Industrial internet of things based device dynamic information collection method and system

Abstract

This application discloses a method and system for acquiring dynamic information of equipment based on the Industrial Internet of Things (IIoT), relating to the field of IIoT technology. Specifically targeting the dynamic information acquisition scenario of factory industrial equipment, this application improves upon existing fixed-frequency acquisition technology by implementing time-block scheduling and dual-channel parallel discrimination. In state discrimination, it considers the relationship between steady-state trend shifts and transient abrupt changes in physical quantities. Through joint decision-making via steady-state feature analysis and transient disturbance detection, it dynamically adjusts the sampling frequency and sensor channel activation strategy for the next time block. It integrates cloud-based and field-based collaborative prediction into the acquisition method and employs seamless switching between dual buffers to ensure data continuity, providing a high-fidelity, energy-saving industrial equipment state perception solution adaptable to all operating conditions.

Description

Technical Field

[0001] This application relates to the field of industrial Internet of Things (IoT) technology, and in particular to a method and system for collecting dynamic information of devices based on industrial IoT. Background Technology

[0002] In modern intelligent manufacturing systems, the Industrial Internet of Things (IIoT) demands increasingly sophisticated perception of the operational status of target equipment. Real-time and accurate acquisition of dynamic equipment information is crucial for ensuring production safety, optimizing energy efficiency, and supporting predictive maintenance. Traditional industrial data acquisition systems employ fixed-frequency polling sampling mechanisms, which, while meeting the needs of early steady-state monitoring, have inherent limitations.

[0003] This type of solution cannot simultaneously guarantee data integrity and resource efficiency. High-frequency sampling during steady-state operation generates a large amount of redundant data, increasing bandwidth, storage, and energy consumption pressure. During transient events, insufficient sampling frequency easily misses key features, leading to anomaly detection failure. Furthermore, it lacks a mechanism for coordinating steady-state and transient data discrimination, and lacks a closed-loop logic of "perception-discrimination-feedback-optimization," resulting in lagging strategy adjustments and difficulty in adapting to dynamic operating conditions of equipment. Therefore, an adaptive dynamic data acquisition method is urgently needed to overcome the current technical bottlenecks, becoming a pressing problem to be solved in this field. Summary of the Invention

[0004] In order to at least overcome the above-mentioned shortcomings in the prior art, the purpose of this application is to provide a method and system for dynamic information acquisition of equipment based on the Industrial Internet of Things to solve the above problems.

[0005] Firstly, this application provides a method for collecting dynamic information about devices based on the Industrial Internet of Things (IIoT), including: Sampling is performed on the target device according to the sampling frequency in effect for the current time block to obtain the first raw data sequence; In response to acquiring the first raw data sequence, the first processing channel obtains the steady-state discrimination result of the target device operation based on the first raw data sequence; In response to acquiring the first raw data sequence, the second processing channel obtains the transient discrimination result of the target device operation based on the first raw data sequence; Based on the steady-state and transient discrimination results, the acquisition strategy for the next time block is obtained; According to the acquisition strategy, the updated sampling frequency is obtained, and when entering the next time block, sampling is performed on the target device according to the updated sampling frequency.

[0006] In one possible implementation, based on the acquisition strategy, an updated sampling frequency is obtained, and upon entering the next time block, sampling is performed on the target device according to the updated sampling frequency, including: Based on the acquisition strategy of the next time block, obtain the instruction control set; According to the preset downlink, the instruction control set is broadcast to all nodes; The node is configured with one or more sensors that perform sampling on the target device; The sensor obtains the updated sampling frequency according to the instruction control set; After sampling the target device at the sampling frequency that is effective for the current time block, the sampling frequency is configured to the updated sampling frequency before entering the next time block. When entering the next time block, the target device is sampled according to the updated sampling frequency.

[0007] In one possible implementation, when entering the next time block, the target device is sampled according to the updated sampling frequency to obtain a second original data sequence; The first and second original data sequences are aligned on the time axis. The first original data sequence and the second original data sequence constitute historical data; At the start of the current time block, sampling is performed on the target device according to the sampling frequency in effect for the current time block to obtain the first raw data sequence, which is then stored in the main buffer. At the start of the next time block, the target device is sampled according to the update sampling frequency to obtain a second raw data sequence, which is then stored in a spare buffer. In response to the state determination based on the first raw data sequence within the current time block and to determine the acquisition strategy for the next time block, the main buffer and the backup buffer are swapped.

[0008] In one possible implementation, in response to acquiring a first raw data sequence, the first processing channel obtains a steady-state discrimination result of the target device operation based on the first raw data sequence, including: In response to acquiring the first raw data sequence, a parallel dual-channel state discrimination process is initiated; The dual channels are configured with a first processing channel; The first processing channel is configured to operate in timer-driven mode; Steady-state characteristic analysis is performed once every at least one time block to determine whether the target device is in steady-state operation; Based on whether it is in steady-state operation, the steady-state discrimination result of the target equipment operation is obtained.

[0009] In one possible implementation, in response to acquiring a first raw data sequence, the first processing channel obtains a steady-state discrimination result of the target device operation based on the first raw data sequence, including: Based on the first original data sequence, the mean and standard deviation of each physical quantity in the current time block are obtained to obtain the coefficient of variation. If the coefficients of variation of all physical quantities are less than the preset steady-state threshold, and no trend shift occurs within at least one consecutive time block, the equipment is determined to be in a steady-state operating state. Otherwise, the equipment is determined to be in an unsteady operating state; The steady-state and non-steady-state operating states are used as the steady-state discrimination results of the target equipment operation.

[0010] In one possible implementation, in response to acquiring a first raw data sequence, the second processing channel obtains a transient discrimination result of the target device operation based on the first raw data sequence, including: In response to acquiring the first raw data sequence, a parallel dual-channel state discrimination process is initiated; The dual channels are configured with a second processing channel; The second processing channel is configured to run in interrupt service routine triggered mode; In response to the detection that any physical quantity in the first raw data sequence exceeds a local extremum, transient change event identification is immediately initiated to determine whether a transient change has occurred in the target device; Based on whether a transient change occurs, the transient discrimination result of the target equipment operation is obtained.

[0011] In one possible implementation, in response to acquiring a first raw data sequence, the second processing channel obtains a transient discrimination result of the target device operation based on the first raw data sequence, including: Based on the first original data sequence, obtain all sampled values ​​of each physical quantity in the current time block to obtain the mean value in the local neighborhood. Based on historical data from steady-state operation, obtain the historical standard deviation of each physical quantity; If the difference between any sampled value of any physical quantity and the mean of the corresponding local neighborhood is greater than a preset multiple of the historical standard deviation, it is determined that the physical quantity has a sudden change point at this time. If any physical quantity has at least one abrupt change point, a transient abrupt change signal is triggered; Based on whether a transient change signal is triggered, the transient discrimination result of the target device operation is obtained.

