PLC remote monitoring data processing method and system based on edge computing

Abstract

The invention belongs to the technical field of digital information transmission, and particularly relates to a PLC remote monitoring data processing method and system based on edge computing, and the method comprises the steps: an edge gateway reads key time sequence data in a PLC register, maintains a real-time sliding window in a memory, and calculates an arithmetic mean value of all sampling data; calculating the product of the fluctuation component and the amplitude weight component to obtain a local non-stability index; obtaining a dynamic dead zone threshold value used for data compression at the current moment; and calculating an absolute value of a difference value between the actually measured data at the current moment and the uploaded data point at the previous moment, comparing the absolute value with the dynamic dead zone threshold value, and uploading the actually measured data at the current moment as effective data if the absolute value is greater than the dynamic dead zone threshold value. According to the invention, the data compression dead zone can be adaptively adjusted according to the current operation state of the equipment, and the bandwidth consumption and fault capture capability can be effectively balanced.

Description

Technical Field

[0001] This invention relates to the field of digital information transmission technology. More specifically, this invention relates to a PLC remote monitoring data processing method and system based on edge computing. Background Technology

[0002] PLCs play a crucial role in complete electrical equipment systems, responsible for collecting key timing data such as voltage, current, and temperature from core components like motors and frequency converters. To achieve efficient remote operation and maintenance, this massive amount of data typically needs to be uploaded in real-time to a cloud platform via an industrial gateway for storage and analysis.

[0003] In existing technology systems, data compression technology typically employs rotating door algorithms or fixed dead-zone algorithms. The core logic of these algorithms is to preset a fixed deviation threshold, and data is only uploaded when the data change exceeds this threshold. However, with long-term operation of the equipment, mechanical wear and aging of electrical components can lead to a gradual increase in background noise during normal operation. If the relatively small threshold set at the factory is continued to be used, the background noise generated by mechanical wear and aging of electrical components will cause critical timing data to be uploaded frequently, resulting in a surge in bandwidth costs. If a large threshold is set directly at the factory, in the early stable operation phase of the equipment, minute feature signals will be filtered out, leading to missed detections.

[0004] Addressing the issues of background noise interference caused by aging operating equipment and insufficient sensitivity of fixed dead-zone algorithms, how to adaptively adjust the compression dead zone according to the aging degree of operating equipment, while saving bandwidth and not losing key fault characteristics, has become an urgent technical problem to be solved in the field of lightweight processing methods and systems for remote monitoring data of programmable logic controllers based on edge computing. Summary of the Invention

[0005] To address the contradiction between the inability of fixed dead-zone algorithms to adapt to equipment aging noise and early feature capture, this invention provides a PLC remote monitoring data processing method and system based on edge computing. This method can adaptively adjust the data compression dead zone according to the current operating status of the equipment, effectively balancing bandwidth consumption and fault detection capabilities.

[0006] In a first aspect, the present invention provides a PLC remote monitoring data processing method based on edge computing, comprising: an edge gateway reading key timing data in a PLC register at a preset sampling frequency; maintaining a real-time sliding window in memory and calculating the arithmetic mean of all sampled data within the real-time sliding window, using the arithmetic mean as the current trend benchmark value; obtaining a local non-stationarity index characterizing the current operating state of the device by calculating the product of the fluctuation component and the amplitude weight component based on the original sampled values, the trend benchmark value, and the range information of the window data within the real-time sliding window; obtaining preset dead zone lower limit, dead zone upper limit, and sensitivity adjustment coefficient; calculating an adjustment factor based on the proportional relationship between the local non-stationarity index and the sensitivity adjustment coefficient; performing interpolation calculation between the dead zone lower limit and the dead zone upper limit based on the adjustment factor to obtain the dynamic dead zone threshold for data compression at the current moment; calculating the absolute value of the difference between the measured data at the current moment and the data points uploaded at the previous moment; comparing the absolute value with the dynamic dead zone threshold; if the absolute value is greater than the dynamic dead zone threshold, uploading the measured data at the current moment as valid data.

