwidth, and latency characteristics, and forming a multi-network capability profile; sampling device operating electrical signals under the constraints of the multi-network capability profile and performing time alignment, performing state discrimination and feature extraction to obtain initial operating status semantics; sending computational contribution sub-semantics in parallel according to a semantic division-of-labor reporting strategy, performing offline caching on failed computational contribution sub-semantics and retransmitting them in order of freshness constraints; and having the processing end align and splice the computational contribution sub-semantics according to their dependencies and freshness constraints to reconstruct a continuous operating status semantic sequence. This invention, by jointly configuring the collection frequency, expression precision, and communication network assignment relationships of computational contribution sub-semantics based on freshness constraints and multi-network capability profiles, forms a semantic division-of-labor reporting strategy.

Description

Technical Field

[0001] This invention relates to the field of network communication technology, and in particular to a method for collecting device operating status that supports multiple network environments. Background Technology

[0002] For industrial internet and intelligent operation and maintenance scenarios, equipment operation status acquisition technology has gradually evolved from single-channel data reporting to a multi-network collaborative edge acquisition paradigm. The acquisition end can perceive the connectivity, bandwidth and latency characteristics of wired networks, wireless WiFi networks and cellular communication networks in parallel, forming a multi-network capability profile. Under a unified time reference, it samples, aligns, judges the status and extracts features of the equipment operation electrical signals, generating structured operation status semantics to support continuous analysis and management of the operation process from a remote location.

[0003] Under conditions where the capabilities of multiple networks fluctuate over time, related technologies often organize data transmission with a single reporting granularity or fixed scheduling rules. This makes it difficult to finely couple the time-sensitive differences of different semantic elements with network latency, bandwidth, and connectivity stability in the same configuration process. As a result, it is difficult to match the reporting density, payload granularity, and network carrying capacity in the long term. The operational status semantics that need to be aligned and spliced ​​according to dependencies are difficult to maintain continuity and consistent time reference within the time limit. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for collecting device operating status in multi-network environments to solve the problem that it is difficult to simultaneously meet time constraints and continuous reconstruction when collecting and reporting operating status semantics under conditions of fluctuating network capabilities.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for collecting device operating status in multi-network environments. The method includes: starting a collection end to detect available communication networks in parallel, acquiring connectivity, bandwidth, and latency characteristics, and forming a multi-network capability profile; sampling device operating electrical signals under the constraints of the multi-network capability profile and performing time alignment, performing state discrimination and feature extraction to obtain initial operating status semantics; constructing a contribution decomposable representation based on the initial operating status semantics, and splitting it into computational contribution sub-semantics, establishing dependencies; allocating short-term and long-term freshness constraints based on the device operating status change characteristics reflected by each computational contribution sub-semantics at different sampling time scales; adapting the collection frequency, expression accuracy, and communication network assignment relationships of the computational contribution sub-semantics to timeliness based on the freshness constraints and the multi-network capability profile, and forming a semantic division of labor reporting strategy; sending the computational contribution sub-semantics in parallel according to the semantic division of labor reporting strategy, performing offline caching on computational contribution sub-semantics that failed to be sent, and retransmitting them in order of freshness constraints; and aligning and splicing the computational contribution sub-semantics according to the dependencies and freshness constraints at the processing end to reconstruct a continuous operating status semantic sequence.

[0007] As a preferred embodiment of the device operation status acquisition method supporting multiple network environments described in this invention, the communication network includes wired network, wireless WiFi network and cellular communication network.

[0008] As a preferred embodiment of the device operation status collection method supporting multiple network environments described in this invention, the specific steps for forming a multi-network capability profile are as follows: Start the data acquisition terminal and perform availability tests on wired networks, wireless WiFi networks, and cellular communication networks respectively to establish a set of currently available communication networks; For each communication network in the communication network set, connectivity parameters, connectivity stability parameters, bandwidth parameters, and time delay parameters are collected to form a corresponding set of network capability parameters. The set of network capability parameters is structured, encapsulated, and normalized according to the communication network identifier, and a multi-network capability profile is output.

[0009] As a preferred embodiment of the device operation status acquisition method supporting multiple network environments described in this invention, the specific steps for obtaining the initial operation status semantics are as follows: Under the constraints of multi-network capability profiling, the device operation electrical signals generated during the operation of the device are sampled, the corresponding sampling timing is configured, and a time identifier is added to the sampled electrical signals. Time alignment processing is performed on the sampled electrical signals with additional time markers to form an electrical signal sequence under a unified time reference; The electrical signal sequence is processed for state discrimination and feature extraction, and the extracted feature information is organized into initial running state semantics according to the field structure.

[0010] As a preferred embodiment of the device operation status acquisition method supporting multiple network environments described in this invention, the construction contribution can be decomposed and represented, and the specific steps are as follows. Read the state feature fields that carry feature information in the initial running state semantics, and divide them according to the role of the state feature information in running state discrimination to form feature parts; For the feature part, the computational dependency relationship is established in the process of operation state discrimination and state evolution analysis; The feature components and their corresponding computational dependencies are structured and organized to form a decomposable contribution representation.

[0011] As a preferred embodiment of the device operation status acquisition method supporting multiple network environments described in this invention, the specific steps for establishing dependencies are as follows: Based on the contribution decomposable representation, the contribution decomposable representation is traversed according to the usage position, reference relationship and calculation order of state features in the operation state discrimination and state evolution analysis, and the contribution part that can participate in the calculation independently is extracted. The contribution components are mapped to corresponding computational contribution sub-semantics. Based on the computational order and reference relationships of the computational contribution sub-semantics in the process of runtime state discrimination and state evolution analysis, the dependency relationships between the computational contribution sub-semantics are established.

[0012] As a preferred embodiment of the device operation status acquisition method supporting multiple network environments described in this invention, the specific process of allocating short-term and long-term freshness constraints is as follows: For each calculated contribution sub-semantic, corresponding state change sequences are constructed under both short-time sampling timescales and long-time sampling timescales; Based on the state change sequence, extract change feature information that characterizes the consistency and degree of deviation of changes across time scales; Generate and assign an index to each computational contribution sub-semantic based on the change feature information; Based on the allocation index, each computational contribution sub-semantic is mapped to the corresponding short-term and long-term freshness constraints.

[0013] As a preferred embodiment of the device operation status collection method supporting multiple network environments described in this invention, the specific process of forming a semantic division of labor reporting strategy is as follows: For each computational contribution sub-semantic, the corresponding freshness constraint is associated with the latency parameter, bandwidth parameter and connectivity stability parameter in the multi-network capability profile to form a time-adaptive description of the computational contribution sub-semantic. Based on the time-adaptation description, the collection frequency configuration, expression precision configuration, and communication network assignment relationship of the computational contribution sub-semantics are adjusted synchronously during the same configuration process. Based on the acquisition frequency configuration, expression accuracy configuration, and communication network assignment relationship, the semantic execution order of all calculated contribution sub-semantics is organized within the same time window to form a semantic division of labor reporting strategy.