[0012] In one possible implementation, if any physical quantity has at least one abrupt change point, a transient abrupt change signal is triggered, and the implementation also includes a multidimensional physical quantity correlation test. If at least two physical quantities are detected to have a sudden change within a set time window, and the direction of change of the physical quantity that has a sudden change matches any entry in the preset fault library of transient events, then the confidence level of the transient change that has occurred at this time is increased. If the confidence level of a transient change occurs at this time is greater than the confidence level of a transient change for the corresponding entry obtained from historical data of steady-state operation, then a transient change signal is triggered.

[0013] In one possible implementation, obtaining the acquisition strategy for the next time block based on the steady-state discrimination result and the transient discrimination result includes: The energy-saving mode acquisition strategy is activated only if the first processing channel determines that the target device is in steady-state operation and the second processing channel does not trigger any transient changes. The energy-saving mode acquisition strategy is configured to: enable a preset base frequency at the sampling frequency of the next time block, and implement intermittent sleep control for non-critical sensor channels; When any channel outputs a non-steady-state signal, the enhanced mode acquisition strategy is activated. The enhanced mode acquisition strategy is configured to enable a preset high-density frequency at the sampling frequency of the next time block, and simultaneously enable a multi-sensor channel collaboration mechanism.

[0014] In one possible implementation, the multi-sensor channel collaboration mechanism is configured with a three-level priority mapping table.

[0015] The three-level priority mapping table includes: When only the first processing channel outputs non-steady-state operation and the second processing channel does not respond, the sampling frequency of the next time block will be set to a preset medium sampling frequency. When only the second processing channel triggers any transient change, while the first processing channel still maintains a steady-state output, in addition to increasing the sampling frequency to enable high-density frequency, it is also necessary to expand the range of sensor channels for acquisition to include sensor channels that were originally in intermittent sleep control. When both indicate an anomaly, the first processing channel outputs non-steady-state operation, and the second processing channel triggers any transient change, the highest level acquisition strategy is fully activated. The sampling frequency of the next time block is set to high density frequency, and all channels of the sensor channel are set to high density frequency for sampling, continuous recording of the original waveform, and local buffer pre-allocation operation.

[0016] In one possible implementation, step S5 is also included.

[0017] The cloud-based storage platform obtains a transfer matrix based on historical data and sends it to the computing layer periodically.

[0018] Based on historical data, a state prediction auxiliary mechanism is set up to obtain historical state transition probabilities, which can be used to predict in advance the operating range that the equipment may enter.

[0019] The state prediction assistance mechanism utilizes a cloud-on-site collaboration mechanism to periodically calculate the state transition matrix from the received historical data, which is the empirical probability of the device transitioning from state A to state B. Then, after each state determination, the calculation layer queries the transition matrix returned to it by the cloud in conjunction with the current state. If it finds that the probability of transitioning to a high-risk state exceeds the third threshold, it activates the preventive enhancement mode collection strategy in advance. The high-risk state is the mutation point where any transient change may occur, avoiding the loss of critical data due to response delay.

[0020] Secondly, this application provides a device dynamic information acquisition system based on the Industrial Internet of Things, including a data acquisition unit, a feature discrimination unit, and a strategy generation unit connected in sequence. The data acquisition unit is configured to: perform sampling on the target device according to the sampling frequency in effect of the current time block, and the computing layer obtains the first original data sequence; The feature discrimination unit is configured to: in response to acquiring a first original data sequence, the first processing channel obtains a steady-state discrimination result of the target device operation based on the first original data sequence; The feature discrimination unit is further configured to: in response to acquiring the first original data sequence, the second processing channel obtains the transient discrimination result of the target device operation based on the first original data sequence; The strategy generation unit is configured to obtain the acquisition strategy for the next time block based on the steady-state discrimination result and the transient discrimination result.

[0021] In summary, the beneficial effects that this application can achieve are: This application proposes a dual-channel parallel discrimination method, which uses a timer-driven steady-state analysis and interrupt-triggered transient detection to solve the problem that traditional acquisition methods cannot simultaneously address steady-state redundancy and transient missed detection, thus achieving target equipment operating status identification. The proposed acquisition strategy and seamless switching between dual buffers, through state feedback to adjust the acquisition frequency and sensor activation mode, solve the problems of data interruption and resource waste, achieving energy-saving and uninterrupted data acquisition. The proposed cloud-and-field collaborative prediction mechanism, which generates a state transition matrix from historical data, solves the problem of acquisition response lag and enables preventative strategy adjustments. The method in this application represents a significant technological advancement and has beneficial effects in improving the efficiency and accuracy of industrial equipment information acquisition. Attached Figure Description

[0022] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the method steps in Embodiment 1 of this application; Figure 2This is a schematic diagram of the method steps in Embodiment 2 of this application; Figure 3 This is a schematic diagram of the system structure of Embodiment 3 of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the accompanying drawings in this application are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this application. Furthermore, it should be understood that the schematic drawings are not drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of this application. It should be understood that the operations in the flowcharts may not be implemented in sequence, and steps without logical contextual relationships may be reversed or implemented simultaneously. In addition, those skilled in the art, guided by the content of this application, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts.

[0024] Furthermore, the described embodiments are merely some, not all, of the embodiments of this application. The components of the embodiments of this application described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0025] Example 1 Please refer to the following: Figure 1 , Figure 1 This is a schematic diagram of the steps of the device dynamic information acquisition method based on the Industrial Internet of Things provided in the embodiment of the present invention. Further, the device dynamic information acquisition method based on the Industrial Internet of Things may specifically include the contents described in steps S1 to S3.

[0026] Step S1: Perform data collection on the target device according to the collection frequency that is effective in the current time block, and the computing layer obtains the first raw data sequence; Step S21: In response to acquiring the first raw data sequence, the first processing channel obtains the steady-state discrimination result of the target device operation based on the first raw data sequence; Step S22: In response to acquiring the first raw data sequence, the second processing channel obtains the transient discrimination result of the target device operation based on the first raw data sequence; Step S3: Based on the steady-state discrimination result and the transient discrimination result, obtain the acquisition strategy for the next time block.

[0027] This application embodiment constructs a dynamic information acquisition system for industrial equipment in factories with dual-channel parallel discrimination. It can introduce closed-loop feedback adjustment and seamless switching structure of buffer zone, which is easy to expand. This embodiment divides the operating state of the target equipment into two modes: steady state and transient state. Based on this, the acquisition frequency and sensor activation strategy are adjusted in real time to achieve high-fidelity, low-overhead, and uninterrupted equipment status acquisition data in the entire operating range.

[0028] First, data is collected from the target device according to the collection frequency effective for the current time block to obtain a first raw data sequence. For example, a fixed collection frequency can be preset, such as collecting data once per second, and data collection can be continuously performed at this frequency within each time block, maintaining this initial collection frequency throughout the entire time block. In this way, continuous operational data of the device within a specific time period can be obtained.

