A method, system, device, and medium for detection enhancement of an industrial device

Abstract

The application relates to a detection enhancement method, system, device and medium of an industrial equipment. The method comprises the following steps: injecting the same global frame sequence number and high-precision collection time stamp into all sensor data in each sampling period to obtain an original synchronous data frame; based on the invariable constraint of the global frame sequence number and the high-precision collection time stamp, signal conditioning is performed on the original synchronous data frame to generate an atomic transmission unit; the reconstructed sensor data frame is stored in a synchronous frame queue in ascending order of the global frame sequence number; time alignment decision is performed according to the high-precision collection time stamp corresponding to the reconstructed sensor data frame at the head of the synchronous frame queue and a target presentation time, so that the current reconstructed sensor data frame to be rendered is obtained; based on the current reconstructed sensor data frame to be rendered, combined with the reconstructed sensor data frame that has been rendered in the last visual rendering cycle, time interpolation calculation is performed to generate a synchronous rendering data set. By adopting the method, cross-platform data synchronous visual presentation can be realized.

Description

Technical Field

[0001] This invention belongs to the field of industrial equipment visualization inspection technology, and in particular relates to an enhanced inspection method, system, equipment and medium for industrial equipment. Background Technology

[0002] With the development of digital technology in industrial production, real-time data monitoring technology for industrial equipment based on LabVIEW has emerged. Through various physical sensors, it can collect and display equipment operating parameters, providing core data support for operation and maintenance personnel.

[0003] In traditional technologies, industrial equipment inspection mainly relies on LabVIEW two-dimensional data monitoring. After sensors collect key parameters such as equipment displacement, vibration, and temperature, the LabVIEW system completes data processing and visualization. Maintenance personnel judge whether there is a fault in the equipment based on abstract numerical changes, waveform trends, and other information, and then carry out fault diagnosis and maintenance work.

[0004] However, the above methods suffer from limited visualization capabilities in two-dimensional monitoring, only able to display data in abstract chart form, failing to reconstruct the three-dimensional spatial structure of the equipment and the installation location of sensors. This results in insufficient intuitiveness in fault location, making it difficult to fully grasp the impact range of faults in large and complex equipment. Furthermore, the models have poor adaptability, making it difficult to directly import three-dimensional design models and lacking cross-scenario scalability. If a cross-platform visualization engine is used, the multi-sensor data at the same physical moment will arrive at the visualization terminal at different times and be rendered asynchronously due to uncertain delays, scheduling differences, transmission jitter, and other factors in each link, resulting in "time misalignment." This seriously affects the accuracy of fault diagnosis and may even mislead the diagnostic direction, failing to meet the actual needs of high-precision fault diagnosis for industrial equipment. Summary of the Invention

[0005] Based on this, it is necessary to provide a detection enhancement method, system, device, and medium for industrial equipment that can solve cross-domain timestamp distortion and realize visualization of multi-source heterogeneous sensors, addressing the aforementioned technical problems.

[0006] In a first aspect, this application provides a method for enhancing the detection of industrial equipment, comprising:

[0007] Based on a unified hardware sampling clock, data from multiple sensors of industrial equipment are synchronously acquired according to a preset sampling period. All sensor data within each sampling period are injected with the same global frame sequence number and the same high-precision acquisition timestamp to obtain the original synchronous data frame.

[0008] Based on the invariant constraints of the global frame sequence number and the high-precision acquisition timestamp, the sensor data of different physical quantities in the original synchronous data frame are subjected to parallel signal conditioning to obtain the reconstructed sensor data frame. When all sensor data belonging to the same global frame sequence number have completed signal conditioning, an atomic transmission unit is generated. The atomic transmission unit includes the global frame sequence number, the high-precision acquisition timestamp, and the reconstructed sensor data frame.

[0009] Based on the global frame sequence number of each atomic transmission unit, the reconstructed sensor data frames corresponding to the global frame sequence number are stored in the synchronization frame queue in ascending order of the global frame sequence number.

[0010] In each visualization rendering loop, the target rendering time of the current visualization rendering loop is calculated based on the current local time of the visualization client and the preset buffer time constant. Time alignment decision is made based on the high-precision acquisition timestamp and target rendering time corresponding to the reconstructed sensor data frame at the head of the synchronization frame queue to obtain the reconstructed sensor data frame to be rendered.

[0011] Based on the reconstructed sensor data frame to be rendered, and combined with the reconstructed sensor data frame rendered in the previous visualization rendering loop, time interpolation calculation is performed to generate a synchronous rendering dataset that is aligned with the target rendering time; the synchronous rendering dataset is used to drive the unified update of the 3D model and the visualization interface.

[0012] In one embodiment, based on a unified hardware sampling clock, data from multiple sensors of an industrial device are synchronously acquired according to a preset sampling period. The same global frame sequence number and the same high-precision acquisition timestamp are injected into all sensor data within each sampling period to obtain the original synchronized data frame, including:

[0013] A global sampling clock signal matching a preset sampling period is generated based on a unified hardware sampling clock. The global sampling clock signal is used to instruct each sensor to trigger an analog-to-digital conversion channel to start sampling once on the valid edge of each sampling clock signal, and to store the sampled value in the analog-to-digital conversion result register.

[0014] Based on the trigger time corresponding to the effective edge of the global sampling clock signal, a preset atomic read instruction is initiated; the atomic read instruction is used to instruct the latest sampled value of each channel to be read from the analog-to-digital conversion result register at once to obtain sensor data;

[0015] Based on a high-precision clock source synchronized with Coordinated Universal Time, the time scale value of the atomic read command is obtained and used as a high-precision acquisition timestamp;

[0016] Based on a monotonically increasing global counter, a unique increment value is assigned to the sensor data corresponding to the atomic read command, which serves as the global frame sequence number;

[0017] The global frame sequence number, high-precision acquisition timestamp, and sensor data are packaged together into a predefined memory data block to obtain the original synchronization data frame.

[0018] In one embodiment, based on the invariant constraints of the global frame sequence number and the high-precision acquisition timestamp, sensor data of different physical quantities in the original synchronized data frame are subjected to parallel signal conditioning to obtain a reconstructed sensor data frame. After all sensor data belonging to the same global frame sequence number have completed signal conditioning, an atomic transmission unit is generated, including:

[0019] Sensor data of different physical quantities in the original synchronization data frame are distributed to the corresponding signal processing channels for signal conditioning to obtain sensor reconstructed data output by each signal processing channel, and the sensor reconstructed data is stored in the frame reconstructed buffer according to the global frame sequence number.

[0020] The frame reassembly buffer is continuously monitored. Only when it is confirmed that all sensor reassembly data corresponding to the global frame sequence number has arrived in the frame reassembly buffer, all sensor reassembly data is encapsulated into a reconstructed sensor data frame, and combined with the global frame sequence number and high-precision acquisition timestamp, an atomic transmission unit is generated.

[0021] In one embodiment, in each visualization rendering loop, the target rendering time of the current visualization rendering loop is calculated based on the current local time of the visualization client and a preset buffer time constant. A time alignment decision is then made based on the high-precision acquisition timestamp corresponding to the reconstructed sensor data frame at the head of the synchronization frame queue and the target rendering time to obtain the currently rendered reconstructed sensor data frame, including:

[0022] The target rendering time for the current visualization rendering cycle is calculated based on the current local time of the visualization client and a preset buffer time constant; the formula for calculating the target rendering time is as follows: ,in, To visualize the client's current local time, This is the preset buffer time constant;

[0023] For the reconstructed sensor data frame at the head of the synchronization frame queue, the visible timestamp is calculated based on the corresponding high-precision acquisition timestamp and the current network latency estimate; the formula for calculating the visible timestamp is as follows: ,in, As a delay compensation factor, For a fixed processing pipeline delay compensation value, For high-precision timestamp collection, This is the current network latency estimate;

[0024] The visible timestamp is compared with the target rendering time of the current loop. If the visible timestamp is less than or equal to the target rendering time, the corresponding reconstructed sensor data frame is removed from the synchronization frame queue and determined as the current frame to be rendered, thus obtaining the current reconstructed sensor data frame to be rendered.