[0007] By adopting the above technical solution, and by maintaining a real-time sliding window in memory and calculating the arithmetic mean as the trend benchmark value, the central trajectory in the original fluctuation data can be effectively removed and high-frequency random noise interference can be eliminated. The local non-stationarity index is derived based on the product of the fluctuation component and the amplitude weight component, which can accurately assess the roughness of equipment operation and identify the background noise enhancement caused by equipment aging. By using the dynamic dead zone threshold to interpolate between the lower limit and the upper limit of the dead zone, the bandwidth consumption and fault detection capability can be effectively balanced, ensuring high sensitivity during the stable period of the equipment and filtering invalid fluctuations during the high noise period of aging.

[0008] Preferably, calculating the arithmetic mean of all sampled data within the real-time sliding window includes: summing all the original sampled data within the real-time sliding window, and dividing the sum by the length of the real-time sliding window to obtain the trend benchmark value.

[0009] Preferably, the adjustment factor is calculated based on the ratio between the local nonstationarity index and the sensitivity adjustment coefficient. The dynamic dead zone threshold for data compression at the current moment is obtained by interpolating the adjustment factor between the lower and upper dead zone limits. Specifically, this includes: calculating the sum of the sensitivity adjustment coefficient and the local nonstationarity index; calculating the ratio obtained by dividing the sensitivity adjustment coefficient by the sum; subtracting the ratio from the value 1 to obtain the adjustment factor; calculating the difference between the upper and lower dead zone limits; multiplying the difference by the adjustment factor to obtain the dynamic increment; and adding the lower dead zone limit to the dynamic increment to obtain the dynamic dead zone threshold for data compression at the current moment.

[0010] By adopting the above technical solution, the adjustment factor is calculated using the proportional relationship between the sensitivity adjustment coefficient and the local non-stationarity index. Based on this, a dynamic increment is calculated between the lower limit and the upper limit of the dead zone to obtain the dynamic dead zone threshold used for data compression at the current moment. An adaptive dynamic dead zone model with reverse suppression logic is constructed, which enables the dynamic dead zone threshold to automatically adjust with the device noise level. During the stable period of the device, the dead zone automatically shrinks to maintain high sensitivity, and during the high noise period of device aging, the dead zone automatically widens to filter out invalid fluctuations. This solves the problem of frequent uploads and soaring bandwidth costs caused by the increase in background noise after long-term operation of the fixed threshold algorithm.

[0011] Preferably, after uploading the measured data at the current moment as valid data, the method further includes: updating the data points uploaded at the previous moment to the measured data at the current moment; setting a maximum forced upload time interval, and forcibly uploading the measured data at the current moment when the time since the last data upload exceeds the maximum forced upload time interval.

[0012] By adopting the above technical solution, the data points uploaded at the previous moment are updated to the measured data at the current moment, and a maximum forced upload interval is set. After determining that the measured data at the current moment is valid data, the reference points are updated in a timely manner. At the same time, the forced upload mechanism with timeout is used to prevent the situation where no data is uploaded for a long time due to the dynamic dead zone threshold being too large. This ensures that the cloud platform can receive the necessary data points and perform linear interpolation based on the timestamp of the data to restore continuous monitoring curves for remote operation and maintenance analysis.

[0013] Preferably, the rated reference constant is taken as the rated parameter of the object being measured, and the rated reference constant is greater than zero.

[0014] Preferably, the sensitivity adjustment coefficient is a positive real number, used to adjust the response speed of the dynamic dead zone threshold as the local nonstationarity index changes.

[0015] Preferably, after the valid data is uploaded to the cloud platform, the method further includes: after the cloud platform receives the valid data with non-equidistant intervals, it performs linear interpolation based on the timestamp of the data to restore a continuous monitoring curve.

[0016] By adopting the above technical solution, the cloud platform is required to perform linear interpolation using timestamps after receiving valid data with non-equidistant intervals. This cloud processing method can effectively restore continuous monitoring curves, solve the problem of discontinuous time axis caused by data simplification, and ensure the continuous monitoring quality of the cloud analysis system while significantly reducing transmission bandwidth costs.