[0029] Then, the operating status of the target device is determined through parallel dual-channel analysis.

[0030] In response to acquiring the first raw data sequence, the first processing channel obtains a steady-state determination result for the target device's operation based on the first raw data sequence. The first processing channel can perform periodic analysis on the collected data, setting a simple statistical analysis at regular intervals and comparing it with a preset fixed threshold. If the data fluctuation is within the threshold range, it is determined to be in steady-state operation; otherwise, it is determined to be in non-steady-state operation. Alternatively, the first processing channel can employ a rule-based expert system, which performs logical judgments on the data based on predefined target device operating rules and empirical knowledge to derive a steady-state determination result.

[0031] Simultaneously, in response to acquiring the first raw data sequence, the second processing channel obtains the transient discrimination result of the target device operation based on the first raw data sequence. The second processing channel can be configured to monitor the data in real time, and a fixed data change rate threshold can be set. When the acquired value of any physical quantity exceeds the change rate threshold within a short period of time, it is determined that a transient change has occurred, triggering the transient discrimination result.

[0032] Finally, based on the steady-state and transient discrimination results, the acquisition strategy for the next time block is obtained. A simple logical judgment table can be used. If the steady-state discrimination result is steady-state and the transient discrimination result does not show a sudden change, the acquisition strategy for the next time block can maintain the current acquisition frequency. If either discrimination result shows a non-steady-state or sudden change, the acquisition strategy for the next time block can increase the acquisition frequency to a preset higher frequency.

[0033] This embodiment introduces a first processing channel and a second processing channel to perform parallel discrimination of the steady-state and transient operating states of the target device. This overcomes the limitation of traditional fixed acquisition strategies in industrial IoT environments, which cannot balance data integrity and resource efficiency. It can dynamically adjust the acquisition strategy according to the actual operating state of the device, avoid the generation of redundant data in the steady state, and improve the ability to capture transient events, thereby optimizing the efficiency and responsiveness of data acquisition to a certain extent.

[0034] Example 2 Based on Example 1, please refer to the following: Figure 2 , Figure 2 This is a schematic diagram of the steps of the device dynamic information acquisition method based on the Industrial Internet of Things provided in the embodiment of the present invention. Further, the device dynamic information acquisition method based on the Industrial Internet of Things may specifically include the contents described in steps S1-S5.

[0035] Step S1: Perform data collection on the target device according to the collection frequency that is effective in the current time block, and the computing layer obtains the first raw data sequence.

[0036] Step S21: In response to acquiring the first raw data sequence, the first processing channel obtains the steady-state discrimination result of the target device operation based on the first raw data sequence; Step S22: In response to acquiring the first raw data sequence, the second processing channel obtains the transient discrimination result of the target device operation based on the first raw data sequence.

[0037] Step S3: Based on the steady-state discrimination result and the transient discrimination result, obtain the acquisition strategy for the next time block.

[0038] Step S4: According to the acquisition strategy, obtain the update sampling frequency, and when entering the next time block, perform sampling on the target device according to the update sampling frequency.

[0039] Step S5: The cloud-based storage platform obtains the transfer matrix based on historical data and periodically sends it to the computing layer.

[0040] In the implementation of this application embodiment, a system is constructed that features dual-channel parallel device status discrimination, acquisition strategy mapping rules, closed-loop feedback adjustment mechanism, seamless buffer switching structure, scheduling method, cloud and field collaboration, and this method is a method for dynamic information acquisition of industrial equipment.

[0041] Then, the sensing layer is configured with sensors based on the actual factory equipment. Both the sensing and computing layers are deployed directly on-site in the factory, which reduces costs. The storage platform is deployed in the cloud, and the sensing and computing layers are effectively connected via the Internet of Things (IoT). Therefore, in a typical deployment scenario of this application, the target equipment is a CNC machine tool operating in a continuous production process. Its key operating parameters include spindle vibration acceleration, cutting force, motor current, and temperature distribution. Multiple industrial IoT nodes are then deployed around this target equipment. Each node integrates different types of sensor channels, and each channel contains several sensors used to collect the aforementioned physical quantities. For example, for data on lubricating oil collected only at a certain location, multiple different types of sensor channels are set up, with multiple sensors deployed on each channel according to the collection location. Since the object of collection here is only lubricating oil, a node is set up at this location.

[0042] All nodes are connected to the computing layer located in the workshop via the Internet of Things (IoT). The computing layer then interacts with the cloud storage platform via the IoT for periodic data synchronization and transfer matrix updates.

[0043] First, within the current time block, a data acquisition operation is performed on the target device according to the acquisition frequency in effect for this time block, and the computing layer obtains the first raw data sequence.

[0044] Then, a dual-channel parallel discrimination process is initiated, wherein the first processing channel performs steady-state feature analysis based on the first original data sequence to output steady-state discrimination results, and the second processing channel simultaneously performs transient disturbance detection based on the first original data sequence to output transient discrimination results.

[0045] Then, the acquisition strategy for the next time block is determined based on the steady-state discrimination result and the transient discrimination result.

[0046] Finally, an instruction control set is generated based on the acquisition strategy and broadcast to all nodes through a preset downlink. Each node is configured with one or more sensors for acquiring data from the target device. After receiving the instruction control set, the sensor updates its acquisition frequency. After completing the acquisition task of the current time block and before entering the next time block, the sensor is configured with the updated acquisition frequency, and at the beginning of the next time block, the target device is acquired based on the updated acquisition frequency to obtain the second raw data sequence.

[0047] Furthermore, the first and second original data sequences are strictly aligned on the timeline and used as historical data to ensure that the entire acquisition process is continuous and without information loss.

[0048] In step S1, the target device is sampled according to the sampling frequency in effect for the current time block, and the computing layer obtains the first raw data sequence.

[0049] In the implementation of this application embodiment, the entire process of collecting dynamic information of the target device is performed. A fixed-length time block is pre-defined as a basic scheduling unit based on different industrial devices. This method is performed periodically according to the basic scheduling unit. Within each time block, the collection period can be obtained based on the current collection frequency. A fixed-length time period is set as the basic scheduling unit, denoted as a time block. Within each time block, the computing layer completes one round of data sequence acquisition, running status determination, strategy generation, and instruction issuance operations.

[0050] Let the time block length be The initial sampling frequency is set to the base frequency. .

[0051] In the current time block At the beginning, k is any natural number, and each node is determined according to... The effective acquisition frequency is used to perform synchronous acquisition on the target device to obtain the first raw data sequence. The first original data sequence It includes all blocks in the current time. Activated sensor channels in time interval The original value of the sampling frequency that is effective in the current time block.