[0014] As a preferred embodiment of the device operation status acquisition method supporting multiple network environments described in this invention, the specific process of performing offline caching and retransmitting the calculated contribution sub-semantics of failed transmissions according to freshness constraints is as follows. Based on the semantic division of labor reporting strategy, the computational contribution sub-semantics are sent in parallel on the corresponding communication network, and the computational contribution sub-semantics that have not been sent are entered into the offline caching process. The computational contribution sub-semantics that enter the offline caching process are written into the offline cache area, and the computational contribution sub-semantics in the offline cache are sorted according to the freshness constraint. When the communication network is ready to transmit, the contribution sub-semantics are read from the offline buffer according to the retransmission order and the retransmission is performed.

[0015] As a preferred embodiment of the device operation status acquisition method supporting multiple network environments described in this invention, the specific process of reconstructing the continuous operation status semantic sequence is as follows: The processing end receives the computational contribution sub-semantics of parallel transmission and retransmission, and organizes the computational contribution sub-semantics in order according to the dependency relationship between them. The processing end calculates the time difference between the receiving time of the calculated contribution sub-semantics and the time identifier attached during generation, and compares the time difference with the allowable timeliness range limited by the freshness constraint corresponding to the calculated contribution sub-semantics to determine whether the calculated contribution sub-semantics meets the timeliness requirements. Computational contribution sub-semantics that do not meet the timeliness requirements are filtered out. The remaining computational contribution sub-semantics are then aligned by time according to the time identifier and spliced ​​together in the order of dependency to form a continuous running state semantic sequence.

[0016] The beneficial effects of this invention are as follows: By combining the acquisition frequency, expression accuracy, and communication network assignment relationship of the computational contribution sub-semantics based on freshness constraints and multi-network capability profiles, a semantic division of labor reporting strategy is formed. Each computational contribution sub-semantics obtains a transmission channel and reporting rhythm that matches the urgency of time within the allowable time frame. At the same time, the sending timetable and queue arrangement relationship are sequentially organized within the same time window, realizing the controllable allocation of reporting density and single payload on the time axis. This reduces congestion and invalid retransmissions during multi-network parallel transmission, improves the operability of arrival order sorting and alignment splicing, and stably supports the reconstruction of semantic sequences in continuous operation. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart for a method to collect device operating status in multi-network environments.

[0019] Figure 2 A flowchart for profiling network capabilities.

[0020] Figure 3 A flowchart for constructing the sub-semantics for computational contribution.

[0021] Figure 4 This is a flowchart for semantic division of labor reporting. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for collecting device operating status supporting multiple network environments, including the following steps: S1: Start the acquisition end to detect available communication networks in parallel, obtain connectivity, bandwidth and latency characteristics and form a multi-network capability profile.

[0026] S1.1: Start the data acquisition terminal and perform availability tests on wired networks, wireless WiFi networks, and cellular communication networks respectively, and establish a set of currently available communication networks.

[0027] Specifically, the acquisition terminal is activated to initiate communication probe operations on wired networks, wireless WiFi networks, and cellular communication networks respectively. By sending basic communication probe messages or calling the corresponding network interfaces to obtain connection response status, it is determined whether each communication network is in a communicable state. Networks that can complete communication probe interactions are recorded as available networks, and a set of currently available communication networks is established.

[0028] It should be noted that communication networks include wired networks, wireless WiFi networks, and cellular communication networks.

[0029] S1.2: For each communication network in the communication network set, collect connectivity parameters, connectivity stability parameters, bandwidth parameters, and time delay parameters to form a corresponding set of network capability parameters.

[0030] Specifically, for each communication network in the communication network set, the acquisition end performs basic communication test operations. It reads the connection establishment status, link maintenance status, and disconnection flag information by calling the connection management interface corresponding to the communication network, and determines whether the communication network is in a communicable state based on the status information, obtains connectivity parameters, records the connection maintenance status, reconnection count, or interruption flag within a continuous time window, and statistically analyzes the changes in connection status over time to form connectivity stability parameters. It sends and receives test data packets with known data volume within a preset time window, calculates the data transmission capacity per unit time based on the total amount of data actually transmitted within the time window, and obtains bandwidth parameters.

[0031] By sending communication probe messages to the corresponding communication network and recording the round-trip time from sending the probe message to receiving the response, the round-trip time is used as the latency parameter of the communication network. The connectivity parameters, connectivity stability parameters, bandwidth parameters, and latency parameters of each communication network are associated and recorded according to the network identifier to form a set of network capability parameters characterizing the communication capabilities of each communication network.

[0032] It should be noted that the preset time window is configured by the acquisition end when starting the communication test based on the communication network type and measurement stability requirements. The value range is limited to the time interval that covers at least one complete data transmission and reception process and avoids occupying communication resources for a long time. It is usually set in the range of milliseconds to seconds.

[0033] Connectivity parameters include connection establishment success flags, connection failure flags, and current connection status flags; connectivity stability parameters include connection hold duration within a continuous time window, number of connection interruptions, number of reconnections, and connection status change sequence; bandwidth parameters include the amount of data transmitted and corresponding time flag records; latency parameters include the sending time flag, receiving time flag, and time difference records.

[0034] S1.3: The set of network capability parameters is structured, encapsulated, and normalized according to the communication network identifier, and a multi-network capability profile is output.

[0035] Specifically, based on the historical statistical range and current observation range of the network capability parameters of each communication network, the acquisition end performs linear interval mapping processing on the connectivity parameters, bandwidth parameters, and time delay parameters respectively, mapping each parameter to a unified dimensionless standard numerical range, eliminating dimensional differences and maintaining the relative size relationship between parameters, forming a multi-network capability profile that can be compared and configured within the same decision space.

[0036] Among them, multi-network capability profiling is formed by structurally encapsulating and normalizing the capability parameters of different communication networks to create a unified standard network performance model.

[0037] S2: Under the constraint of multi-network capability profile, sample the device's operating electrical signals and perform time alignment, execute state discrimination and feature extraction to obtain the initial operating state semantics.

[0038] S2.1: Under the constraints of multi-network capability profiling, sample the device operation electrical signals generated during device operation, configure the corresponding sampling timing, and add time markers to the sampled electrical signals.