[0025] In one embodiment, for the reconstructed sensor data frame at the head of the synchronization frame queue, a visible timestamp is calculated based on the corresponding high-precision acquisition timestamp and the current network latency estimate, including:

[0026] The local time of arrival of the reconstructed sensor data frame at the visualization client is recorded, and combined with the data server's sending timestamp in the atomic transmission unit, the network round-trip delay time of the reconstructed sensor data frame is calculated; the formula for calculating the network round-trip delay time is as follows: ,in, To reach the local time of the visualization client, Send timestamps to the data server;

[0027] Based on the high-precision acquisition timestamp and the data server's transmission timestamp, the queuing delay of the reconstructed sensor data frame at the data server is calculated; the formula for calculating the queuing delay is as follows: ,in, For high-precision timestamp collection;

[0028] The sliding window averaging method is used to calculate the average value of the network round-trip delay and queuing delay of multiple reconstructed sensor data frames to obtain the current network delay estimate.

[0029] In one embodiment, based on the currently rendered reconstructed sensor data frame and combined with the reconstructed sensor data frames rendered in the previous visualization rendering loop, time interpolation is performed to generate a synchronized rendering dataset aligned with the target rendering time, including:

[0030] The visual timestamp corresponding to the reconstructed sensor data frame rendered in the previous visualization rendering loop is determined as the first visual timestamp, and the visual timestamp corresponding to the currently rendered reconstructed sensor data frame is determined as the second visual timestamp.

[0031] Based on the target rendering time of the current visualization rendering cycle, the normalized interpolation factor between the first and second visual timestamps is calculated; the formula for calculating the normalized interpolation factor is as follows: ,in, Presentation time for the target This is the first visible timestamp. This is the second visible timestamp;

[0032] For the reconstructed sensor data of all different physical quantities in the current reconstructed sensor data frame to be rendered and the reconstructed sensor data frame rendered in the previous visualization rendering loop, linear interpolation calculation is performed based on the normalized interpolation factor to generate a synchronous rendering dataset.

[0033] In one embodiment, the method further includes:

[0034] Based on the preset industrial equipment 3D model driving event processing function, the synchronous rendering dataset is transformed into a visual detection and positioning; the industrial equipment 3D model driving event processing function correspondingly sets the displacement interpolation data in the synchronous rendering dataset as the spatial offset of the rotor model, calculates and superimposes the high-frequency vibration effect based on the vibration interpolation data, and updates the color of the motor model through color mapping based on the temperature interpolation data.

[0035] Based on the preset user interface update event handling function, the physical quantity values ​​in the synchronous rendering dataset and the corresponding global frame sequence number are displayed in real time in the screen control.

[0036] Secondly, this application also provides an industrial equipment inspection enhancement system, comprising:

[0037] The real-time data monitoring module is used to synchronously collect data from multiple sensors of industrial equipment according to a preset sampling period based on a unified hardware sampling clock, and inject the same global frame sequence number and the same high-precision acquisition timestamp into all sensor data in each sampling period to obtain the original synchronous data frame.

[0038] The LabVIEW Data Processing module is used to perform parallel signal conditioning on sensor data of different physical quantities in the original synchronous data frame based on the invariant constraints of the global frame sequence number and the high-precision acquisition timestamp, so as to obtain the reconstructed sensor data frame. After all sensor data belonging to the same global frame sequence number has completed signal conditioning, an atomic transmission unit is generated. The atomic transmission unit includes the global frame sequence number, the high-precision acquisition timestamp, and the reconstructed sensor data frame.

[0039] The synchronization module is used to store the reconstructed sensor data frames corresponding to the global frame sequence numbers into the synchronization frame queue in ascending order of the global frame sequence numbers, based on the global frame sequence numbers of each atomic transmission unit.

[0040] The time distortion correction module is used to calculate the target rendering time of the current visualization rendering cycle based on the current local time of the visualization client and the preset buffer time constant in each visualization rendering cycle. It also makes a time alignment decision based on the high-precision acquisition timestamp and target rendering time corresponding to the reconstructed sensor data frame at the head of the synchronization frame queue to obtain the reconstructed sensor data frame to be rendered.

[0041] The UE5 visualization module is used to perform time interpolation calculations based on the reconstructed sensor data frame to be rendered, combined with the reconstructed sensor data frame rendered in the previous visualization rendering loop, to generate a synchronous rendering dataset that is aligned with the target rendering time; the synchronous rendering dataset is used to drive the unified update of the 3D model and the visualization interface.

[0042] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the detection enhancement method for any of the above-described industrial devices.

[0043] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the detection enhancement method for any of the above-described industrial devices.

[0044] The aforementioned industrial equipment detection enhancement method, system, equipment, and medium synchronously acquire multi-sensor data from industrial equipment at preset cycles based on a unified hardware sampling clock. All data within each sampling cycle are injected with the same global frame sequence number and high-precision acquisition timestamp to form original synchronized data frames. In the LabVIEW data processing terminal, multi-physical quantity data are subjected to parallel signal conditioning with the frame sequence number and timestamp as invariant constraints, generating reconstructed sensor data frames containing the same identifier. After conditioning all data under the same frame sequence number, these are encapsulated into atomic transmission units. The reconstructed data frames are stored in a synchronized frame queue according to the ascending order of the frame sequence number. In the UE5 visualization rendering loop, the target rendering time is calculated by combining the client's local time with a preset buffer constant. Alignment decisions are made by comparing the high-precision acquisition timestamp of the head data frame with the target time, selecting the current frame to be rendered. Time interpolation is performed between the current frame to be rendered and the previous rendered frame to generate a synchronized rendering dataset precisely aligned with the target rendering time, used to uniformly drive the 3D model and visualization interface. By injecting a time reference from the data source, maintaining frame integrity during transmission processing, and achieving alignment on the client side through time-series queues and interpolation, strict time consistency of multi-source heterogeneous sensor data is ensured in long-chain transmission and visualization rendering. This effectively overcomes the problem of cross-domain timestamp distortion, enabling the digital twin of device status monitoring to realistically and synchronously reflect the operating logic of the physical world, significantly improving the accuracy of fault diagnosis and the reliability of system monitoring. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a schematic flowchart of the detection enhancement method for industrial equipment according to the present invention;

[0047] Figure 2 This is a structural diagram of the detection enhancement system for industrial equipment according to the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0049] In one embodiment, such as Figure 1 As shown, an enhanced detection method for industrial equipment is provided. This embodiment illustrates the method using a LabVIEW terminal and an Unreal Engine 5 (UE5) terminal. It is understood that this method can also be applied to a server, or to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0050] S101. Based on a unified hardware sampling clock, data from multiple sensors of industrial equipment are synchronously acquired according to a preset sampling period. All sensor data within each sampling period are injected with the same global frame sequence number and the same high-precision acquisition timestamp to obtain the original synchronous data frame.

[0051] This illustration demonstrates how the cRIO real-time controller hardware works in conjunction with the LabVIEW Real-Time module to establish a unified and accurate time base and data acquisition logic. Specifically, the cRIO's FPGA timing loop or scan engine is configured, using an onboard high-stability crystal oscillator as the reference source for the global sampling clock signal. The frequency stability of this reference source directly determines the consistency of the sampling time. The preset sampling period must follow the Nyquist theorem, using the highest frequency sensor signal in industrial equipment as a reference, ensuring that the sampling frequency is not less than twice the highest frequency of that signal, thereby avoiding signal aliasing. For temperature sensors with slower response rates, oversampling is implemented through cRIO hardware channel configuration or driver layer parameter settings. Subsequently, downsampling is performed in the LabVIEW Real-Time module, aligning the sampling time points of the temperature sensor with those of high-speed sensors such as displacement and vibration sensors, thus physically eliminating the initial time deviation caused by independent sampling clocks between different types of sensors.

[0052] At the start of each preset sampling period, the LabVIEW Real-Time module triggers an atomic data read operation. This operation is performed through the FPGA buffer of cRIO or the direct I / O interface, ensuring that the latest data from all connected sensors is completely read within the same time window, including displacement 3D vectors, vibration acceleration 3D vectors, temperature scalars, and rotational speed parameters. The width of this time window is strictly consistent with the preset sampling period. After the data read is completed, LabVIEW's built-in high-precision timing function is immediately called to obtain the timestamp of the acquisition completion moment. This timestamp is synchronized with Coordinated Universal Time (UTC) via the PTP protocol, achieving microsecond-level accuracy, and is used as a high-precision acquisition timestamp for subsequent end-to-end time tracking. Simultaneously, the LabVIEW Real-Time module maintains a 64-bit monotonically increasing counter. After each atomic read of a set of sensor data, the current counter value is assigned to that set of data as a global frame sequence number. After assignment, the counter automatically increments by 1. The global frame sequence number serves as a unique logical identifier for the data frame, unaffected by physical clock transitions or resets, and is used solely to characterize the order in which the data was generated. By using LabVIEW's Cluster data structure, the global frame sequence number, high-precision acquisition timestamp, and raw data from all sensors are encapsulated into raw synchronization data frames, achieving a tight binding between the time stamp and the data itself.