[0052] Step S21: In response to acquiring the first raw data sequence, the first processing channel obtains the steady-state discrimination result of the target device operation based on the first raw data sequence; Step S22: In response to acquiring the first raw data sequence, the second processing channel obtains the transient discrimination result of the target device operation based on the first raw data sequence.

[0053] When this application embodiment is implemented, in the current time block Internally, the computation layer initiates a dual-channel parallel discrimination process. The first processing channel operates in timer-driven mode, performing a complete steady-state feature extraction and trend modeling every K time blocks; the second processing channel is configured in interrupt service routine triggered mode to respond in real time to sudden signals. Both channels use the first raw data sequence. It serves as an input source, but its processing logic and triggering conditions are completely different.

[0054] In step S21, in response to acquiring the first raw data sequence, the first processing channel obtains the steady-state discrimination result of the target device operation based on the first raw data sequence, including: In response to acquiring the first raw data sequence, a parallel dual-channel state discrimination process is initiated; The dual channels are configured with a first processing channel; The first processing channel is configured to operate in timer-driven mode; Steady-state characteristic analysis is performed once every at least one time block to determine whether the target device is in steady-state operation; Based on whether it is in steady-state operation, the steady-state discrimination result of the target equipment operation is obtained.

[0055] In the implementation of this application embodiment, the first processing channel operates in time-driven mode. Within each time block, steady-state characteristic analysis is performed periodically. Each time the timer is set, a response is made, and then a steady-state judgment is performed. This is also to analyze the long-term trend characteristics of the target device's operation.

[0056] In one possible implementation, in response to acquiring a first raw data sequence, the first processing channel obtains a steady-state discrimination result of the target device operation based on the first raw data sequence, including: Based on the first original data sequence, the mean and standard deviation of each physical quantity in the current time block are obtained to obtain the coefficient of variation. If the coefficients of variation of all physical quantities are less than the preset steady-state threshold, and no trend shift occurs within at least one consecutive time block, the equipment is determined to be in a steady-state operating state. Otherwise, the equipment is determined to be in an unsteady operating state; The steady-state and non-steady-state operating states are used as the steady-state discrimination results of the target equipment operation.

[0057] In the implementation of this application embodiment, the trend deviation is determined by whether the absolute value of the linear regression slope exceeds a preset trend deviation threshold; if the absolute value of the slope is greater than the trend deviation threshold, it is determined that the target device has slow performance degradation or operating condition drift.

[0058] right Each physical quantity in Sliding window statistical analysis is applied to its corresponding time subsequence. The window length is set to... Step size is Covers the entire current time block Then, for each window location, calculate all physical quantities within that window. mean Standard deviation Then the coefficient of variation is calculated. .

[0059] If the coefficients of variation of all physical quantities are less than the preset first steady-state threshold If the condition is met, the primary condition for steady-state operation is satisfied; if the condition is greater than or equal to the preset first steady-state threshold, it is directly determined to be unsteady-state operation.

[0060] When the coefficients of variation of all physical quantities are less than the preset first steady-state threshold, the first processing channel also needs to verify whether there is a slow trend shift.

[0061] For each physical quantity Perform a univariate linear fit on historical data from the most recent M consecutive time blocks, including the current time block. A univariate linear regression model built using historical data can be expressed as: ; in, It is the regression slope. It is the intercept. Residual term.

[0062] Then the absolute value of the regression slope offset threshold from the preset first trend Compare, if for all i... Then determine the device in the current time block. It is in a steady-state operation; otherwise, it is determined to be unsteady because the target equipment is now experiencing slow performance degradation or operating condition drift.

[0063] This yields the steady-state discrimination result of the target equipment's operation, and the steady-state discrimination result is denoted as... , where 1 represents steady-state operation and 0 represents unsteady-state operation.

[0064] In step S22, in response to acquiring the first raw data sequence, the second processing channel obtains the transient discrimination result of the target device operation based on the first raw data sequence, including: In response to acquiring the first raw data sequence, a parallel dual-channel state discrimination process is initiated; The dual channels are configured with a second processing channel; The second processing channel is configured to run in interrupt service routine triggered mode; In response to the detection that any physical quantity in the first raw data sequence exceeds a local extremum, transient change event identification is immediately initiated to determine whether a transient change has occurred in the target device; Based on whether a transient change occurs, the transient discrimination result of the target equipment operation is obtained.

[0065] In the implementation of this application embodiment, the second processing channel operates in an interrupt service routine triggered mode. Within each time block, as long as any physical quantity in the first original data sequence exceeds the local extreme value, the interrupt service routine responds and immediately performs a transient discrimination.

[0066] In one possible implementation, in response to acquiring a first raw data sequence, the second processing channel obtains a transient discrimination result of the target device operation based on the first raw data sequence, including: Based on the first raw data sequence, obtain all the collected values ​​of each physical quantity in the current time block to obtain the mean value in the local neighborhood. Based on historical data from steady-state operation, obtain the historical standard deviation of each physical quantity; If the difference between any collected value of any physical quantity and the mean value in the corresponding local neighborhood is greater than a preset multiple of the historical standard deviation, it is determined that there is a sudden change point in the physical quantity at this time. If any physical quantity has at least one abrupt change point, a transient abrupt change signal is triggered; Based on whether a transient change signal is triggered, the transient discrimination result of the target device operation is obtained.

[0067] In one possible implementation, if any physical quantity has at least one abrupt change point, a transient abrupt change signal is triggered, and the implementation also includes a multidimensional physical quantity correlation test. If at least two physical quantities are detected to have a sudden change within a set time window, and the direction of change of the physical quantity that has a sudden change matches any entry in the preset fault library of transient events, then the confidence level of the transient change that has occurred at this time is increased. If the confidence level of a transient change occurs at this time is greater than the confidence level of a transient change for the corresponding entry obtained from historical data of steady-state operation, then a transient change signal is triggered.

[0068] In the implementation of this application embodiment, the transient discrimination process of the second processing channel focuses on local mutation detection, mainly in order to respond to sudden, short-duration abnormal events in the operation of the target device.

[0069] right Each physical quantity in traverse all its collection points , calculation The mean value within a local neighborhood formed by N sampling points before and after the center. The mutation criterion can be expressed as:

[0070] in, It is the standard deviation obtained from the historical steady-state operation statistics of this physical quantity. It is a preset multiplier factor, ranging from 3 to 5.

[0071] If there is at least one collection point If the above inequality is satisfied, then mark the physical quantity in the current time block. A sudden change occurs within, triggering a transient change signal.

[0072] In this embodiment, before triggering the transient mutation signal, the correlation test of multidimensional physical quantities can be introduced in the second processing channel after the mutation has occurred.