[0039] Specifically, under the constraints of multi-network capability profiles, the acquisition end uses the latency and bandwidth parameters in the multi-network capability profiles as the basis for sampling scheduling, sets the sampling period and sampling trigger time of the device's operating electrical signals, so that the amount of data generated by sampling matches the available transmission capacity, and calls the unified time base established by the acquisition end to read the clock count value during each sampling, writes the clock count value into the corresponding sampling electrical signal and adds a time identifier.

[0040] A unified time base can be established through a network time protocol server, GPS synchronization signals, and an internal high-precision clock.

[0041] S2.2: Perform time alignment processing on the sampled electrical signals with additional time markers to form an electrical signal sequence under a unified time reference.

[0042] Specifically, when performing time alignment processing on the sampled electrical signals with additional time markers, the acquisition end sorts and corrects the offset of the sampled electrical signals according to the time markers, maps the electrical signals obtained at different sampling times onto a unified time reference axis, and adjusts the alignment of sampled electrical signals with time deviations to form an electrical signal sequence based on the same time reference.

[0043] S2.3: Perform state discrimination and feature extraction processing on the electrical signal sequence, and organize the extracted feature information into initial running state semantics according to the field structure.

[0044] Specifically, the electrical signal sequence is divided into continuous time periods according to a unified time reference. Within each time period, the sampled amplitude sequence is read, and the amplitude concentration, amplitude fluctuation range, and amplitude change direction are statistically analyzed to form amplitude statistical characteristics. The amplitude differences between adjacent sampling points are calculated segment by segment to form change rate characteristics. At the same time, the duration of continuous maintenance of the electrical signal within the same amplitude range and with the same change trend is statistically analyzed to form state persistence characteristics. The amplitude statistical characteristics, change rate characteristics, and state persistence characteristics are matched with the set of operating states in the state discrimination rules. When the amplitude statistical characteristics, change rate characteristics, and state persistence characteristics meet the discrimination range of the corresponding operating state, the time period is marked as the corresponding equipment operating state.

[0045] The acquisition end writes the device's operating status marker, amplitude statistical feature, rate of change feature, and state persistence feature into the status marker field, amplitude statistical feature field, rate of change feature field, and state persistence feature field, respectively, and encapsulates them with time identifiers according to a unified field order to form the initial operating status semantics.

[0046] The historical electrical signals are divided into state samples such as normal operation state, standby state, state switching state, and abnormal fluctuation state according to known operating states. The amplitude statistical characteristic range, change rate characteristic range, and state duration characteristic range corresponding to each state sample are calculated respectively. The characteristic value range that stably covers the state sample under the same operating state is used as the discrimination range of the corresponding operating state.

[0047] S3: Construct a decomposable representation of contributions based on the initial running state semantics, and break it down into computational contribution sub-semantics to establish dependencies.

[0048] S3.1: Read the state feature fields that carry feature information in the initial running state semantics, and divide them according to the role of the state feature information in running state discrimination to form feature parts.

[0049] Specifically, when reading the state feature fields that carry feature information in the initial running state semantics, the acquisition end extracts the amplitude statistical features, change rate features, and state persistence features item by item according to the field identifier, and groups and organizes them according to the usage position of each state feature in the state discrimination rules. The state features involved in state switching identification are grouped into one group, the state features involved in state stability judgment are grouped into another group, and the state features involved in trend change assessment are grouped into another group to form the feature part.

[0050] Among them, the characteristic information of the carrier refers to the various electrical signal characteristic data related to the operating status of the equipment.

[0051] The status discrimination rules are set by the acquisition end during the initialization phase based on the device's operating characteristics and historical electrical signal behavior features. The status discrimination rules include at least time window division rules, feature statistics and comparison rules, and status mapping rules, and are based on a predefined set of operating states for discrimination.

[0052] For example, when the amplitude of the electrical signal of the device changes beyond the preset range, it may be judged as a fault state; when the rate of change of the device remains stable within a certain time window, it may be judged as a normal state; when the electrical signal state of the device remains stable within a certain range, it may be judged as a standby state. The preset range of change is set by analyzing the historical electrical signal data, operating characteristics and technical specifications of the device, combined with statistical methods. The specific range of values ​​depends on the type of device and the working conditions.

[0053] State characteristics include at least amplitude statistical characteristics, rate of change characteristics, and state persistence characteristics.

[0054] S3.2: For the feature part, establish the computational dependency relationship in the process of operation state discrimination and state evolution analysis.

[0055] Specifically, based on the rules for judging the operating state and the state evolution analysis process, the data acquisition end sorts out the order of use and data reference relationships of the feature parts in the calculation process. The feature parts for state switching identification are determined as the starting feature parts without any prior dependencies; the feature parts for judging state stability are set as the subsequent feature parts that depend on the starting feature parts; and the feature parts for evaluating trend changes are set as the combined feature parts that depend on the feature parts for judging state switching and the feature parts for judging state stability. The prior dependencies of each feature part are clarified, forming the computational dependencies of sequential calculation and splicing reconstruction.

[0056] S3.3: Organize the feature parts and their corresponding computational dependencies in a structured manner to form a contribution decomposable representation.

[0057] Specifically, each feature is associated with its corresponding source state feature field, running state discrimination function type, state evolution analysis function type, pre-dependency identifier, and subsequent reference identifier, and encapsulated according to a unified field structure. The encapsulated feature parts are arranged in order according to the pre-dependency identifier and subsequent reference identifier, and a semantic mapping identifier corresponding to the subsequent calculation contribution sub-semantics is established for each feature part, forming a decomposable representation of the contribution.

[0058] Among them, structured organization refers to writing "feature parts" and "computational dependencies" into a unified field structure, including feature part identifier, source state feature field, runtime state discrimination function type, state evolution analysis function type, preceding dependency identifier, subsequent reference identifier, and semantic mapping identifier.

[0059] S3.4: Based on the contribution decomposable representation, the contribution decomposable representation is traversed according to the position, reference relationship and calculation order of the state features in the operation state discrimination and state evolution analysis, and the contribution parts that can participate in the calculation independently are extracted.

[0060] Specifically, based on the contribution decomposable representation, the acquisition end performs traversal processing on each feature part according to the structural template recorded in the contribution decomposable representation, reads the source state feature field and calculation dependency information corresponding to each feature part item by item, and identifies the feature parts that are not subject to the constraints of the preceding dependencies in the current calculation stage according to the decomposition principle of "independent source fields, independent discrimination effects, and independent reference of dependency relationships", and extracts the feature parts as contribution parts that can participate in the calculation independently.