[0053] S102. Based on the invariant constraints of the global frame sequence number and the high-precision acquisition timestamp, the sensor data of different physical quantities in the original synchronous data frame are subjected to parallel signal conditioning to obtain the reconstructed sensor data frame. When all sensor data belonging to the same global frame sequence number have completed signal conditioning, an atomic transmission unit is generated. The atomic transmission unit includes the global frame sequence number, the high-precision acquisition timestamp, and the reconstructed sensor data frame.

[0054] Furthermore, the LabVIEW host receives raw synchronous data frames transmitted via CRIO through high-speed buses such as Ethernet. A multi-channel signal conditioning pipeline is built based on LabVIEW's parallel processing architecture. Each signal conditioning channel corresponds to the processing logic of a specific physical quantity, and all channel inputs and outputs strictly retain the global frame sequence number and high-precision acquisition timestamp of the original synchronous data frame, ensuring that timing identifiers are not lost or tampered with during processing. For displacement and vibration sensor data, the signal conditioning channel calls LabVIEW's digital filtering function library to perform low-pass or band-pass filtering to remove environmental noise, while simultaneously loading preset calibration coefficients to correct sensor errors. The processing algorithm for this type of signal involves complex calculations such as spectrum analysis, resulting in some uncertainty in processing time. For temperature sensor data, the signal conditioning channel performs linearization processing to correct sensor nonlinearity errors and uses simple low-pass filtering to eliminate random interference. The processing flow is relatively simple and the processing time is stable. Although the signal conditioning time varies for different physical quantities, due to the invariant constraints of the global frame sequence number and high-precision acquisition timestamp, the processed sensor data can still be clearly associated with the same physical acquisition time.

[0055] Furthermore, downstream of the parallel signal conditioning channel, a LabVIEW state machine structure is deployed to implement the frame reassembler function. The frame reassembler internally maintains a cache dictionary with the global frame sequence number as the key. Each key in this dictionary corresponds to a data cache slot, used to store the conditioned data of different physical quantities under the same global frame sequence number. The frame reassembler monitors the output of each parallel conditioning channel in real time through LabVIEW's event listening mechanism. When all physical quantity data, such as displacement, vibration, temperature, and rotational speed, corresponding to a certain global frame sequence number have been conditioned and stored in the corresponding cache slot, the data frame for that global frame sequence number is determined to be "ready". For a ready data frame, the LabVIEW host first obtains the current system time as the transmission timestamp. Then, it integrates the global frame sequence number, high-precision acquisition timestamp, transmission timestamp, and the conditioned data of all physical quantities—that is, the reconstructed sensor data frame—into a unified data structure. The LabVIEW Flatten To JSON function is then called to convert this data structure into a JSON format string, generating atomic transmission units. As the smallest unit for cross-platform data transmission, the atomic transmission unit ensures the temporal consistency of its internal data through global frame sequence numbers and high-precision acquisition timestamps, thus guaranteeing the integrity of data frames during subsequent network transmission and visualization processing.

[0056] S103. Based on the global frame sequence number of each atomic transmission unit, store the reconstructed sensor data frames corresponding to the global frame sequence number into the synchronization frame queue in ascending order of the global frame sequence number.

[0057] Optionally, leveraging the characteristics of the TCP protocol and its caching and sorting mechanism, the global frame sequence number order of atomic transmission units can be maintained. On the LabVIEW host side, an independent TCP sending thread is built. This thread uses LabVIEW's While loop structure to strictly retrieve the JSON strings corresponding to the atomic transmission units from the sending queue in sequence. The execution logic of the sending thread is consistent with the incrementing order of the global frame sequence numbers, ensuring that the order in which atomic transmission units enter the network perfectly matches the order in which the data is generated. Because the TCP protocol itself has a byte stream order guarantee feature, as long as the sending end sends in order, the receiving end can receive the byte stream in the sending order, providing underlying support for the sequential transmission of data frames.

[0058] In the UE5 client, an asynchronous TCP receive thread is created using C++ or Blueprints. This thread continuously listens to the network port and reads the TCP byte stream. After converting the read byte stream into a string, it calls UE5's JSON parsing interface to attempt to parse the complete JSON object. During parsing, the global frame sequence number is extracted first from the JSON object, and the complete parsing result is stored in the UE5's receive buffer. This receive buffer uses a priority queue or doubly linked list data structure sorted by the global frame sequence number. For example, even if the TCP protocol guarantees the order of the byte stream, if the JSON data volume of a single atomic transmission unit is too large, causing TCP fragmentation, or if the UE5 receive thread's system scheduling causes the parsing order of multiple JSON objects to be inconsistent with the receiving order, the receive buffer can perform a secondary sorting using the global frame sequence number to ensure that the output data strictly follows the ascending order of the global frame sequence number. The receive buffer then injects the sorted atomic transmission units sequentially into the UE5's Synchronized Frame Queue (SFQ). The Synchronized Frame Queue uses a first-in, first-out storage logic to provide ordered data input for subsequent visualization rendering time alignment.

[0059] S104. In each visualization rendering loop, the target rendering time of the current visualization rendering loop is calculated based on the current local time of the visualization client and the preset buffer time constant. Time alignment decision is made based on the high-precision acquisition timestamp and target rendering time corresponding to the reconstructed sensor data frame at the head of the synchronization frame queue to obtain the reconstructed sensor data frame to be rendered.

[0060] This example illustrates how precise alignment of data frames with the visualization rendering loop is achieved in the UE5 client's Game Thread through state management and time correction algorithms of the synchronized frame queue. UE5's synchronized frame queue comprises a buffer queue and a synchronized state machine. The buffer queue uses UE5's TQueue data structure to store atomic transmission units sorted by the receive buffer. The synchronized state machine is responsible for executing the time alignment decision logic, and its triggering cycle is consistent with UE5's visualization rendering loop, triggering once per frame tick. For example, the target rendering time is... ,in, This is the current local time for the UE5 game thread. This time is provided by the UE5 system clock and synchronized with the LabVIEW host time to millisecond precision via the NTP protocol. The preset buffer time constant must be chosen in a trade-off between real-time performance and synchronization stability. A certain time buffer is reserved to offset the impact of network transmission jitter and client processing latency. Furthermore, before making time alignment decisions, the one-way network latency is estimated to optimize time correction accuracy. This is done by assuming that the network round-trip latency is approximately symmetrical, and the one-way latency is taken as half of the round-trip latency. ,in, The current local time of the UE5 client, and Based on the same system clock; This is the high-precision acquisition timestamp of the data frame at the head of the synchronization frame queue. Based on the estimated one-way network latency, the corrected visible timestamp of the data frame is calculated, i.e. ,in, This is an adjustment factor with a value range between 0.5 and 1.5. The default setting is 1.0. It is used to adjust the magnitude of time prediction or lag and can be dynamically optimized according to the actual network environment. It is a fixed time offset, obtained through offline calibration, used to compensate for the delay of fixed processing links such as signal conditioning on the LabVIEW side and JSON parsing on the UE5 side.

[0061] Specifically, the synchronization state machine reads the corrected visible timestamp of the head data frame in the synchronization frame queue. ,like If the data frame meets the rendering time requirements and is ready to be dequeued, a continuity check is performed to verify whether the global frame sequence number of the head data frame is the next expected sequence number. If the sequence numbers are consecutive, the data frame is directly identified as the current reconstructed sensor data frame to be rendered. If the sequence numbers are not consecutive, a timeout is started. If the timeout exceeds two preset sampling periods, the data frame is still identified as the current reconstructed sensor data frame to be rendered, and the packet loss event is recorded.

[0062] S105. Based on the reconstructed sensor data frame to be rendered, and combined with the reconstructed sensor data frame rendered in the previous visualization rendering loop, perform time interpolation calculation to generate a synchronous rendering dataset aligned with the target rendering time; the synchronous rendering dataset is used to drive the unified update of the 3D model and visualization interface.

[0063] Furthermore, it executes synchronously with the visualization rendering loop in the UE5 game thread, using interpolation calculations to eliminate screen stuttering caused by a mismatch between the data frame update frequency and the rendering frame rate, while ensuring the temporal consistency of all data. Specifically, it obtains the corrected visual timestamp corresponding to the currently rendered reconstructed sensor data frame. And the corrected visual timestamps of the data frames rendered in the previous visualization rendering loop. Based on target presentation time Calculate interpolation factor ,in, It is a linear interpolation factor, whose value range is restricted to [0,1] through clamping operations. and When they overlap, This indicates that the data rendered in the previous frame is used entirely; when and When they overlap, This indicates that the current data to be rendered is used entirely; when When it is in between, The weights of the data to be rendered are reflected proportionally.