[0073] If within the time window Within a given context, m physical quantities simultaneously exhibit abrupt changes. If the direction of change (rising or falling) matches a certain entry in the preset fault coupling mode library, the confidence of the transient event is increased. A typical example is the detection of a sudden increase in vibration acceleration accompanied by a sudden drop in cutting force, which corresponds to the tool breakage mode of the target equipment. At this time, the confidence is immediately increased.

[0074] If the confidence level at this time is higher than the confidence level obtained from historical data of steady-state operation for the occurrence of a transient change in the corresponding entry, then a transient change signal is triggered.

[0075] Finally, the second processing channel outputs the transient discrimination result. 1 indicates no transient mutation, and 0 indicates transient mutation.

[0076] In step S3, obtaining the acquisition strategy for the next time block based on the steady-state discrimination result and the transient discrimination result includes: The energy-saving mode acquisition strategy is activated only if the first processing channel determines that the target device is in steady-state operation and the second processing channel does not trigger any transient changes. The energy-saving mode acquisition strategy is configured to: enable a preset base frequency at the acquisition frequency of the next time block, and implement intermittent sleep control for non-critical sensor channels; When any channel outputs a non-steady-state signal, the enhanced mode acquisition strategy is activated. The enhanced mode acquisition strategy is configured to enable a preset high-density frequency in the acquisition frequency of the next time block, and simultaneously enable a multi-sensor channel collaboration mechanism to ensure data redundancy for cross-validation in the spatial dimension.

[0077] In one possible implementation, the multi-sensor channel collaboration mechanism is configured with a three-level priority mapping table.

[0078] The three-level priority mapping table is used to refine the different acquisition intensity levels of the enhanced mode acquisition strategy under different combinations of conditions. The three-level priority mapping table is defined by the following rules: When only the first processing channel outputs non-steady-state operation and the second processing channel does not respond, the sampling frequency of the next time block will be set to the preset medium sampling frequency. When only the second processing channel triggers any transient change, while the first processing channel continues to operate in an unsteady state, in addition to increasing the acquisition frequency to enable high-density frequency, it is also necessary to expand the range of sensor channels for acquisition to include sensor channels that were originally in intermittent sleep control. When both indicate an anomaly, the first processing channel outputs non-steady-state operation, and the second processing channel triggers any transient change, the highest level acquisition strategy is fully activated. The acquisition frequency of the next time block is set to high density frequency, and all channels of the sensor channel are set to high density frequency for acquisition, continuous recording of raw waveforms, and local buffer pre-allocation.

[0079] In the implementation of this application embodiment, based on the steady-state discrimination result and the transient discrimination result, that is, based on... and The combination of these elements yields the next time block. The data collection strategy and its decision-making logic can be represented by four cases: Scenario 1: When the system is in steady-state operation without any transient changes, the energy-saving mode acquisition strategy is activated.

[0080] The sampling frequency for the next time block is set to the base frequency. Meanwhile, key channels remain continuously activated, according to Continuous operation is performed, and intermittent sleep control is implemented for non-critical sensor channels. Taking the target device in this embodiment as an example, non-critical sensor channels include ambient temperature and humidity, and auxiliary coolant flow rate, while critical sensor channels include spindle vibration and motor current.

[0081] Intermittent sleep control is in the next time block Within this system, the power is turned off at set acquisition cycles, and the system is only woken up in specific cycles to maintain a minimum level of state awareness for non-critical sensor channels. For example, in this embodiment, the system is only woken up when entering the next time block. The first cycle activates non-critical sensor channels to acquire data, and then puts them into a dormant state during that time block.

[0082] Situations two through four all involve activating the enhanced mode acquisition strategy.

[0083] In the enhanced mode, high-density frequency acquisition is always enabled. In this embodiment, a medium acquisition frequency is also set. The high-density frequency includes the medium acquisition frequency. and high density frequency .

[0084] Scenario 2: This refers to non-steady-state operation without transient changes, corresponding to slow-evolution anomalies. Taking the target device in this embodiment as an example, slow-evolution anomalies include early wear or thermal drift of the bearing.

[0085] The sampling frequency for the next time block is set to , .

[0086] Simultaneously, the multi-sensor channel collaboration mechanism is configured to: maintain continuous activation of all activated key sensor channels, according to... It operates continuously, but does not extend the range of sensor channels.

[0087] Scenario 3: This means that the system is in steady state but has transient changes, such as cutting impact or external vibration disturbance.

[0088] The sampling frequency for the next time block is set to a high-density frequency. .

[0089] At this time, the multi-sensor channel collaboration mechanism is configured as follows: expand the range of acquisition sensor channels, temporarily activate non-critical sensor channels that were originally in a dormant state, such as non-critical sensor channels of acoustic emission sensors and infrared thermal imagers, and set them for the next time block. Internal data collection strategies to enhance coverage of emergencies.

[0090] Scenario 4: This refers to non-steady-state operation with transient changes, which is a compound anomaly. In this case, the target equipment has both long-term performance degradation and a transient change event. For example, the bearing has already worn out and the cutting impact is too large.

[0091] The sampling frequency for the next time block is set to a high-density frequency. .

[0092] At this time, the multi-sensor channel collaboration mechanism is configured as follows: a full activation of the highest-level acquisition strategy is adopted: all sensor channels are... Frequency acquisition; raw data is written directly to the local cache of the sensing layer without compression; At the same time, the computing layer sends cache allocation instructions to each node, reserving additional local storage space to cope with the raw data collected from abnormal events that may last for multiple time blocks, and uploading the raw data of each time block and the first raw data sequence to the cloud storage platform.

[0093] After each successful upload of the original data of the current time block and the first original data sequence to the cloud storage platform, the data of the earliest time block stored in the local storage space is deleted.

[0094] In step S4, according to the acquisition strategy, the updated sampling frequency is obtained, and when entering the next time block, sampling is performed on the target device according to the updated sampling frequency, including: Based on the acquisition strategy of the next time block, obtain the instruction control set; According to the preset downlink, the instruction control set is broadcast to all nodes; The node is configured with one or more sensors that perform data acquisition on the target device; The sensor obtains the update sampling frequency according to the instruction control set; After collecting data from the target device at the collection frequency that is effective for the current time block, the perception layer is configured to update the collection frequency before entering the next time block. When entering the next time block, the target device is collected according to the updated collection frequency.

[0095] In the implementation of this application embodiment, the computing layer generates an instruction control set according to the acquisition strategy of the next time block it generates. Command and control set At least including: the target acquisition frequency of the sensor channel The target acquisition frequency of the sensor channel is determined based on the specific acquisition strategy. The specific value, that is base frequency Medium sampling frequency or high density frequency Any one of the sensor channel activation lists This requires explicitly listing the next time block. For the sensor channels that require internal operation, active channels are set to 1, and inactive channels are set to 0. The local storage space of the computing layer is allocated to the cache size of different sensor channels. .

[0096] Then the instruction control set of the computing layer It is broadcast to all nodes in the perception layer via the downlink.