[0017] Preferably, the key timing data includes voltage, current, and temperature data of the motor or frequency converter; the edge gateway reads the key timing data via Modbus TCP protocol or OPC UA protocol.

[0018] Secondly, the present invention provides a PLC remote monitoring data processing system based on edge computing, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned PLC remote monitoring data processing method based on edge computing is implemented.

[0019] By adopting the above technical solution, the above-mentioned edge computing-based PLC remote monitoring data processing method generates a computer program and stores it in the memory so that it can be loaded and executed by the processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.

[0020] The beneficial effects of this invention are as follows: This invention effectively decouples equipment aging characteristics from data anomaly characteristics by introducing a local nonstationarity index. It can accurately identify background noise fluctuations caused by equipment aging and automatically relax the dynamic dead zone threshold to filter invalid data. This solves the technical problem of bandwidth consumption surges in the later stages of equipment aging in traditional fixed threshold algorithms. At the same time, during the stable operation phase of the equipment, the dynamic dead zone threshold automatically shrinks to the lower limit of the dead zone, ensuring that small voltage drops or current pulses and early fault signals can be keenly captured and uploaded. This achieves intelligent data processing that balances bandwidth saving and fault detection capabilities.

[0021] Furthermore, this invention only involves basic mathematical operation logic, with low overall computational complexity. It can be adapted to industrial edge gateways with limited computing resources and run efficiently on the local end, reducing dependence on hardware performance. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the PLC remote monitoring data processing method based on edge computing in this invention. Figure 2 This is a schematic diagram illustrating the trend of motor stator current and noise characteristics analysis; Figure 3 This is a schematic diagram illustrating the relationship between local nonstationarity and dynamic dead zone threshold response curves; Figure 4 This is a schematic diagram illustrating the performance comparison between the dynamic dead-time algorithm and the existing fixed dead-time algorithm. Detailed Implementation

[0023] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0024] This invention discloses a PLC remote monitoring data processing method based on edge computing. The method runs in an edge computing gateway that connects the PLC and the cloud platform. In one specific embodiment, the edge computing gateway can be an industrial IoT gateway based on ARM architecture. In the following description, the edge computing gateway is simply referred to as an edge gateway.

[0025] Reference Figure 1 A PLC remote monitoring data processing method based on edge computing includes steps S1-S4: S1. The edge gateway reads key timing data from the PLC register at a preset sampling frequency, maintains a real-time sliding window in memory, calculates the arithmetic mean of all sampled data within the real-time sliding window, and uses the arithmetic mean as the current trend benchmark value.

[0026] In an optional embodiment, the edge gateway first reads key timing data from the PLC register at a sampling frequency of 10ms to 100ms. Since single-point data lacks statistical significance and is easily affected by transient noise, the system maintains a data set with a length of [missing information - likely a value] in memory. The real-time sliding window is configured such that, in one specific implementation, N is set to 50, and its operating logic is as follows: whenever a new data point enters the real-time sliding window, the oldest data point is removed, meaning that the real-time sliding window always retains the latest 50 data points.

[0027] In one specific implementation, to distinguish between noise and real changes, it is first necessary to calculate the trend baseline of the data. Specifically, this is done by first defining the original data set within the current real-time sliding window as... The system calculates the moving average of the data within the window, which serves as the current trend baseline. The formula is: ; in, Indicates the current time Push forward The original values ​​of each sampling point.

[0028] To more clearly illustrate the processing method of the original data in this embodiment, the following example will be used for explanation: Assume the current sliding window length N is 5, and the original current data sequence collected at this time is as follows: Given [10.1, 10.2, 9.9, 10, 10.3], the current trend benchmark value is calculated as follows: .

[0029] Reference Figure 1 This moving average-based algorithm is represented in the graph as a trend baseline that reflects the smooth change of the equipment's operating center trajectory.

[0030] In this way, by constructing a sliding window and extracting the trend benchmark value, the central trajectory of equipment operation can be effectively extracted from the original fluctuation data, eliminating the interference of high-frequency random noise, and providing a stable reference benchmark for subsequent accurate assessment of the dispersion of the data.