[0064] Optionally, the appropriate interpolation method is selected based on the type of sensor data. For physical quantities requiring smooth transitions, such as three-dimensional displacement vectors, three-dimensional vibration acceleration vectors, and rotational speed, the linear interpolation (Lerp) algorithm is used, weighting the data from the previous frame with the data to be rendered using an interpolation factor. For parameters involving rotational motion, such as rotor rotation angle, the spherical linear interpolation (Slerp) algorithm is used to ensure the smoothness and continuity of the rotational transition. For physical quantities that do not require smooth transitions, such as temperature, the current data to be rendered can be used directly, or a transition process can be performed using a easing function to avoid abrupt changes in temperature display. Through the above interpolation calculations, a synchronized rendering dataset precisely aligned with the target rendering time is generated. This dataset contains the time-consistency processing results of all sensors, ensuring that the physical times corresponding to parameters such as displacement, vibration, temperature, and rotational speed are completely consistent.

[0065] The synchronized rendering dataset is broadcast via UE5's event dispatcher (Delegate), and the update logic of both the 3D model and the visualization interface is bound to this event for unified triggering. Regarding 3D model updates, after receiving the synchronized rendering dataset, UE5's Actor Blueprint directly applies the displacement interpolation results to the device's 3D model using the SetActorRelativeLocation function, such as for position offset control of rotors and motor housings. Vibration interpolation results, combined with phase information derived from high-precision acquisition timestamps, are input to a sine wave superposition function to calculate a micro-amplitude vibration offset synchronized with the physical vibration signal, which is then superimposed onto the model's position or rotation parameters to restore the actual vibration state of the device. Temperature interpolation results are converted into color gradient values ​​through linear mapping, such as blue corresponding to low temperature and red to high temperature, updating the model's dynamic material parameters in real time to keep model color changes synchronized with temperature changes. Rotational speed interpolation results are used to calculate the rotational angle increment between the current rendering frame and the previous rendering frame, and applied to the rotor model using the SetActorRelativeRotation function to achieve a smooth presentation of rotational motion. Regarding the update of the visualization interface, after the UMG user interface control in UE5 binds an event, it will synchronously render the global frame sequence number, high-precision acquisition timestamp, and interpolation results of each physical quantity in the dataset, and update the corresponding text display control (Text Block) in real time to ensure that the digital readings on the interface are strictly synchronized with the motion and state changes of the 3D model.

[0066] In the aforementioned industrial equipment detection enhancement method, multi-sensor data of the industrial equipment are synchronously collected at a preset cycle using a unified hardware sampling clock. Simultaneously, the same global frame sequence number and the same high-precision acquisition timestamp are injected into all sensor data within each sampling cycle to form an original synchronous data frame. Based on the invariant constraints of the global frame sequence number and the high-precision acquisition timestamp, signal conditioning is performed in parallel on sensor data of different physical quantities within the original synchronous data frame. After all sensor data corresponding to the same global frame sequence number has been conditioned, an atomic transmission unit is generated containing the global frame sequence number, the high-precision acquisition timestamp, and the reconstructed sensor data frame. Subsequently, the reconstructed sensor data frames are stored in the synchronous frame queue in ascending order of the global frame sequence number. This process is repeated in each visualization rendering loop. In this process, the target rendering time is calculated by combining the current local time of the visualization client with the preset buffer time constant. Time alignment decision is made based on the high-precision acquisition timestamp of the data frame at the head of the synchronous frame queue and the target rendering time to determine the current data frame to be rendered. Finally, time interpolation calculation is performed by combining the data frames rendered in the previous rendering cycle to generate a synchronous rendering dataset aligned with the target rendering time and drive the unified update of the 3D model and visualization interface. This effectively solves the problem of inconsistent timing in cross-platform multi-source sensor data transmission and visualization process, ensures the inherent physical time logic relationship of the data, realizes high-fidelity visualization of the operating status of industrial equipment, provides a reliable data foundation for fault diagnosis, and improves the smoothness and real-time performance of system visualization.

[0067] In one embodiment, based on a unified hardware sampling clock, data from multiple sensors of an industrial device are synchronously acquired according to a preset sampling period. The same global frame sequence number and the same high-precision acquisition timestamp are injected into all sensor data within each sampling period to obtain the original synchronized data frame, including:

[0068] S11. Generate a global sampling clock signal that matches the preset sampling period based on a unified hardware sampling clock; the global sampling clock signal is used to instruct each sensor to trigger the analog-to-digital conversion channel to start sampling once on the effective edge of each sampling clock signal, and store the sampled value in the analog-to-digital conversion result register.

[0069] This diagram illustrates how the cRIO FPGA hardware logic collaborates with the LabVIEW Real-Time module to establish a synchronous triggering mechanism between the global sampling clock signal and the sensor's analog-to-digital converter (ADC) channels. The unified hardware sampling clock is generated by a high-stability crystal oscillator on the cRIO board. After frequency division and delay calibration by the FPGA's internal logic, it outputs a global sampling clock signal that strictly matches the preset sampling period. The effective edge of this signal, either the rising or falling edge, serves as the sole trigger reference for all sensor sampling actions. Each sensor's corresponding ADC channel establishes a binding relationship with the global sampling clock signal through the cRIO hardware configuration interface. When the effective edge arrives, the ADC channel immediately initiates the analog-to-digital conversion process, converting the analog signals acquired by the sensor, such as electrical signals corresponding to displacement, vibration, and temperature, into digital signals. The converted sampled value is then written to a dedicated analog-to-digital conversion result register in real time. The analog-to-digital conversion result register employs an independent address mapping design, with each sensor channel corresponding to a unique register address. This ensures that no data conflicts or overwriting occur during sampled value storage, providing hardware-level data isolation and integrity guarantees for subsequent atomic readings.

[0070] S12. Based on the trigger time corresponding to the effective edge of the global sampling clock signal, start the preset atomic read instruction; the atomic read instruction is used to instruct the latest sampled value of each channel to be read from the analog-to-digital conversion result register at one time to obtain sensor data.

[0071] Furthermore, after the global sampling clock signal triggers the ADC channel to complete sampling and store the sampled value along its effective edge, the trigger moment corresponding to this effective edge will be synchronously fed back to the LabVIEW Real-Time module, triggering the execution of a preset atomic read instruction. The atomic read instruction is pre-compiled by the LabVIEW FPGA module and embedded in the cRIO FPGA logic. During instruction execution, a bus burst transmission mode is used, traversing the analog-to-digital conversion result register addresses corresponding to all sensor channels at once, and reading the latest sampled value stored in each register in a preset order, such as displacement channel, vibration channel, and temperature channel. The entire read process is completed within a single FPGA clock cycle, avoiding time differences caused by reading channel by channel. After the read is completed, the sampled values ​​of all channels are integrated into a unified sensor data set, ensuring that all data in this set corresponds to the sampling result at the same trigger moment.

[0072] S13. Based on a high-precision clock source synchronized with Coordinated Universal Time, obtain the time scale value of the atomic read command, and use the time scale value as a high-precision acquisition timestamp.

[0073] Specifically, the high-precision clock source synchronizes with a network-wide time server via the cRIO PTP protocol interface, calibrating the local hardware clock in real time to ensure that the clock source's time accuracy remains consistent with UTC at the microsecond level. Once the atomic read command is executed, the LabVIEW Real-Time module immediately triggers a time scale value acquisition operation, reading the current time from the high-precision clock source through a dedicated clock interface. This time is the time scale value indicating the completion of the atomic read command. The time scale value is stored in a 64-bit floating-point data format, accurately recording the time information of the read completion moment. This information is directly used as a high-precision acquisition timestamp and bound to the sensor data, ensuring that each sensor data set can be traced back to a unique physical time point.

[0074] S14. Based on a monotonically increasing global counter, assign a unique increment value to the sensor data corresponding to the atomic read command, which serves as the global frame sequence number.