[0097] Each node in the perception layer has a built-in instruction parsing and execution unit. It receives instruction control sets. Then, the node first verifies its time and only receives data for the next time block. The instructions.

[0098] Then, based on the internal synchronization clock, in the current time block Before ending, complete the configuration update and match the acquisition clock to the target acquisition frequency of the sensor channel. According to the active list of sensor channels Control the power switches of each sensor channel.

[0099] Ensure in the next time block The new strategy can be implemented from the start.

[0100] In one possible implementation, when entering the next time block, the target device is sampled according to the update sampling frequency to obtain a second raw data sequence; The first and second original data sequences are aligned on the time axis. The first original data sequence and the second original data sequence constitute historical data; At the start of the current time block, data is collected from the target device according to the collection frequency in effect for the current time block, the first raw data sequence is obtained, and stored in the main buffer. At the start of the next time block, data is collected from the target device according to the update collection frequency to obtain a second raw data sequence, which is then stored in a backup buffer. In response to the state determination based on the first raw data sequence within the current time block and to determine the acquisition strategy for the next time block, the main buffer and the backup buffer are swapped.

[0101] In this embodiment, a dual-port static access memory structure can be used to introduce dual buffers, which satisfies the requirement that the method can collect dynamic information from the target device in real time without interruption. Then, at the moment of time block switching, the roles are switched, and the dual buffers can be switched seamlessly, so that the original primary buffer becomes the standby state and the original standby buffer becomes the primary state. In this way, at any given time, one buffer is in the writing state while the other is in the reading or processing state, which constitutes a device dynamic information collection method that performs collection from the target device in real time throughout the process.

[0102] In the implementation of this application embodiment, in order to ensure the continuity and integrity of data at the computing layer, in the current time block... During this period, the first original data sequence Write to the main buffer, next time block Initially, the newly collected data is the second raw data sequence. It was written to the original spare buffer.

[0103] In the current time block Internal completion status determination and generation of instruction control set Then, the buffers immediately switch roles. This role swap means that the original primary buffer becomes the current backup buffer and switches to read-only mode; similarly, the original backup buffer becomes the current primary buffer and switches to write mode.

[0104] Because it uses a dual-port memory, it responds to the generation of instruction control sets. The address mapping of the dual-port memory is switched, meaning the original primary buffer is switched to read-only mode for subsequent data compression, feature extraction, or uploading; the original backup buffer is switched to primary write mode to receive the second raw data sequence. .

[0105] This swapping operation is completed within a single time block, with no risk of data loss and no introduced processing delay. At the time block boundary, the write pointer instantly jumps to another buffer, while the read pointer can continue reading from the original buffer until processing is complete, achieving a seamless switch between the primary and backup buffers.

[0106] In one possible implementation, step S5 is also included.

[0107] Step S5: The cloud-based storage platform obtains the transfer matrix based on historical data and periodically sends it to the computing layer.

[0108] Based on historical data, a state prediction auxiliary mechanism is set up to obtain historical state transition probabilities, which can be used to predict in advance the operating range that the equipment may enter.

[0109] The state prediction assistance mechanism utilizes a cloud-field collaboration mechanism to periodically calculate a state transition matrix from historical data, which is the empirical probability of the device transitioning from state i to state j. Then, after each state determination, the calculation layer queries the transition matrix returned to it from the cloud based on the current state. If it finds that the probability of transitioning from a steady-state operation with no transient changes to other states exceeds the third threshold, it activates the preventive enhancement mode acquisition strategy in advance to avoid the loss of critical data due to response delays.

[0110] In implementing the embodiments of this application, in order to improve the adaptability and speed of steady-state discrimination and transient discrimination, a cloud-local collaborative mechanism is considered for optimization.

[0111] The cloud-based storage platform periodically extracts device state transition sequences from historical data uploaded by the computing layer; for example, this could be set to once a day for the target device in this embodiment.

[0112] Based on the steady-state and transient discrimination results of a large number of time blocks, i.e. and Statistical analysis is performed to construct a Markov transition model and obtain a state transition matrix P.

[0113] Then the elements of this matrix This represents the probability of a device transitioning from state i to state j. The state space includes four basic states: steady-state operation with no transient changes, non-steady-state operation with no transient changes, steady-state operation with transient changes, and non-steady-state operation with transient changes. Furthermore, it can be further subdivided into anomaly subclasses based on specific operating conditions.

[0114] The probability that the device transitions from state i to state j at this point At this point, state i is a steady-state operation with no transient changes, and state j is another state, which can be denoted as... , and then if Activate the enhanced mode acquisition strategy.

[0115] Furthermore, the state transition matrix P is sent to the computation layer, which injects it as prior knowledge into the dual-channel discrimination. In the steady-state discrimination of the first processing channel, a preset first trend offset threshold is used. It can be dynamically adjusted according to the transfer probability; in the mutation discrimination of the second processing channel, a preset fold factor is used. It can adaptively scale based on the probability of transitioning from the current state to a high-risk state, thereby improving sensitivity to early anomalies while ensuring a low false alarm rate.

[0116] Example 3 This is the third embodiment of the present invention. Based on embodiments 1 and 2, please refer to the following references. Figure 3 , Figure 3 This is a schematic diagram of the structure of a device dynamic information acquisition system based on the Industrial Internet of Things (IIoT) provided in an embodiment of the present invention. Further, the device dynamic information acquisition system based on the Industrial Internet of Things may specifically include a sequential electrical connection data acquisition unit, a feature discrimination unit, and a strategy generation unit. The data acquisition unit is configured to: perform data acquisition on the target device according to the acquisition frequency in effect of the current time block, and the computing layer obtains the first raw data sequence; The feature discrimination unit is configured to: in response to acquiring a first original data sequence, the first processing channel obtains a steady-state discrimination result of the target device operation based on the first original data sequence; The feature discrimination unit is further configured to: in response to acquiring the first original data sequence, the second processing channel obtains the transient discrimination result of the target device operation based on the first original data sequence; The strategy generation unit is configured to obtain the acquisition strategy for the next time block based on the steady-state discrimination result and the transient discrimination result.

[0117] In the implementation of this application embodiment, the system further includes a perception layer electrically connected to the data acquisition unit. The perception layer is configured with at least one node, and each node deploys a sensor channel. The sensor channel converts the collected raw data into a first raw data sequence and a second raw data sequence by an analog-to-digital converter. Both the raw data and the first raw data sequence are sent to the data acquisition unit. The policy generation unit is also electrically connected to the perception layer. The policy generation unit broadcasts the instruction control set to the node.

[0118] The first and second original data sequences are combined to form historical data, which is then uploaded to the cloud storage platform along with the original data. The cloud storage platform periodically distributes the calculated state transition matrix to the feature discrimination unit.