[0061] During the initialization phase, a unified field arrangement is performed based on the field structure of the initial running state semantics, the role type in the state discrimination rules, the reference relationship in the state evolution analysis process, as well as the pre-dependency identifier and the subsequent reference identifier, to form a structure template. The structure template includes feature part identifier, source state feature field, running state discrimination role type, state evolution analysis role type, pre-dependency identifier, subsequent reference identifier, and semantic mapping identifier.

[0062] The decomposition principle refers to dividing the feature parts in the decomposable contribution representation into contribution parts that can participate in the calculation independently, based on whether the source state feature fields are independent, whether the running state discrimination function is independent, whether the state evolution analysis function is independent, whether the pre-dependency relationship is clear, and whether the subsequent reference relationship can be called independently. If a feature part has an independent source field, an independent discrimination function, and can be called independently by subsequent calculations, it is decomposed into a separate contribution part. If multiple feature parts must participate in the same state evolution analysis and share the same pre-dependency relationship, they are combined and decomposed into a single contribution part.

[0063] S3.5: Map the contribution parts to the corresponding computational contribution sub-semantics, and establish the dependency relationship between computational contribution sub-semantics based on the computational order and reference relationship of computational contribution sub-semantics in the process of running state discrimination and state evolution analysis.

[0064] Specifically, the contribution components are mapped to a unified semantic description structure according to the source state feature fields, the operational state discrimination function type, and the state evolution analysis function type. Corresponding computational contribution sub-semantic identifiers are generated based on the source state feature field identifier, function type identifier, and pre-dependency identifier. When a contribution component has an independent source field and can participate in operational state discrimination or state evolution analysis as a separate reference object, it is mapped to a computational contribution sub-semantic. When multiple contribution components participate in the same state evolution analysis and have common pre-dependencies, they are combined and mapped to a single computational contribution sub-semantic. The acquisition end writes the computational contribution sub-semantic identifier into the semantic mapping identifier field in the contribution decomposable representation and establishes dependencies between computational contribution sub-semantics based on their computational order and reference relationships.

[0065] S4: Based on the characteristics of equipment operating status changes reflected by each calculated contribution sub-semantic at different sampling time scales, allocate short-term and long-term freshness constraints.

[0066] S4.1: For each computational contribution sub-semantic, construct the corresponding state change sequence under both short-time sampling timescales and long-time sampling timescales.

[0067] Specifically, for each computational contribution sub-semantic, the state feature fields corresponding to the computational contribution sub-semantic in the initial running state semantics are read, and the state feature fields are arranged in chronological order based on the unified time reference formed by time alignment processing. The state feature fields are then subjected to time-by-time difference calculation with the adjacent sampling times corresponding to the short time sampling time scale as the boundary to form a state change sequence under the short time sampling time scale. The state change sequence under the short time sampling time scale is then subjected to window-based summary calculation with the aggregation time window corresponding to the long time sampling time scale as the boundary, and arranged in window order to form a state change sequence under the long time sampling time scale.

[0068] It should be noted that the short-time sampling timescale refers to the timescale formed by taking the sampling sequence of the device's operating electrical signal as a reference and using adjacent sampling times as the basis for continuous change characterization, describing the state changes of the calculated contribution sub-semantics between adjacent sampling points; the long-time sampling timescale refers to the timescale formed by aggregating the state changes corresponding to multiple consecutive sampling times based on the short-time sampling timescale, describing the overall change behavior of the calculated contribution sub-semantics across multiple sampling times; the time aggregation is obtained by the acquisition end combining continuous sampling intervals under a unified time reference.

[0069] S4.2: Based on the state change sequence, extract change feature information that characterizes the consistency and degree of deviation of changes across time scales.

[0070] Specifically, based on the state change sequences under short-time sampling timescales and the state change sequences under long-time sampling timescales, the aggregation time window under the long-time sampling timescale is mapped to the corresponding short-time sampling timescale coverage interval. Within each coverage interval, the state change sequences under the short-time sampling timescale are summed up within the window to generate the corresponding window change representation quantity.

[0071] The window change representation is compared window by window with the state change sequence under the long sampling time scale within the corresponding aggregated time window. The consistency of change across time scales is characterized by recording the consistency of change direction. The difference is calculated for the corresponding window change representation and accumulated within multiple aggregated time windows to characterize the degree of deviation of change across time scales, thus forming change feature information.

[0072] Among them, change feature information refers to extracting features that characterize the consistency and degree of deviation of changes across time scales by comparing state change sequences under short-term and long-term sampling time scales.

[0073] The window change characteristic is represented by the following formula: ; In the formula, Represents the semantics of calculating contribution sub-components In the Each long sampling timescale corresponds to an aggregation time window The window change within the timeframe represents the quantity. This indicates a long sampling timescale. The status characteristic identifier is represented. Indicates the aggregation time window The corresponding set of short-time sampling timescale coverage intervals, Represents the semantics of calculating contribution sub-components At a short sampling timescale, the first The state change obtained between adjacent sampling times Indicates the sampling time index.

[0074] The formula for consistency of change across time scales is: ; In the formula, Represents the semantics of calculating contribution sub-components In the Each long sampling timescale corresponds to an aggregation time window Consistency indicator of changes across time scales within the region. Represents the semantics of calculating contribution sub-components In the Each long sampling timescale corresponds to an aggregation time window The change in state within. This represents the state change over a long sampling timescale, with values ​​ranging from [value missing]. This indicates that the direction of state change is consistent under both short-term and long-term sampling timescales, and the value is [value missing]. This indicates that the direction of state change is inconsistent between short-term and long-term sampling timescales.

[0075] The degree of deviation across time scales is expressed by the formula: ; In the formula, Represents the semantics of calculating contribution sub-components The cumulative deviation of cross-timescale changes over multiple long-term sampling timescales This represents the total number of aggregation time windows involved in the calculation.

[0076] The changes in short-time sampling timescale, window change representation, and long-time sampling timescale originate from the same state characteristic system, differing only in timescale. Addition, subtraction, averaging, and cumulative operations are all performed between variables of the same dimension. Therefore, the formulas for window change representation, cross-timescale change consistency, and cross-timescale change deviation maintain the same dimension.

[0077] S4.3: Generate an assigned index for each computational contribution sub-semantic based on the change feature information.

[0078] Specifically, the acquisition end reads the corresponding change feature information for each calculated contribution sub-semantic, summarizes the absolute changes of the window change representation quantity within multiple aggregated time windows to obtain the window change intensity, and performs linear normalization processing on the deviation of the window change intensity and the cross-time scale change based on the minimum and maximum values ​​of the corresponding calculated contribution sub-semantic in historical data, and uniformly maps them to the numerical range of 0 to 1 to form normalized window change intensity and normalized deviation.