[0075] For example, a global counter maintained by the LabVIEW Real-Time module is used to uniquely assign global frame sequence numbers. The global counter uses a 64-bit unsigned integer type, is initially set to 0, and is configured to retain the sequence number even after a power failure to prevent duplicate sequence numbers due to system restarts. Each time an atomic read instruction completes and generates a sensor data set, the global counter automatically increments, using the incremented value as the global frame sequence number for the current sensor data set. This allocation process is atomic, ensuring that each sensor data set obtains a unique sequence number, and the sequence number strictly follows the order in which the data was generated, incrementing without being affected by system clock jumps, task scheduling delays, or other factors.

[0076] S15. Pack the global frame sequence number, high-precision acquisition timestamp, and sensor data together into a predefined memory data block to obtain the original synchronization data frame.

[0077] The raw synchronization data frame is generated using LabVIEW's structured data encapsulation feature. The predefined memory data block is designed based on LabVIEW's cluster data type, and its structure is predefined during system initialization using the LabVIEW programming environment, ensuring data storage standardization and compatibility. During the encapsulation process, the LabVIEW Real-Time module sequentially writes the global frame sequence number, high-precision acquisition timestamp, and sensor data into the corresponding fields of the memory data block in a preset order. The sensor data is stored in a structured manner according to channel type. The resulting memory data block is the raw synchronization data frame. This data frame achieves fast access through memory address mapping, ensuring efficient extraction of timing identifiers and sensor data during subsequent signal conditioning and transmission, maintaining the binding relationship between the three without disruption.

[0078] In one embodiment, based on the invariant constraints of the global frame sequence number and the high-precision acquisition timestamp, sensor data of different physical quantities in the original synchronized data frame are subjected to parallel signal conditioning to obtain a reconstructed sensor data frame. After all sensor data belonging to the same global frame sequence number have completed signal conditioning, an atomic transmission unit is generated, including:

[0079] S21. Distribute the sensor data of different physical quantities in the original synchronization data frame to the corresponding signal processing channels for signal conditioning, obtain the sensor reconstructed data output by each signal processing channel, and store the sensor reconstructed data into the frame reconstructed buffer according to the global frame sequence number.

[0080] In a schematic representation, after receiving the raw synchronization data frame, the LabVIEW host uses its built-in data routing logic to precisely distribute different types of sensor data to preset dedicated signal processing channels based on the physical quantity attributes of the sensor data. Each signal processing channel is configured with dedicated conditioning logic for the characteristics of the corresponding physical quantity. For example, bandpass filtering and phase correction algorithms are used for high-frequency vibration data, and nonlinear error correction algorithms are used for temperature data, ensuring that the conditioned data accurately reflects the device's operating status. During signal conditioning, all processing channels strictly retain the global frame sequence number and high-precision acquisition timestamp of the original synchronization data frame, ensuring that the sensor reconstructed data generated after conditioning remains uniquely associated with the original acquisition time. The frame reconstructing buffer adopts LabVIEW's key-value dictionary data structure design, using the global frame sequence number as the unique index key. Each index key corresponds to an independent buffer slot. After conditioning, the sensor reconstructed data is directly written to the buffer slot corresponding to the global frame sequence number, achieving centralized storage and index management of different physical quantity data at the same acquisition time.

[0081] S22. Continuously monitor the frame reassembly buffer. Only when it is confirmed that all sensor reassembly data corresponding to the global frame sequence number has arrived at the frame reassembly buffer, encapsulate all sensor reassembly data into a reconstructed sensor data frame, and generate an atomic transmission unit by combining the global frame sequence number and the high-precision acquisition timestamp.

[0082] Furthermore, the frame reassembly buffer is monitored in real time using LabVIEW's event-driven listening mechanism or state machine module. The core monitoring logic is to verify the integrity of the cache slot data corresponding to each global frame sequence number. The monitoring module traverses all index keys of the frame reassembly buffer at a preset period, and for each global frame sequence number, it verifies whether the sensor reassembly data of all physical quantities such as displacement, vibration, temperature, and rotational speed corresponding to it has been stored in the cache slot. When it is detected that all physical quantity data of a certain global frame sequence number is complete, the encapsulation process is immediately triggered. Specifically, LabVIEW's structured data integration function is called to integrate all sensor reassembly data under that global frame sequence number into a unified reconstructed sensor data frame according to a preset format. The reconstructed sensor data frame is then bound and encapsulated with the corresponding global frame sequence number and high-precision acquisition timestamp to form an atomic transmission unit. This atomic transmission unit achieves tight coupling between the timing identifier and the data body through a fixed data structure, ensuring that the timing correlation of the data is not destroyed during the transmission between LabVIEW and UE5 platforms.

[0083] In one embodiment, in each visualization rendering loop, the target rendering time of the current visualization rendering loop is calculated based on the current local time of the visualization client and a preset buffer time constant. A time alignment decision is then made based on the high-precision acquisition timestamp corresponding to the reconstructed sensor data frame at the head of the synchronization frame queue and the target rendering time to obtain the currently rendered reconstructed sensor data frame, including:

[0084] S31. Based on the current local time of the visualization client and a preset buffer time constant, calculate the target rendering time for the current visualization rendering cycle; the formula for calculating the target rendering time is as follows: ,in, To visualize the client's current local time, This is the preset buffer time constant.

[0085] This is illustrative, triggered every frame tick in the UE5 game thread to determine the ideal time to render data in the current rendering loop. Provided in real time by the UE5 system clock, this time is synchronized with the LabVIEW host time to millisecond-level accuracy via the NTP protocol, ensuring consistency of the time base across platforms. As a preset buffer time constant, its value needs to comprehensively consider the real-time requirements of industrial equipment monitoring and the stability of data transmission and processing. By reserving a fixed time buffer, the impact of unpredictable factors such as network transmission jitter and UE5 client data parsing delay is offset. The target presentation time of the current visualization rendering cycle is obtained by subtracting the buffer time constant from the current local time. This time point not only ensures the timeliness of data presentation but also provides a clear reference benchmark for subsequent time alignment decisions.

[0086] S32. For the reconstructed sensor data frame at the head of the synchronization frame queue, calculate the visible timestamp based on the corresponding high-precision acquisition timestamp and the current network latency estimate; the formula for calculating the visible timestamp is as follows: ,in, As a delay compensation factor, For a fixed processing pipeline delay compensation value, For high-precision timestamp collection, This is the current network latency estimate.

[0087] Furthermore, computational logic is executed on the reconstructed sensor data frame at the head of the synchronization frame queue, wherein, The high-precision acquisition timestamp carried by this data frame is directly associated with the physical acquisition time of the data, and has microsecond-level accuracy and UTC time synchronization characteristics. This is the current network latency estimate, based on the UE5 client's current local time and... The difference derivation reflects the time it takes for data to be transmitted from the acquisition end to the UE5 client. As a delay compensation factor, it can be dynamically adjusted according to the actual network environment to achieve flexible adaptation of the delay compensation range. This is a fixed pipeline latency compensation value, obtained through offline calibration, covering the cumulative latency of fixed processing links such as signal conditioning on the LabVIEW side, data encapsulation, and JSON parsing on the UE5 side. Substituting these parameters into the formula yields the visible timestamp. This timestamp accurately represents the ideal moment when a data frame should be presented on the UE5 device, providing a core basis for time alignment decisions.

[0088] S33. Compare the visible timestamp with the target rendering time of the current loop. If the visible timestamp is less than or equal to the target rendering time, remove the corresponding reconstructed sensor data frame from the synchronization frame queue and determine it as the current frame to be rendered, thus obtaining the current reconstructed sensor data frame to be rendered.

[0089] Optionally, a time alignment decision is performed via the UE5 synchronization state machine to verify whether the data frame meets the timing requirements of the current rendering cycle. The synchronization state machine extracts the visible timestamp of the data frame at the head of the synchronization frame queue in real time. And compared with the target rendering time calculated in the current rendering loop. Perform numerical comparisons. When a determination is made... When this occurs, it indicates that the ideal rendering time for the data frame has been reached or matches the current rendering rhythm. At this point, the synchronization state machine triggers a queue removal operation, stripping the reconstructed sensor data frame from the synchronization frame queue. Simultaneously, a state marking mechanism clearly designates it as the current frame to be rendered, ensuring that subsequent interpolation calculations and rendering operations are performed based on this time-aligned valid data frame, logically guaranteeing the temporal consistency between the visualization presentation and the physical acquisition time. When the determination... At that time, according to the synchronization frame queue order, check whether the visible timestamp of the next reconstructed sensor data frame meets the conditions.