[0119] In this embodiment of the application, a CNC machine tool operating in a continuous production process is located in a smart factory. Real-time monitoring of its operating status is crucial for ensuring production safety and optimizing maintenance. Traditional fixed-frequency monitoring systems generate a large amount of redundant data during long-term stable operation of the equipment, while critical transient information may be missed due to insufficient sampling frequency when the equipment experiences a sudden failure. This system aims to resolve this contradiction.

[0120] Therefore, the CNC machine tool is the target device of this system. First, within the current time block of 1 minute, according to the currently effective acquisition frequency of 10Hz, the operating status of the machine tool is collected by multiple sensors deployed on the machine tool, including spindle vibration acceleration, cutting force, motor current, and temperature distribution. The first raw data sequence containing physical quantities of vibration, pressure, current, and temperature is obtained from the data acquisition unit.

[0121] In response to acquiring the first raw data sequence, the feature discrimination unit immediately initiates a parallel dual-channel state discrimination process.

[0122] The first processing channel operates in timer-driven mode to perform steady-state discrimination of the target device. At the end of the current time block, it performs steady-state characteristic analysis on the first raw data sequence.

[0123] First, the mean and standard deviation of each physical quantity, including vibration acceleration and bearing temperature, are calculated within the current time block to obtain the coefficient of variation. If the coefficients of variation of all key physical quantities are less than the preset steady-state threshold, and no continuous increase in bearing temperature or vibration amplitude has been detected in the past five consecutive time blocks (i.e., no significant trend shift), then the first processing channel determines that the target device is in a steady-state operating state. Otherwise, it is determined to be in a non-steady-state operating state, and this steady-state determination result is then output.

[0124] The second processing channel also starts in parallel and runs in interrupt service routine triggered mode to perform transient discrimination of the target device's operation. It monitors every acquired value in the first raw data sequence in real time, including when the vibration sensor data suddenly shows a spike, causing the difference between a certain acquired value and the mean of the physical quantity in the local neighborhood to exceed three times the preset historical standard deviation calculated based on historical steady-state data, the second processing channel immediately determines that there is a sudden change point in the physical quantity at this time.

[0125] To improve the accuracy of the judgment, a multidimensional physical quantity correlation test will also be performed. If, within a set very short time window, a sudden change is detected in both vibration and current within 100 milliseconds, with vibration suddenly increasing and current suddenly fluctuating, and their direction of change matches the entry for "early stage of bearing wear" in the preset transient event fault database, then the confidence level of this transient change will be increased.

[0126] Only when this confidence level is greater than the confidence level of the transient change corresponding to the entry obtained from historical data of steady-state operation can the transient change signal be finally triggered and the transient discrimination result be output.

[0127] When the steady-state discrimination result of the first processing channel and the transient discrimination result of the second processing channel are obtained, the strategy generation unit obtains the acquisition strategy for the next time block.

[0128] If the first processing channel determines that the device is operating in a steady state, and the second processing channel does not trigger any transient change signals throughout the entire time block, the system will activate the energy-saving mode acquisition strategy.

[0129] At this point, the acquisition frequency of the next time block is configured to a preset 5Hz as the base frequency. Meanwhile, intermittent sleep control is implemented for non-critical sensor channels, including temperature sensors, to significantly reduce the energy consumption and communication bandwidth usage of data acquisition. Compared with traditional fixed high-frequency acquisition, this avoids generating a large amount of redundant data under stable operating conditions and improves resource utilization efficiency.

[0130] If the first processing channel determines that the device is in an unsteady operating state, such as detecting a vibration trend shift, or if the second processing channel triggers a transient change signal, such as detecting a current change, the system will immediately activate the enhanced mode acquisition strategy.

[0131] At this point, the acquisition frequency of the next time block is increased to a preset high-density frequency, such as 20Hz and 50Hz, and a multi-sensor channel coordination mechanism is simultaneously enabled to ensure that all key sensors, including vibration, temperature, pressure, and current, acquire data synchronously at the highest frequency to capture the device status. Compared with traditional systems that may miss key information due to insufficient acquisition frequency during transient events, this method can rapidly increase the density of acquired data and ensure the complete capture of key transient features.

[0132] After obtaining the acquisition strategy for the next time block, the system generates a corresponding instruction control set and broadcasts it to all sensor nodes connected to the machine tool via a preset downlink. The sensors in these nodes update their acquisition frequencies according to the instruction control set. For example, if the strategy is energy-saving mode, the acquisition frequency is updated to 5Hz; if it is enhanced mode, it is updated to 20Hz or 50Hz depending on the specific situation.

[0133] After the data acquisition for the current time block is completed, the system configures the sensor's acquisition frequency to the updated frequency before moving to the next time block. When the next time block begins, the sensor performs data acquisition on the machine tool according to the new acquisition frequency, obtaining the second raw data sequence.

[0134] To ensure data continuity and processing efficiency, the system employs a dual-buffer mechanism. At the start of the current time block, the first raw data sequence, acquired according to the currently effective acquisition frequency, is stored in the primary buffer. At the start of the next time block, the second raw data sequence, acquired according to the updated acquisition frequency, is stored in the backup buffer. After completing state determination and identifying the acquisition strategy for the next time block within the current time block, the primary and backup buffers exchange roles, allowing new data to be processed seamlessly.

[0135] Furthermore, the first and second original data sequences are aligned on the timeline, together forming continuous historical data that facilitates monitoring by the cloud storage platform.

[0136] This closed-loop feedback mechanism ensures smooth switching of acquisition strategies and time alignment of data, solving the problem of lack of dynamic coupling in traditional systems.

[0137] In addition, the cloud-based storage platform continuously collects and analyzes historical data, and periodically calculates and updates the transition matrix of the target device's operating state. This matrix describes the empirical probability of the device transitioning from a steady-state operation to another state without transient changes, and sends it to the on-site computing layer.

[0138] The computation layer employs a state prediction assistance mechanism. After determining the state of the current time block, the computation layer combines the current device state with the transition matrix provided by the cloud. If it predicts that the device is transitioning to a high-risk state, potentially resulting in a non-steady-state or transient change exceeding a preset third threshold of 70%, the system will proactively activate a preventative enhanced mode acquisition strategy even if no actual transient change has been detected. This proactive adjustment avoids response delays caused by waiting for actual changes to occur, effectively preventing the loss of critical transient data and solving the problem of response lag in traditional systems.

[0139] Example 4 The fourth embodiment of the present invention differs from the previous embodiments in that: Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0140] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through some interfaces, devices, or units, or they may be electrical, mechanical, or other forms of connection.