[0031] S2. Based on the original sampled values ​​within the real-time sliding window, the trend benchmark value, and the range information of the window data, the local non-stationarity index characterizing the current operating state of the device is obtained by calculating the product of the fluctuation component and the amplitude weight component.

[0032] In an optional embodiment, when the equipment is aging, the collected data will exhibit two characteristics: first, the frequency and amplitude of fluctuations around the trend line will increase; second, there is a probability of large spikes appearing. Therefore, this embodiment of the invention constructs a local nonstationarity index. This is used to combine the normalized feature of the mean square error and the log-range feature, and its calculation method is as follows: ; in, For the current moment The local nonstationarity index; These are the original sampled values ​​within the sliding window; This is the trend baseline value at the corresponding time point; The rated reference constant is taken as the rated parameter of the object being measured, and ; This is a natural constant, which is set to 2.718 in this invention; This represents the window range.

[0033] To more clearly illustrate the role and calculation process of the local nonstationarity index, the following example will demonstrate its application: exist Given a value of 10.1, set the rated current. The value is 20; the absolute values ​​of the deviations of each point from the benchmark are 0, 0.1, 0.2, 0.1, and 0.2 respectively; then... ; Right now .

[0034] When equipment ages and vibrations increase The value will increase significantly.

[0035] Thus, by combining the normalized characteristics of the mean square error and the logarithmic range characteristics to calculate the local nonstationarity index, the roughness of the current equipment operation can be accurately assessed. This not only reflects the fluctuation range of the data around the trend line, but also takes into account the influence of occasional spikes, thereby enabling the identification of background noise enhancement caused by equipment aging.

[0036] S3. Obtain the preset lower dead zone limit, upper dead zone limit, and sensitivity adjustment coefficient. Calculate the adjustment factor based on the ratio between the local non-stationarity index and the sensitivity adjustment coefficient. Based on the adjustment factor, perform interpolation calculation between the lower dead zone limit and the upper dead zone limit to obtain the dynamic dead zone threshold used for data compression at the current moment.

[0037] In an optional embodiment, after obtaining the local nonstationarity exponent, a dynamic mapping is constructed using a variant of the sigmoid function, which is then transformed into the dead zone threshold required by the compression algorithm, thereby making it possible for the local nonstationarity exponent to be determined when the local nonstationarity exponent is obtained. When the value is small, the dead zone threshold approaches the minimum; when... When the value is large, the dead zone threshold transitions towards its maximum value. The method for calculating the dead zone threshold at the current time is as follows: ; in, This represents the dynamic dead zone threshold at the current moment; The lower limit of the dead zone preset by the system; The maximum dead zone limit is preset for the system. This is the sensitivity adjustment coefficient, which is used to adjust the response speed.

[0038] To more clearly illustrate the role and calculation process of the dead zone threshold at the current moment, the following example will demonstrate this: exist Given a value of 0.00804, set 0.1 0.5 If the value is 1, then: ; The calculations above show that when the equipment is running smoothly, the equipment's... The value is small, and the calculated dead zone threshold Very close to the lower limit (0.1), maintaining high sensitivity.

[0039] When equipment shows signs of severe aging. The value becomes 1, at which point ; The calculations above show that when equipment aging is severe, the calculated dead zone threshold value increases significantly to accommodate the noise caused by equipment aging.

[0040] Thus, by establishing an adaptive dynamic dead-zone model with reverse suppression logic, the dead-zone threshold is automatically adjusted according to the noise level of the equipment: the dead zone automatically shrinks during the stable period of the equipment to maintain high sensitivity, and the dead zone automatically widens during the high noise period of equipment aging to filter out invalid fluctuations, thereby solving the rigidity problem of the fixed threshold algorithm.

[0041] S4. Calculate the absolute value of the difference between the measured data at the current moment and the data points uploaded at the previous moment. Compare the absolute value with the dynamic dead zone threshold. If the absolute value is greater than the dynamic dead zone threshold, then upload the measured data at the current moment as valid data.

[0042] In an optional embodiment, the edge gateway reads the measured data at the current time t in real time. And retrieve the data points uploaded in the previous moment. And calculate the absolute deviation between the two. .