[0090] In one embodiment, for the reconstructed sensor data frame at the head of the synchronization frame queue, a visible timestamp is calculated based on the corresponding high-precision acquisition timestamp and the current network latency estimate, including:

[0091] S41. Record the local time when the reconstructed sensor data frame arrives at the visualization client, and combine it with the data server's sending timestamp in the atomic transmission unit to calculate the network round-trip delay time of the reconstructed sensor data frame; the formula for calculating the network round-trip delay time is as follows: ,in, To reach the local time of the visualization client, Send timestamps to the data server.

[0092] For example, The data is recorded in real time by the UE5 system clock after the sensor data frame is fully received and the JSON parsing is completed. This moment is consistent with the UE5 local time base and has millisecond-level accuracy. This is a transmission timestamp embedded in the LabVIEW client when generating the atomic transport unit. This timestamp is recorded by the LabVIEW host system clock and is synchronized with the high-precision acquisition timestamp. Based on the same time synchronization reference, the network round-trip time is calculated by the difference between the two. It comprehensively reflects the round-trip time of the data frame from being sent from the LabVIEW end, transmitted through the TCP network to the UE5 end for reception and parsing, including the time consumption caused by non-deterministic factors such as network queuing delay, routing jitter, and packet fragmentation and reassembly.

[0093] S42. Based on the high-precision acquisition timestamp and the data server's transmission timestamp, the queuing delay of the reconstructed sensor data frame at the data server is calculated; the formula for calculating the queuing delay is as follows: ,in, This is for high-precision data collection timestamps.

[0094] For example, This provides a high-precision timestamp for the atomic acquisition of data frames at the cRIO end. The difference between the timestamp of the data frame sent by the LabVIEW side and the timestamp of the data frame sent by the LabVIEW side. It covers fixed and semi-fixed delays such as signal conditioning and queuing for transmission of data frames in LabVIEW, and transmission queue scheduling. These delays are unrelated to network transmission and only reflect the time consumption characteristics of data flow within the server, providing data support for the subsequent separation of pure network delays.

[0095] S43. Using the sliding window averaging method, the network round-trip delay and queuing delay of multiple reconstructed sensor data frames are statistically analyzed and the average value is calculated to obtain the current network delay estimate.

[0096] Specifically, the stability of the delay estimation is optimized by using a statistical smoothing algorithm, which involves setting a fixed-length sliding window and continuously storing the data frames corresponding to the latest received reconstructed sensor data frames within the window. and Whenever a new data frame arrives and its corresponding two latency values ​​are calculated, it is added to the window, while the oldest expired latency data in the window is removed to ensure that the window always contains the latest statistical samples. The arithmetic mean of the network round-trip latency and queuing latency of all samples in the window is calculated. This averaging process smooths out the latency deviation caused by single network jitter and temporary server scheduling fluctuations. The final estimated current network latency value can more accurately reflect the stable time of actual network transmission, providing a reliable basis for the accurate calculation of subsequent visible timestamps.

[0097] In one embodiment, based on the currently rendered reconstructed sensor data frame and combined with the reconstructed sensor data frames rendered in the previous visualization rendering loop, time interpolation is performed to generate a synchronized rendering dataset aligned with the target rendering time, including:

[0098] S51. Determine the visual timestamp corresponding to the reconstructed sensor data frame rendered in the previous visualization rendering loop as the first visual timestamp, and determine the visual timestamp corresponding to the currently rendered reconstructed sensor data frame as the second visual timestamp.

[0099] Indicative, first visible timestamp This is the visual timestamp corresponding to the reconstructed sensor data frame that was rendered in the previous visualization rendering loop. This timestamp has been calculated using the previous time correction logic and is stored in UE5's frame state cache to represent the ideal presentation time corresponding to the previous rendered data. Second visual timestamp The visible timestamp, obtained after time correction of the reconstructed sensor data frame to be rendered, is directly extracted from the timing identifier field of that data frame. Together, they constitute the time interval for interpolation calculation, ensuring that subsequent interpolation operations are strictly based on physical timing logic and conform to the temporal evolution of device operating data.

[0100] S52. Based on the target rendering time of the current visualization rendering cycle, calculate the normalized interpolation factor between the first and second visual timestamps; the formula for calculating the normalized interpolation factor is as follows: ,in, Presentation time for the target This is the first visible timestamp. This is the second visible timestamp.

[0101] Furthermore, a unified interpolation factor is obtained through normalization. The target rendering time for the current visualization rendering cycle is calculated by UE5 using local time and buffer time constants, representing the ideal moment when the data should be rendered. In the formula, the numerator represents the time difference between the target rendering time and the ideal rendering time of the previous rendered data, while the denominator represents the time interval between the ideal rendering times of two consecutive data frames to be interpolated. The ratio of these two values ​​is the normalized interpolation factor. Optionally, the value range of this factor is limited to [0,1] through clamping, which intuitively reflects the relative position of the target presentation time within the time interval formed by the two visible timestamps, providing a unified weight benchmark for interpolation of different physical quantity data.

[0102] S53. For the reconstructed sensor data of all different physical quantities in the current reconstructed sensor data frame to be rendered and the reconstructed sensor data frame rendered in the previous visualization rendering cycle, perform linear interpolation calculation based on the normalized interpolation factor to generate a synchronous rendering dataset.

[0103] Optionally, for physical quantities such as displacement 3D vectors, vibration acceleration 3D vectors, and rotational speeds that require smooth transition between the two data frames, linear interpolation operations are performed separately, and the data from the previous frame and the data to be rendered are compared using a normalized interpolation factor. Perform a weighted summation, meaning the current physical quantity interpolation result is equal to the previous frame's data multiplied by (1- Multiply by the current data to be rendered The accumulated value. For physical quantities such as temperature, the current data to be rendered can be used directly or the same linear interpolation logic can be executed according to the actual presentation requirements to avoid data mutation. After interpolation calculation, all physical quantity data are integrated into a unified synchronous rendering dataset. This dataset is precisely aligned with the target rendering time to ensure that when UE5 drives the 3D model motion and the visualization interface is updated, all data corresponds to the same physical time sequence node.

[0104] In one embodiment, the method further includes:

[0105] S61. Based on the preset industrial equipment 3D model driving event processing function, the synchronous rendering dataset is transformed into a visual detection and positioning; the industrial equipment 3D model driving event processing function sets the displacement interpolation data in the synchronous rendering dataset as the spatial offset of the rotor model, calculates and superimposes the high-frequency vibration effect based on the vibration interpolation data, and updates the color of the motor model through color mapping based on the temperature interpolation data.

[0106] In a schematic representation, within the UE5 game thread, pre-defined blueprints or C++ event handling functions are used to transform quantized data from the synchronously rendered dataset into a visually intuitive 3D model. A pre-defined industrial equipment 3D model-driven event handling function is bound to the broadcast event of the synchronously rendered dataset; upon receiving the dataset, the model-driven logic is immediately triggered. For displacement interpolation data, the function calls UE5's `SetActorRelativeLocation` interface to directly map the data to the spatial coordinate offset of the industrial equipment rotor model, ensuring real-time matching between the model's position and the actual displacement state of the equipment. For vibration interpolation data, the function extracts the amplitude and frequency parameters of the data, combines them with time-related phase information, and calculates the micro-amplitude vibration offset using a sine wave superposition algorithm. This offset is then superimposed onto the rotor model's position or rotation parameters to recreate the high-frequency vibration characteristics of the equipment during operation. For temperature interpolation data, the function uses a linear mapping algorithm to map temperature values ​​to a pre-defined color gradient range and updates the surface color of models such as the motor housing in real-time through a dynamic material instance interface. This allows the model's color changes to intuitively reflect the equipment's temperature distribution and trends, achieving visualized detection and positioning of the equipment's operating status.

[0107] S62. Based on the preset user interface update event handling function, the physical quantity values ​​in the synchronous rendering dataset and the corresponding global frame sequence number are displayed in the screen control in real time.

[0108] Furthermore, the preset user interface update event handler extracts the interpolation results of various physical quantities such as displacement, vibration, temperature, and rotational speed from the synchronously rendered dataset, along with the corresponding global frame sequence numbers. Through UE5's control binding mechanism, this data is written in real-time to preset display controls such as TextBlocks on the interface. The physical quantity values ​​are displayed in a preset precision format, and the global frame sequence number identifies the acquisition frame corresponding to the currently displayed data, facilitating data tracing by maintenance personnel. Because the interface update and the 3D model drive use the same data source and are triggered synchronously, the values ​​displayed on the screen controls strictly correspond to the spatial offset, vibration effects, and color states of the 3D model.