[0079] The acquisition end uses the intensity of normalized window change as the basic quantity of change in the allocation index, the consistency of change across time scales as the enhancement quantity of the allocation index, and the degree of normalization deviation as the suppression quantity of the allocation index. In the same processing flow, a single index value is generated according to the combination rule of "the basic quantity of change and the enhancement quantity together improve the allocation index, and the suppression quantity reduces the allocation index". The single index value is then associated with the corresponding computational contribution sub-semantics to generate the allocation index.

[0080] A better approach is to classify or weight changes in operating status based on fixed rules, empirical thresholds, or a single time scale. Instead, by normalizing and combining the consistency and deviation of changes across time scales to generate an allocation index, the state change characteristics of short-term and long-term sampling time scales are uniformly mapped to a computable index space, thus achieving a refined and adaptive allocation of freshness constraints.

[0081] S4.4: Based on the allocation index, each computational contribution sub-semantic is mapped to the corresponding short-term and long-term freshness constraints respectively.

[0082] Specifically, the acquisition end reads the corresponding allocation index for each computational contribution sub-semantic, and reads the short-term freshness constraint configuration range and the long-term freshness constraint configuration range. It completes this by comparing the boundary value relationship between the allocation index and the short-term freshness constraint configuration range and the boundary value relationship between the allocation index and the long-term freshness constraint configuration range. Based on the matching relationship of interval matching processing, it establishes a binding association between the computational contribution sub-semantic and the short-term or long-term freshness constraint, and writes the constraint field of the short-term or long-term freshness constraint into the constraint record field of the computational contribution sub-semantic.

[0083] It should be noted that the freshness constraint refers to the allowed time range for each computational contribution sub-semantic. The allowed time range is an allocation index calculated by the acquisition end based on the state change characteristics of the computational contribution sub-semantic under different sampling time scales, and is set by interval matching with the short-term freshness constraint configuration range and the long-term freshness constraint configuration range. The constraint information corresponding to the allowed time range includes a time identifier field, a time span field, and a unified time reference field.

[0084] Both the short-term and long-term freshness constraint configuration ranges are set by the acquisition end based on the device's operating characteristics and communication network capability parameters. Device operating characteristics include the frequency and persistence of state changes in the computational contribution sub-semantics at different sampling time scales. Communication network capability parameters include latency, bandwidth, and connectivity parameters. The acquisition end determines the short-term and long-term freshness constraint configuration ranges based on the computational contribution sub-semantics with high state change frequency and sensitivity to timeliness, corresponding to smaller time spans, and those with low state change frequency or strong persistence, corresponding to larger time spans. The short-term freshness constraint configuration range corresponds to the time span allowed for computational contribution sub-semantics with strict timeliness requirements, while the long-term freshness constraint configuration range corresponds to the time span allowed for computational contribution sub-semantics with more lenient timeliness requirements.

[0085] S5: Based on freshness constraints and multi-network capability profiles, the collection frequency, expression accuracy, and communication network assignment relationships of the computational contribution sub-semantics are adapted in a timely manner, and a semantic division of labor reporting strategy is formed.

[0086] S5.1: For each computational contribution sub-semantic, the corresponding freshness constraint is associated with the latency parameter, bandwidth parameter and connectivity stability parameter in the multi-network capability profile to form a time-adaptive description of the computational contribution sub-semantic.

[0087] Specifically, for each computational contribution sub-semantic, the acquisition end reads the short-term freshness constraint and long-term freshness constraint corresponding to the computational contribution sub-semantic. Then, in the multi-network capability profile, it sequentially reads the latency parameters, bandwidth parameters, and connectivity stability parameters corresponding to wired networks, wireless WiFi networks, and cellular communication networks according to the communication network identifier. Using the allowable timeframe defined by the freshness constraint corresponding to the computational contribution sub-semantic as the time benchmark, and combined with the latency parameters of the communication network, it calculates the time required for the communication network to complete one computational contribution sub-semantic transmission in the current state, obtains the differential computation amount, and writes it into the timeliness record field. Based on the transmission load characteristics formed by the computational contribution sub-semantic under the current acquisition frequency configuration and expression precision configuration, and combined with the bandwidth parameters of the communication network, it matches and judges the communication network's ability to carry the transmission of the computational contribution sub-semantic in the current state, obtains the matching computation amount, and writes it into the carrying record field.

[0088] Simultaneously, connectivity stability parameters are written into the stability record field. The acquisition end associates and encapsulates the timeliness record field, bearer record field, and stability record field according to the communication network identifier and the computational contribution sub-semantic identifier. The latency adaptation information, bandwidth adaptation information, and connectivity stability adaptation information corresponding to the computational contribution sub-semantic under different communication networks are kept aligned in the same structure to form the timeliness adaptation description of the computational contribution sub-semantic.

[0089] A better approach, compared to conducting a unified capability assessment on a network-by-network basis and selecting communication networks or adjusting transmission parameters at a coarse-grained task level, is to sink the freshness constraint down to the semantic level of computational contribution sub-concepts and associate it with the latency, bandwidth, and connectivity stability parameters of different communication networks to form a time-adaptive description, thereby achieving a fine coupling between semantic time value and multi-network capabilities and a more targeted scheduling advantage.

[0090] The formula for calculating the computational cost of the difference is: ; In the formula, Indicates the first The computational contribution sub-semantics in the th... Differential computation complexity under a communication network Indicates the first The allowable timeframe defined by the freshness constraints corresponding to each computational contribution sub-semantic. Indicates the first The latency parameters of a communication network Indicates the first The amount of payload data formed by each computational contribution sub-semantic under the current expression precision configuration. Indicates the first Bandwidth parameters of a communication network, Indicates the first The computational contribution sub-semantics in the th... The queue waiting time between entering the sending queue and starting to send in a communication network.

[0091] when When the value is ≥0, it indicates that the time required for the communication network to complete one computational contribution sub-semantic transmission in the current state does not exceed the allowable time limit, and the communication network is marked as meeting the time limit requirement; when A value less than 0 indicates that the time required for the communication network to complete a single computational contribution sub-semantic transmission in the current state exceeds the allowable time limit, and the communication network is marked as not meeting the time limit requirements.

[0092] S5.2: Based on the time-adaptive description, the acquisition frequency configuration, expression precision configuration, and communication network assignment relationship of the computational contribution sub-semantics are adjusted synchronously during the same configuration process.