[0141] The units described as separate components may or may not be physically separate. As will be apparent to those skilled in the art, the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0142] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0143] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or grid device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0144] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for collecting dynamic equipment information based on the Industrial Internet of Things, characterized in that, include: Sampling is performed on the target device according to the sampling frequency in effect for the current time block to obtain the first raw data sequence; In response to acquiring the first raw data sequence, the first processing channel obtains the steady-state discrimination result of the device operation based on the first raw data sequence; In response to acquiring the first raw data sequence, the second processing channel obtains the transient discrimination result of the device operation based on the first raw data sequence; Based on the steady-state and transient discrimination results, the acquisition strategy for the next time block is obtained; According to the acquisition strategy, the updated sampling frequency is obtained, and when entering the next time block, sampling is performed on the target device according to the updated sampling frequency.

2. The method for collecting dynamic equipment information based on the Industrial Internet of Things according to claim 1, characterized in that, According to the acquisition strategy, an update sampling frequency is obtained, and sampling of the target device is performed according to the update sampling frequency when entering the next time block, including: Based on the acquisition strategy of the next time block, obtain the instruction control set; According to the preset downlink, the instruction control set is broadcast to all nodes; The node is configured with one or more sensors that perform sampling on the target device; The sensor obtains the updated sampling frequency according to the instruction control set; After sampling the target device at the sampling frequency that is effective for the current time block, the sampling frequency is configured to the updated sampling frequency before entering the next time block. When entering the next time block, the target device is sampled according to the updated sampling frequency.

3. The method for collecting dynamic equipment information based on the Industrial Internet of Things according to claim 2, characterized in that, When entering the next time block, the target device is sampled according to the updated sampling frequency to obtain the second original data sequence; The first and second original data sequences are aligned on the time axis. The first original data sequence and the second original data sequence constitute historical data; At the start of the current time block, sampling is performed on the target device according to the sampling frequency in effect for the current time block to obtain the first raw data sequence, which is then stored in the main buffer. At the start of the next time block, the target device is sampled according to the update sampling frequency to obtain a second raw data sequence, which is then stored in a spare buffer. In response to the state determination based on the first raw data sequence within the current time block and to determine the acquisition strategy for the next time block, the main buffer and the backup buffer are swapped.

4. The method for collecting dynamic equipment information based on the Industrial Internet of Things according to claim 1, characterized in that, In response to acquiring the first raw data sequence, the first processing channel obtains a steady-state discrimination result of device operation based on the first raw data sequence, including: In response to acquiring the first raw data sequence, a parallel dual-channel state discrimination process is initiated; The dual channels are configured with a first processing channel; The first processing channel is configured to operate in timer-driven mode; Steady-state characteristic analysis is performed once every at least one time block to determine whether the target device is in steady-state operation; Based on whether it is in steady-state operation, the steady-state judgment result of the equipment operation is obtained.

5. The method for collecting dynamic equipment information based on the Industrial Internet of Things according to claim 4, characterized in that, In response to acquiring the first raw data sequence, the first processing channel obtains a steady-state discrimination result of device operation based on the first raw data sequence, including: Based on the first original data sequence, the mean and standard deviation of each physical quantity in the current time block are obtained to obtain the coefficient of variation. If the coefficients of variation of all physical quantities are less than the preset steady-state threshold, and no trend shift occurs within at least one consecutive time block, the equipment is determined to be in a steady-state operating state. Otherwise, the equipment is determined to be in an unsteady operating state; The steady-state and non-steady-state operating states are used as the steady-state discrimination results of the equipment operation.

6. The method for collecting dynamic equipment information based on the Industrial Internet of Things according to claim 1, characterized in that, In response to acquiring the first raw data sequence, the second processing channel obtains the transient discrimination result of device operation based on the first raw data sequence, including: In response to acquiring the first raw data sequence, a parallel dual-channel state discrimination process is initiated; The dual channels are configured with a second processing channel; The second processing channel is configured to run in interrupt service routine triggered mode; In response to the detection that any physical quantity in the first raw data sequence exceeds a local extremum, transient change event identification is immediately initiated to determine whether a transient change has occurred in the target device; Based on whether a transient change occurs, the transient discrimination result of the target equipment operation is obtained.

7. The method for collecting dynamic equipment information based on the Industrial Internet of Things according to claim 6, characterized in that, In response to acquiring the first raw data sequence, the second processing channel obtains the transient discrimination result of the target device operation based on the first raw data sequence, including: Based on the first original data sequence, obtain all sampled values ​​of each physical quantity in the current time block to obtain the mean value in the local neighborhood. Based on historical data from steady-state operation, obtain the historical standard deviation of each physical quantity; If the difference between any sampled value of any physical quantity and the mean of the corresponding local neighborhood is greater than a preset multiple of the historical standard deviation, it is determined that the physical quantity has a sudden change point at this time. If any physical quantity has at least one abrupt change point, a transient abrupt change signal is triggered; Based on whether a transient change signal is triggered, the transient discrimination result of the target device operation is obtained.

8. The method for collecting dynamic equipment information based on the Industrial Internet of Things according to claim 7, characterized in that, If any physical quantity has at least one abrupt change point, a transient abrupt change signal is triggered, which also includes a multidimensional physical quantity correlation test: If at least two physical quantities are detected to have a sudden change within a set time window, and the direction of change of the physical quantity that has a sudden change matches any entry in the preset fault library of transient events, then the confidence level of the transient change that has occurred at this time is increased. If the confidence level of a transient change occurs at this time is greater than the confidence level of a transient change for the corresponding entry obtained from historical data of steady-state operation, then a transient change signal is triggered.

9. The method for collecting dynamic equipment information based on the Industrial Internet of Things according to claim 1, characterized in that, The step of obtaining the acquisition strategy for the next time block based on the steady-state discrimination result and the transient discrimination result includes: The energy-saving mode acquisition strategy is activated only if the first processing channel determines that the target device is in steady-state operation and the second processing channel does not trigger any transient changes. The energy-saving mode acquisition strategy is configured to: enable a preset base frequency at the sampling frequency of the next time block, and implement intermittent sleep control for non-critical sensor channels; When any channel outputs a non-steady-state signal, the enhanced mode acquisition strategy is activated. The enhanced mode acquisition strategy is configured to enable a preset high-density frequency at the sampling frequency of the next time block, and simultaneously enable a multi-sensor channel collaboration mechanism.

10. A device dynamic information acquisition system based on the Industrial Internet of Things, characterized in that, It includes a data acquisition unit, a feature discrimination unit, and a strategy generation unit for sequential electrical connections; The data acquisition unit is configured to: perform sampling on the target device according to the sampling frequency in effect of the current time block, and acquire the first raw data sequence; The feature discrimination unit is configured to: in response to acquiring a first original data sequence, the first processing channel obtains a steady-state discrimination result of the target device operation based on the first original data sequence; The feature discrimination unit is further configured to: in response to acquiring the first original data sequence, the second processing channel obtains the transient discrimination result of the target device operation based on the first original data sequence; The strategy generation unit is configured to obtain the acquisition strategy for the next time block based on the steady-state discrimination result and the transient discrimination result.

Images