[0109] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0110] Based on the same inventive concept, this application also provides an industrial equipment detection enhancement system for implementing the aforementioned industrial equipment detection enhancement method. The solution provided by this system is similar to the implementation described in the above method; therefore, the specific limitations of one or more industrial equipment detection enhancement system embodiments provided below can be found in the limitations of the industrial equipment detection enhancement method described above, and will not be repeated here.

[0111] In one exemplary embodiment, such as Figure 2 As shown, an industrial equipment inspection enhancement system is provided, comprising:

[0112] The real-time data monitoring module 201 is used to synchronously collect data from multiple sensors of industrial equipment according to a preset sampling period based on a unified hardware sampling clock, and inject the same global frame sequence number and the same high-precision acquisition timestamp into all sensor data in each sampling period to obtain the original synchronous data frame.

[0113] LabVIEW Data Processing Module 202 is used to perform parallel signal conditioning on sensor data of different physical quantities in the original synchronous data frame based on the invariant constraints of global frame sequence number and high-precision acquisition timestamp, so as to obtain reconstructed sensor data frame. When all sensor data belonging to the same global frame sequence number has completed signal conditioning, an atomic transmission unit is generated. The atomic transmission unit includes global frame sequence number, high-precision acquisition timestamp and reconstructed sensor data frame.

[0114] Synchronization module 203 is used to store the reconstructed sensor data frames corresponding to the global frame sequence numbers into the synchronization frame queue in ascending order of the global frame sequence numbers, based on the global frame sequence numbers of each atomic transmission unit.

[0115] The time distortion correction module 204 is used to calculate the target rendering time of the current visualization rendering cycle based on the current local time of the visualization client and the preset buffer time constant in each visualization rendering cycle, and to make a time alignment decision based on the high-precision acquisition timestamp and target rendering time corresponding to the reconstructed sensor data frame at the head of the synchronization frame queue, so as to obtain the reconstructed sensor data frame to be rendered.

[0116] The UE5 visualization module 205 is used to perform time interpolation calculations based on the reconstructed sensor data frame to be rendered and the reconstructed sensor data frame rendered in the previous visualization rendering loop, to generate a synchronous rendering dataset that is aligned with the target rendering time; the synchronous rendering dataset is used to drive the unified update of the 3D model and the visualization interface.

[0117] In one embodiment, the real-time data monitoring module 201 is further configured to:

[0118] A global sampling clock signal matching a preset sampling period is generated based on a unified hardware sampling clock. The global sampling clock signal is used to instruct each sensor to trigger an analog-to-digital conversion channel to start sampling once on the valid edge of each sampling clock signal, and to store the sampled value in the analog-to-digital conversion result register.

[0119] Based on the trigger time corresponding to the effective edge of the global sampling clock signal, a preset atomic read instruction is initiated; the atomic read instruction is used to instruct the latest sampled value of each channel to be read from the analog-to-digital conversion result register at once to obtain sensor data;

[0120] Based on a high-precision clock source synchronized with Coordinated Universal Time, the time scale value of the atomic read command is obtained and used as a high-precision acquisition timestamp;

[0121] Based on a monotonically increasing global counter, a unique increment value is assigned to the sensor data corresponding to the atomic read command, which serves as the global frame sequence number;

[0122] The global frame sequence number, high-precision acquisition timestamp, and sensor data are packaged together into a predefined memory data block to obtain the original synchronization data frame.

[0123] In one embodiment, the LabVIEW data processing module 202 is further configured to:

[0124] Sensor data of different physical quantities in the original synchronization data frame are distributed to the corresponding signal processing channels for signal conditioning to obtain sensor reconstructed data output by each signal processing channel, and the sensor reconstructed data is stored in the frame reconstructed buffer according to the global frame sequence number.

[0125] The frame reassembly buffer is continuously monitored. Only when it is confirmed that all sensor reassembly data corresponding to the global frame sequence number has arrived in the frame reassembly buffer, all sensor reassembly data is encapsulated into a reconstructed sensor data frame, and combined with the global frame sequence number and high-precision acquisition timestamp, an atomic transmission unit is generated.

[0126] In one embodiment, in each visualization rendering loop, the target rendering time of the current visualization rendering loop is calculated based on the current local time of the visualization client and a preset buffer time constant. A time alignment decision is then made based on the high-precision acquisition timestamp corresponding to the reconstructed sensor data frame at the head of the synchronization frame queue and the target rendering time to obtain the currently rendered reconstructed sensor data frame, including:

[0127] The target rendering time for the current visualization rendering cycle is calculated based on the current local time of the visualization client and a preset buffer time constant; the formula for calculating the target rendering time is as follows: ,in, To visualize the client's current local time, This is the preset buffer time constant;

[0128] For the reconstructed sensor data frame at the head of the synchronization frame queue, the visible timestamp is calculated based on the corresponding high-precision acquisition timestamp and the current network latency estimate; the formula for calculating the visible timestamp is as follows: ,in, As a delay compensation factor, For a fixed processing pipeline delay compensation value, For high-precision timestamp collection, This is the current network latency estimate;

[0129] The visible timestamp is compared with the target rendering time of the current loop. If the visible timestamp is less than or equal to the target rendering time, the corresponding reconstructed sensor data frame is removed from the synchronization frame queue and determined as the current frame to be rendered, thus obtaining the current reconstructed sensor data frame to be rendered.

[0130] In one embodiment, a visual timestamp module is also included for:

[0131] The local time of arrival of the reconstructed sensor data frame at the visualization client is recorded, and combined with the data server's sending timestamp in the atomic transmission unit, the network round-trip delay time of the reconstructed sensor data frame is calculated; the formula for calculating the network round-trip delay time is as follows: ,in, To reach the local time of the visualization client, Send timestamps to the data server;

[0132] Based on the high-precision acquisition timestamp and the data server's transmission timestamp, the queuing delay of the reconstructed sensor data frame at the data server is calculated; the formula for calculating the queuing delay is as follows: ,in, For high-precision timestamp collection;

[0133] The sliding window averaging method is used to calculate the average value of the network round-trip delay and queuing delay of multiple reconstructed sensor data frames to obtain the current network delay estimate.

[0134] In one embodiment, the UE5 visualization module 205 is further configured to:

[0135] The visual timestamp corresponding to the reconstructed sensor data frame rendered in the previous visualization rendering loop is determined as the first visual timestamp, and the visual timestamp corresponding to the currently rendered reconstructed sensor data frame is determined as the second visual timestamp.

[0136] Based on the target rendering time of the current visualization rendering cycle, the normalized interpolation factor between the first and second visual timestamps is calculated; the formula for calculating the normalized interpolation factor is as follows: ,in, Presentation time for the target This is the first visible timestamp. This is the second visible timestamp;

[0137] For the reconstructed sensor data of all different physical quantities in the current reconstructed sensor data frame to be rendered and the reconstructed sensor data frame rendered in the previous visualization rendering loop, linear interpolation calculation is performed based on the normalized interpolation factor to generate a synchronous rendering dataset.

[0138] In one embodiment, the UE5 visualization module 205 is further configured to:

[0139] Based on the preset industrial equipment 3D model driving event processing function, the synchronous rendering dataset is transformed into a visual detection and positioning; the industrial equipment 3D model driving event processing function correspondingly sets the displacement interpolation data in the synchronous rendering dataset as the spatial offset of the rotor model, calculates and superimposes the high-frequency vibration effect based on the vibration interpolation data, and updates the color of the motor model through color mapping based on the temperature interpolation data.

[0140] Based on the preset user interface update event handling function, the physical quantity values ​​in the synchronous rendering dataset and the corresponding global frame sequence number are displayed in real time in the screen control.

[0141] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps in the above method embodiments.

[0142] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0143] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0144] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for enhancing the detection of industrial equipment, characterized in that, The method includes: Based on a unified hardware sampling clock, data from multiple sensors of industrial equipment are synchronously acquired according to a preset sampling period. All sensor data within each sampling period are injected with the same global frame sequence number and the same high-precision acquisition timestamp to obtain the original synchronous data frame. Based on the invariant constraints of the global frame sequence number and the high-precision acquisition timestamp, the sensor data of different physical quantities in the original synchronous data frame are subjected to parallel signal conditioning to obtain a reconstructed sensor data frame. After all sensor data belonging to the same global frame sequence number have completed signal conditioning, an atomic transmission unit is generated. The atomic transmission unit includes the global frame sequence number, the high-precision acquisition timestamp, and the reconstructed sensor data frame. Based on the global frame sequence number of each of the atomic transmission units, the reconstructed sensor data frames corresponding to the global frame sequence number are stored in the synchronization frame queue in ascending order of the global frame sequence number. In each visualization rendering loop, the target rendering time of the current visualization rendering loop is calculated based on the current local time of the visualization client and the preset buffer time constant. Time alignment decision is made based on the high-precision acquisition timestamp corresponding to the reconstructed sensor data frame at the head of the synchronization frame queue and the target rendering time to obtain the reconstructed sensor data frame to be rendered. Based on the currently rendered reconstructed sensor data frame, combined with the reconstructed sensor data frame rendered in the previous visualization rendering loop, time interpolation calculation is performed to generate a synchronous rendering dataset aligned with the target presentation time; the synchronous rendering dataset is used to drive the unified update of the 3D model and visualization interface.