[0093] Specifically, based on the timeliness adaptation description, the acquisition end sequentially reads the timeliness record field, bearer record field, and stability record field for each calculation contribution sub-semantic, and performs communication network assignment relationship filtering processing in wired networks, wireless WiFi networks, and cellular communication networks. The acquisition end reads the timeliness difference calculation amount in the timeliness record field, the transmission load matching calculation amount in the bearer record field, and the connectivity stability parameter in the stability record field. When the timeliness difference calculation amount corresponding to the currently evaluated communication network is not less than zero and the transmission load matching calculation amount is not less than zero, the identifier of the currently evaluated communication network is included in the available communication network identifiers.

[0094] The computational load matching computation is obtained by comparing the available bandwidth parameter of the communication network at the acquisition end with the data load corresponding to the semantic contribution sub-computation.

[0095] While ensuring that the computational contribution sub-semantics meets the freshness constraint, the sampling period and triggering rhythm of the computational contribution sub-semantics are adjusted in combination with the communication network bearer adaptation reflected in the bearer record field. Based on the freshness constraint corresponding to the computational contribution sub-semantics, the sampling period level is matched first, and the computational load is matched with the transmission load in the bearer record field to determine the current network bearer capacity. When the bearer capacity meets the current reporting density, the current sampling period is maintained and sampling is triggered at the predetermined sending time. When the bearer capacity is insufficient to support the current reporting density, the sampling period is switched to the extended period level, and the sequential triggering is adjusted to the merged triggering to reduce the number of times the computational contribution sub-semantics are generated and reported per unit time.

[0096] The extended period range is formed by expanding the current sampling period by a multiple, which is selected based on the matching relationship between the available bandwidth of the communication network, the amount of data to be sent, and the current reporting density.

[0097] Based on the communication network connectivity stability reflected in the stability record field and the bandwidth adaptation reflected in the bearer record field, the granularity of field retention and numerical expression in the calculation contribution sub-semantic is adjusted. The expression precision level is selected according to the connectivity stability parameters in the stability record field and the transmission load in the bearer record field to match the computational load. When the communication network carrying capacity meets the current single reporting load, the complete fields of the calculation contribution sub-semantic are retained and the original numerical expression is used. When the communication network carrying capacity is insufficient, the calculation contribution sub-semantic identifier, time identifier, dependency identifier, freshness constraint field, and core state feature field are retained first, and unnecessary extended fields are deleted. The original numerical expression is converted into a fixed-point expression to reduce the single reporting load. In the same configuration process, the acquisition end uniformly associates the updated acquisition frequency configuration, expression precision configuration, and communication network assignment relationship with the calculation contribution sub-semantic identifier.

[0098] The reporting density refers to the frequency or density of data generated and sent within the same time window. The acquisition end counts the number of samples within the time window according to the sampling period, counts the number of actual reporting triggers according to the trigger rhythm, and combines the set of sending payload fields corresponding to each reporting to form the reporting density.

[0099] The reporting load refers to the amount of data that needs to be transmitted in a single reporting transmission to calculate the contribution sub-semantics. It is determined by the number of reserved fields, the data length of each field, the precision of the numerical expression, and the additional identification information.

[0100] Adjustments are made to the field retention granularity and numerical expression granularity in the contribution sub-semantics calculation, thereby controlling the single-report load; for example, from the original... Downgraded to Remove redundant fields to reduce data volume; the granularity of numerical representation can be reduced by converting floating-point numbers to fixed-point format or retaining numerical values ​​to a specified number of decimal places, thereby reducing the precision of data representation and reducing transmission load.

[0101] S5.3: Based on the acquisition frequency configuration, expression accuracy configuration and communication network assignment relationship, organize the execution order of all calculated contribution sub-semantics within the same time window to form a semantic division of labor reporting strategy.

[0102] Specifically, based on the timeliness adaptation description, the acquisition end compares the freshness constraints corresponding to the calculated contribution sub-semantics with the timeliness record field, carrier record field, and stability record field under each communication network. Under the premise of satisfying the timeliness constraints, it selects an available communication network and simultaneously obtains the acquisition frequency configuration and expression precision configuration that are compatible with the capabilities of the selected communication network, thus obtaining the acquisition frequency configuration, expression precision configuration, and communication network assignment relationship. The acquisition end reads the acquisition frequency configuration, expression precision configuration, and communication network assignment relationship corresponding to each calculated contribution sub-semantics one by one.

[0103] The sampling period, trigger rhythm, and time window start and end identifiers in the acquisition frequency configuration are read. Starting from the time window start identifier, the transmission trigger time is generated by recursively generating the transmission trigger time according to the sampling period. Then, the transmission trigger time is processed by retention, merging, and sequential arrangement according to the trigger rhythm, and the acquisition frequency configuration is converted into a transmission time table within the time window. The field retention granularity and numerical expression granularity in the expression precision configuration are read. The fields to be reported are selected from the calculation contribution sub-semantics according to the field retention granularity. Then, the numerical content in the fields to be reported is processed by format conversion and length adjustment according to the numerical expression granularity, and the expression precision configuration is converted into a transmission payload field set. The communication network assignment relationship is converted into a transmission channel identifier.

[0104] The acquisition end merges and sorts the transmission timetables of all computational contribution sub-semantics according to time sequence. At the same transmission time, the acquisition end performs queue orchestration on the computational contribution sub-semantics under the same communication network assignment relationship, obtains the queue orchestration relationship, and arranges them according to the priority order configured by the acquisition frequency and the load constraint relationship configured by the expression precision during the queue orchestration process. The acquisition end encapsulates the sorted transmission timetable, queue orchestration relationship, transmission channel identifier and transmission load field set in a structured manner to form a semantic division of labor reporting strategy.

[0105] Specifically, if the allowable timeframe defined by the freshness constraint corresponding to the calculated contribution sub-semantic is not less than the expected transmission and reception time span reflected by the timeframe record field under the current evaluation communication network, the timeframe constraint is deemed to be satisfied.

[0106] The priority order of the acquisition frequency configuration is obtained based on the timeliness requirement in the freshness constraint of each calculated contribution sub-semantic, with those having higher timeliness requirements being sent first.

[0107] S6: The computational contribution sub-semantics are sent in parallel according to the semantic division of labor reporting strategy. The computational contribution sub-semantics that fail to be sent are cached offline and retransmitted in order of freshness constraints. The processing end aligns and splices the computational contribution sub-semantics according to the dependency relationship and freshness constraints to reconstruct the continuous running state semantic sequence.

[0108] S6.1: Based on the semantic division of labor reporting strategy, the computational contribution sub-semantics are sent in parallel on the corresponding communication network, and the computational contribution sub-semantics that have not been sent are entered into the offline caching process.

[0109] Specifically, the acquisition end reads the communication network assignment relationship and transmission time arrangement corresponding to each computational contribution sub-semantic according to the semantic division of labor reporting strategy, and triggers transmission operations simultaneously on wired network, wireless WiFi network and cellular communication network. During the transmission process, the transmission status indicator is monitored. When it is detected that the computational contribution sub-semantic has not been transmitted within the scheduled transmission time, the corresponding computational contribution sub-semantic, together with freshness constraints and identification information, is written into the offline cache process.

[0110] The scheduled transmission time refers to the specific transmission time point generated by the semantic division of labor reporting strategy based on the collection frequency configuration and under a unified time reference, for each calculation contributing sub-semantic calculation.

[0111] The offline caching process refers to writing the calculated contribution sub-semantics, along with the identification information, generation time identifier, and freshness constraint, into the offline cache area and marking it as pending retransmission when the calculation of the contribution sub-semantics is not completed and sent at the scheduled time.

[0112] S6.2: Write the computational contribution sub-semantics that enter the offline caching process into the offline cache area, and perform supplementary transmission sorting on the computational contribution sub-semantics in the offline cache according to the freshness constraint.

[0113] Specifically, after detecting that the computational contribution sub-semantics has not been sent, the acquisition end writes the computational contribution sub-semantics, along with the corresponding freshness constraints, identification information, and generation time identifier, into the offline cache. In the offline cache, the computational contribution sub-semantics are sorted according to the time span field corresponding to the freshness constraints, prioritizing the arrangement according to the upper limit of the allowed time range from small to large. When multiple computational contribution sub-semantics correspond to the same upper limit of the time span, they are further sorted according to the generation time identifier, with the computational contribution sub-semantics generated earlier placed at the front of the retransmission queue.

[0114] S6.3: When the communication network is ready to transmit, read the calculated contribution sub-semantics from the offline buffer according to the retransmission order and perform the retransmission.

[0115] Specifically, the acquisition end periodically detects the connectivity, bandwidth, and latency of wired networks, wireless WiFi networks, and cellular communication networks. When a communication network is detected that meets the basic communication conditions for the current retransmission, the acquisition end reads the calculated contribution sub-semantics in the order of the retransmissions generated in the offline buffer, and triggers the transmission operation of the corresponding communication network according to the communication network assignment relationship associated with the calculated contribution sub-semantics, sending the calculated contribution sub-semantics from the offline buffer to the processing end.

[0116] It should be noted that the basic communication conditions refer to the communication network being in a communicable state within the current detection period, and the currently observed bandwidth and delay states meeting the requirements of the freshness constraint corresponding to the semantics of the calculated contribution sub-transmission on the transmission timeliness. The acquisition end determines whether the communication network has the conditions to perform supplementary transmission by matching the connectivity status identifier, bandwidth parameters, and delay parameters of the communication network with the timeliness adaptation description associated with the semantics of the calculated contribution sub-transmission.

[0117] S6.4: The processing end receives the computational contribution sub-semantics of parallel transmission and supplementary transmission, and organizes the computational contribution sub-semantics in order according to the dependency relationship between them.

[0118] Specifically, the processing end receives the computational contribution sub-semantics sent in parallel and in retransmission, reads the computational contribution sub-semantics identifier and dependency relationship identifier carried by each computational contribution sub-semantics, temporarily stores the computational contribution sub-semantics in the pending processing queue, and checks one by one, according to the dependency relationship identifier, whether the preceding dependency corresponding to each computational contribution sub-semantics has arrived and been received.

[0119] The processing end writes the received computational contribution sub-semantic identifiers into the received identifier set, and compares the pre-dependency identifiers of the computational contribution sub-semantic to be processed with the received identifier set. If all pre-dependency identifiers exist, it is determined that the process has been completed; if there are missing identifiers, it is determined that the process has not been completed. The computational contribution sub-semantic that has satisfied all pre-dependencies is added to the processable sequence and arranged in the order in which the dependencies between the computational contribution sub-semantices are satisfied.

[0120] S6.5: The processing end calculates the time difference between the receiving time of the calculated contribution sub-semantic and the time identifier attached during generation, and compares the time difference with the allowable timeliness range limited by the freshness constraint corresponding to the calculated contribution sub-semantic to determine whether the calculated contribution sub-semantic meets the timeliness requirements.

[0121] Specifically, the processing end reads the generation time identifier carried by the computational contribution sub-semantic and obtains the corresponding reception time; it performs time difference calculation on the reception time and the generation time identifier to obtain the actual transmission delay of the computational contribution sub-semantic; the processing end reads the allowable timeliness range limited in the freshness constraint corresponding to the computational contribution sub-semantic, and directly compares the actual transmission delay with the allowable timeliness range. When the actual transmission delay falls within the allowable timeliness range, it is determined that the computational contribution sub-semantic meets the timeliness requirement; when it exceeds the allowable timeliness range, it is determined that the computational contribution sub-semantic does not meet the timeliness requirement.

[0122] S6.6: Filter out computational contribution sub-semantics that do not meet the timeliness requirements, perform time alignment on the retained computational contribution sub-semantics according to the time identifier, and splice them in the dependency order to form a continuous running state semantic sequence.

[0123] Specifically, the freshness constraint and time identifier corresponding to each computational contribution sub-semantic are read one by one, and the time identifier is compared with the allowable timeliness range represented by the freshness constraint to identify computational contribution sub-semantic that does not meet the timeliness requirements and perform elimination processing; the computational contribution sub-semantic that has not been eliminated is rearranged and time-aligned according to the time identifier under a unified time reference, and the computational contribution sub-semantic arriving from different communication networks is aligned to a consistent position on the time axis. The time-aligned computational contribution sub-semantic is then spliced ​​and organized according to the order of the dependencies between the computational contribution sub-semantic, generating a continuous running state semantic sequence.

[0124] It should be noted that the allowable time limit is directly given by the freshness constraint written into the semantic constraint record field of the computational contribution sub-constraint. The allowable time limit is expressed in the form of time span, and the maximum acceptable time length from the generation of the computational contribution sub-semantic identifier to the completion of reception and entry into the processing flow is limited.

[0125] In summary, this invention, based on freshness constraints and multi-network capability profiles, jointly configures the acquisition frequency, expression accuracy, and communication network assignment relationships of computational contribution sub-semantics to form a semantic division of labor reporting strategy. This ensures that each computational contribution sub-semantics obtains a transmission channel and reporting rhythm matching the urgency of time within the allowable timeframe. Simultaneously, it sequentially organizes the sending timetable and queue arrangement relationships within the same time window, achieving controllable allocation of reporting density and single-load on the time axis. This reduces congestion and invalid retransmissions during multi-network parallel transmission, improves the operability of arrival order sorting and alignment splicing, and stably supports the reconstruction of semantic sequences in continuous operation.