2. The method according to claim 1, characterized in that, The method involves synchronously acquiring data from multiple sensors of industrial equipment based on a unified hardware sampling clock and a preset sampling period. All sensor data within each sampling period are injected with the same global frame sequence number and the same high-precision acquisition timestamp to obtain the original synchronized data frame, including: A global sampling clock signal matching the preset sampling period is generated based on the unified hardware sampling clock; the global sampling clock signal is used to instruct each sensor to trigger an analog-to-digital conversion channel to start sampling once on the valid edge of each sampling clock signal, and to store the sampled value in the analog-to-digital conversion result register; Based on the trigger time corresponding to the effective edge of the global sampling clock signal, a preset atomic read instruction is initiated; the atomic read instruction is used to instruct the latest sampled value of each channel to be read from the analog-to-digital conversion result register at once to obtain the sensor data; Based on a high-precision clock source synchronized with Coordinated Universal Time, the time scale value for completing the atomic read instruction is obtained, and the time scale value is used as the high-precision acquisition timestamp; Based on a monotonically increasing global counter, a unique increment value is assigned to the sensor data corresponding to the atomic read instruction, which serves as the global frame sequence number; The global frame sequence number, the high-precision acquisition timestamp, and the sensor data are packaged together into a predefined memory data block to obtain the original synchronization data frame.

3. The method according to claim 1, characterized in that, Based on the invariant constraints of the global frame sequence number and the high-precision acquisition timestamp, the sensor data of different physical quantities in the original synchronous data frame are subjected to parallel signal conditioning to obtain a reconstructed sensor data frame. After all sensor data belonging to the same global frame sequence number have completed signal conditioning, an atomic transmission unit is generated, including: Sensor data of different physical quantities in the original synchronization data frame are distributed to the corresponding signal processing channels for signal conditioning to obtain sensor reconstructed data output by each signal processing channel, and the sensor reconstructed data is stored in the frame reconstructed buffer according to the global frame sequence number. The frame reassembly buffer is continuously monitored. Only when it is confirmed that all sensor reassembly data corresponding to the global frame sequence number has arrived at the frame reassembly buffer is all the sensor reassembly data encapsulated into the reconstructed sensor data frame, and the atomic transmission unit is generated by combining the global frame sequence number and the high-precision acquisition timestamp.

4. The method according to claim 1, characterized in that, In each visualization rendering loop, the target rendering time for the current visualization rendering loop is calculated based on the current local time of the visualization client and a preset buffer time constant. A time alignment decision is then made based on the high-precision acquisition timestamp corresponding to the reconstructed sensor data frame at the head of the synchronization frame queue and the target rendering time to obtain the currently rendered reconstructed sensor data frame, including: The target rendering time for the current visualization rendering cycle is calculated based on the current local time of the visualization client and a preset buffer time constant; the formula for calculating the target rendering time is as follows: ,in, To visualize the client's current local time, This is the preset buffer time constant; For the reconstructed sensor data frame at the head of the synchronization frame queue, a visible timestamp is calculated based on the corresponding high-precision acquisition timestamp and the current network latency estimate; the calculation formula for the visible timestamp is as follows: ,in, As a delay compensation factor, For a fixed processing pipeline delay compensation value, This refers to the high-precision acquisition timestamp. This is the current network latency estimate; The visible timestamp is compared with the target rendering time of the current loop. If the visible timestamp is less than or equal to the target rendering time, the corresponding reconstructed sensor data frame is removed from the synchronization frame queue and determined as the current frame to be rendered, thus obtaining the reconstructed sensor data frame to be rendered.

5. The method according to claim 4, characterized in that, The reconstructed sensor data frame at the head of the synchronization frame queue is used to calculate a visible timestamp based on the corresponding high-precision acquisition timestamp and the current network latency estimate, including: The local time of arrival of the reconstructed sensor data frame at the visualization client is recorded, and combined with the data server's transmission timestamp in the atomic transmission unit, the network round-trip delay time of the reconstructed sensor data frame is calculated; the formula for calculating the network round-trip delay time is as follows: ,in, To reach the local time of the visualization client, Send timestamps to the data server; Based on the high-precision acquisition timestamp and the data server's transmission timestamp, the queuing delay of the reconstructed sensor data frame at the data server is calculated; the formula for calculating the queuing delay is as follows: ,in, This refers to the high-precision acquisition timestamp; The sliding window averaging method is used to calculate the average value of the network round-trip time and queuing delay of multiple reconstructed sensor data frames to obtain the current network delay estimate.

6. The method according to claim 4, characterized in that, The process involves performing time interpolation calculations based on the currently rendered reconstructed sensor data frame, combined with the reconstructed sensor data frames rendered in the previous visualization rendering loop, to generate a synchronized rendering dataset aligned with the target rendering time. This includes: The visual timestamp corresponding to the reconstructed sensor data frame rendered in the previous visualization rendering loop is determined as the first visual timestamp, and the visual timestamp corresponding to the currently rendered reconstructed sensor data frame is determined as the second visual timestamp. Based on the target rendering time of the current visualization rendering cycle, a normalized interpolation factor between the first visual timestamp and the second visual timestamp is calculated; the formula for calculating the normalized interpolation factor is as follows: ,in, Presentation time for the target This is the first visible timestamp. This is the second visible timestamp; For the reconstructed sensor data of all different physical quantities in the currently rendered reconstructed sensor data frame and the reconstructed sensor data frame rendered in the previous visualization rendering cycle, linear interpolation calculation is performed based on the normalized interpolation factor to generate the synchronous rendering dataset.

7. The method according to claim 6, characterized in that, The method further includes: Based on the preset industrial equipment 3D model driving event processing function, the synchronous rendering dataset is transformed into a visual detection and positioning; the industrial equipment 3D model driving event processing function correspondingly sets the displacement interpolation data in the synchronous rendering dataset as the spatial offset of the rotor model, calculates and superimposes the high-frequency vibration effect based on the vibration interpolation data, and updates the color of the motor model through color mapping based on the temperature interpolation data. Based on the preset user interface update event handling function, the physical quantity values ​​in the synchronous rendering dataset and the corresponding global frame sequence number are displayed in the screen control in real time.

8. A detection enhancement system for industrial equipment, characterized in that, The system includes: The real-time data monitoring module is used to synchronously collect data from multiple sensors of industrial equipment according to a preset sampling period based on a unified hardware sampling clock, and inject the same global frame sequence number and the same high-precision acquisition timestamp into all sensor data in each sampling period to obtain the original synchronous data frame. The LabVIEW data processing module is used to perform parallel signal conditioning on sensor data of different physical quantities in the original synchronous data frame based on the invariant constraints of the global frame sequence number and the high-precision acquisition timestamp, to obtain a reconstructed sensor data frame. After all sensor data belonging to the same global frame sequence number have completed signal conditioning, an atomic transmission unit is generated. The atomic transmission unit includes the global frame sequence number, the high-precision acquisition timestamp, and the reconstructed sensor data frame. The synchronization module is used to store the reconstructed sensor data frames corresponding to the global frame sequence numbers into the synchronization frame queue in ascending order of the global frame sequence numbers, based on the global frame sequence numbers of each of the atomic transmission units. The time distortion correction module is used to calculate the target rendering time of the current visualization rendering cycle based on the current local time of the visualization client and a preset buffer time constant in each visualization rendering cycle, and to make a time alignment decision based on the high-precision acquisition timestamp corresponding to the reconstructed sensor data frame at the head of the synchronization frame queue and the target rendering time, so as to obtain the reconstructed sensor data frame to be rendered. The UE5 visualization module is used to perform time interpolation calculations based on the currently rendered reconstructed sensor data frame and the reconstructed sensor data frame rendered in the previous visualization rendering loop, to generate a synchronous rendering dataset that is aligned with the target presentation time; the synchronous rendering dataset is used to drive the unified update of the 3D model and the visualization interface.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images