[0126] 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, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for collecting a running state of a device supporting a multi-network environment, the method comprising the steps of: include, ​ The acquisition end is activated to detect available communication networks in parallel, obtain connectivity, bandwidth and latency characteristics, and form a multi-network capability profile. Under the constraint of multi-network capability profiling, the sampling device operation electrical signal is sampled and time-aligned, and the execution state discrimination and feature extraction are performed to obtain the initial operation state semantics; Contribution decomposability representation is constructed based on the initial running state semantics, and then broken down into sub-semantics for calculating contribution, and dependencies are established. Based on the characteristics of equipment operating status changes reflected by each calculated contribution sub-semantic at different sampling time scales, short-term and long-term freshness constraints are assigned. Based on freshness constraints and multi-network capability profiles, the collection frequency, expression accuracy, and communication network assignment relationships of the computational contribution sub-semantics are adapted in a timely manner, and a semantic division of labor reporting strategy is formed. The computational contribution sub-semantics are sent in parallel according to the semantic division of labor reporting strategy. The computational contribution sub-semantics that fail to be sent are cached offline and retransmitted in order of freshness constraints. The processing end aligns and splices the computational contribution sub-semantics according to the dependency relationship and freshness constraints to reconstruct the continuous running state semantic sequence.

2. The device operation status acquisition method supporting multiple network environments as described in claim 1, characterized in that: The communication network includes wired networks, wireless WiFi networks, and cellular communication networks.

3. The method of claim 2, wherein the device operation state collection method for supporting a multi-network environment is characterized by: The specific steps for forming a multi-network capability profile are as follows: Start the data acquisition terminal and perform availability tests on wired networks, wireless WiFi networks, and cellular communication networks respectively to establish a set of currently available communication networks; For each communication network in the communication network set, connectivity parameters, connectivity stability parameters, bandwidth parameters, and time delay parameters are collected to form a corresponding set of network capability parameters. The set of network capability parameters is structured, encapsulated, and normalized according to the communication network identifier, and a multi-network capability profile is output.

4. The method of claim 3, wherein the device operation state collection method for supporting a multi-network environment is characterized by: The specific steps for obtaining the initial running state semantics are as follows: Under the constraints of multi-network capability profiling, the device operation electrical signals generated during the operation of the device are sampled, the corresponding sampling timing is configured, and a time identifier is added to the sampled electrical signals. Time alignment processing is performed on the sampled electrical signals with additional time markers to form an electrical signal sequence under a unified time reference; The electrical signal sequence is processed for state discrimination and feature extraction, and the extracted feature information is organized into initial running state semantics according to the field structure.

5. The method of claim 4, wherein the device operation state collection method for supporting a multi-network environment is characterized by: The construction contribution can be decomposed and represented, and the specific steps are as follows. Read the state feature fields that carry feature information in the initial running state semantics, and divide them according to the role of the state feature information in running state discrimination to form feature parts; For the feature part, the computational dependency relationship is established in the process of operation state discrimination and state evolution analysis; The feature components and their corresponding computational dependencies are structured and organized to form a decomposable contribution representation.

6. The method of claim 5, wherein the device operation state collection method for supporting a multi-network environment is characterized by: The specific steps for establishing dependencies are as follows. Based on the contribution decomposable representation, the contribution decomposable representation is traversed according to the usage position, reference relationship and calculation order of state features in the operation state discrimination and state evolution analysis, and the contribution part that can participate in the calculation independently is extracted. The contribution components are mapped to corresponding computational contribution sub-semantics. Based on the computational order and reference relationships of the computational contribution sub-semantics in the process of runtime state discrimination and state evolution analysis, the dependency relationships between the computational contribution sub-semantics are established.

7. The method of claim 6, wherein the device operation state collection method for supporting a multi-network environment is characterized by: The specific process for allocating short-term and long-term freshness constraints is as follows. For each calculated contribution sub-semantic, corresponding state change sequences are constructed under both short-time sampling timescales and long-time sampling timescales; Based on the state change sequence, extract change feature information that characterizes the consistency and degree of deviation of changes across time scales; Generate and assign an index to each computational contribution sub-semantic based on the change feature information; Based on the allocation index, each computational contribution sub-semantic is mapped to the corresponding short-term and long-term freshness constraints.

8. The method of claim 7, wherein the device operation state collection method for supporting a multi-network environment is characterized by: The specific process for forming a semantic division of labor reporting strategy is as follows. For each computational contribution sub-semantic, the corresponding freshness constraint is associated with the latency parameter, bandwidth parameter and connectivity stability parameter in the multi-network capability profile to form a time-adaptive description of the computational contribution sub-semantic. Based on the time-adaptation description, the collection frequency configuration, expression precision configuration, and communication network assignment relationship of the computational contribution sub-semantics are adjusted synchronously during the same configuration process. Based on the acquisition frequency configuration, expression accuracy configuration, and communication network assignment relationship, the semantic execution order of all calculated contribution sub-semantics is organized within the same time window to form a semantic division of labor reporting strategy.

9. The method of claim 8, wherein the device operation state collection method for supporting a multi-network environment is characterized by: The specific process of performing offline caching and retransmitting the calculated contribution sub-semantics for failed transmissions, sorted by freshness constraints, is as follows. Based on the semantic division of labor reporting strategy, the computational contribution sub-semantics are sent in parallel on the corresponding communication network, and the computational contribution sub-semantics that have not been sent are entered into the offline caching process. The computational contribution sub-semantics that enter the offline caching process are written into the offline cache area, and the computational contribution sub-semantics in the offline cache are sorted according to the freshness constraint. When the communication network is ready to transmit, the contribution sub-semantics are read from the offline buffer according to the retransmission order and the retransmission is performed.

10. The method of claim 9, wherein the device operation state collection method for supporting a multi-network environment is characterized by: The specific process for reconstructing the semantic sequence of continuous operating states is as follows. The processing end receives the computational contribution sub-semantics of parallel transmission and retransmission, and organizes the computational contribution sub-semantics in order according to the dependency relationship between them. The processing end calculates the time difference between the receiving time of the calculated contribution sub-semantics and the time identifier attached during generation, and compares the time difference with the allowable timeliness range limited by the freshness constraint corresponding to the calculated contribution sub-semantics to determine whether the calculated contribution sub-semantics meets the timeliness requirements. Computational contribution sub-semantics that do not meet the timeliness requirements are filtered out. The remaining computational contribution sub-semantics are then aligned by time according to the time identifier and spliced ​​together in the order of dependency to form a continuous running state semantic sequence.

Images