[0043] like If the measured data is not greater than the dynamic dead zone threshold, it is determined that the measured data at the current time t is redundant data or background noise caused by equipment aging. In this case, the measured data at the current time t is discarded and not uploaded. like If the measured data is greater than the dynamic dead zone threshold, then the measured data at the current time t is determined to be valid data, and the current point is set to... Store in the send queue and update .

[0044] In addition, to prevent excessively large dead zones from causing prolonged periods without data uploads, a maximum forced upload interval is set, which forces an upload of the current value upon timeout. After receiving the data, the cloud server reconstructs a continuous curve using linear interpolation.

[0045] Reference Figure 2 The curve representing the trend baseline eliminated noise from the original data and successfully extracted the center trajectory.

[0046] Reference Figure 3 The value indicates that the local nonstationarity index increases with the increase of noise, which in turn increases the dynamic dead zone threshold, thus verifying the adaptability of the algorithm.

[0047] Reference Figure 4 ,exist Figure 4 In part a, existing fixed dead-time technology retains a large number of glitches in the aging, high-noise region, wasting bandwidth. Figure 4In curve b, the curve representing the technology of this invention is smooth in the same noise region, successfully filtering out invalid fluctuations. At the end of the curve, a fault mutation pulse is completely preserved, proving that key fault features are not lost while the data is compressed significantly.

[0048] In this way, intelligent data filtering is achieved through the judgment logic based on dynamic dead zone. This not only filters out redundant aging background noise and saves network transmission bandwidth, but also ensures that subtle early fault signals are captured when the dead zone tightens, thus avoiding missed detections.

[0049] This invention also discloses a PLC remote monitoring data processing system based on edge computing, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the PLC remote monitoring data processing method based on edge computing according to this invention is implemented.

[0050] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0051] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise expressly and specifically defined.

Claims

1. A PLC remote monitoring data processing method based on edge computing, characterized in that, include: The edge gateway reads key timing data from the PLC register at a preset sampling frequency, maintains a real-time sliding window in memory, calculates the arithmetic mean of all sampled data within the real-time sliding window, and uses the arithmetic mean as the current trend benchmark value. Based on the original sampled values, trend benchmark values, and range information of the window data within the real-time sliding window, a local non-stationarity index characterizing the current operating state of the equipment is derived by calculating the product of the fluctuation component and the amplitude weight component. Specifically, this includes: obtaining a preset rated benchmark constant; calculating the difference between each original sampled value within the real-time sliding window and the corresponding trend benchmark value; dividing the difference by the rated benchmark constant to obtain the normalized deviation; calculating the arithmetic mean of the squares of all normalized deviations; and taking the square root of this arithmetic mean to obtain the fluctuation component; calculating the difference between the maximum and minimum values ​​of the sampled data within the real-time sliding window as the window range value; calculating the natural logarithm of the sum of the natural constant and the window range value as the amplitude weight component; and multiplying the fluctuation component and the amplitude weight component to obtain the local non-stationarity index. Obtain the preset lower dead zone limit, upper dead zone limit, and sensitivity adjustment coefficient. Calculate the adjustment factor based on the ratio between the local nonstationarity index and the sensitivity adjustment coefficient. Based on the adjustment factor, perform interpolation calculation between the lower dead zone limit and the upper dead zone limit to obtain the dynamic dead zone threshold used for data compression at the current moment. Calculate the absolute value of the difference between the measured data at the current moment and the data points uploaded at the previous moment. Compare the absolute value with the dynamic dead zone threshold. If the absolute value is greater than the dynamic dead zone threshold, the measured data at the current moment is uploaded as valid data.

2. The PLC remote monitoring data processing method based on edge computing according to claim 1, characterized in that, The calculation of the arithmetic mean of all sampled data within the real-time sliding window includes: summing all the original sampled data within the real-time sliding window, dividing the summation result by the length of the real-time sliding window, and obtaining the trend benchmark value.

3. The PLC remote monitoring data processing method based on edge computing according to claim 1, characterized in that, The adjustment factor is calculated based on the proportional relationship between the local non-stationarity index and the sensitivity adjustment coefficient. Interpolation is then performed between the lower and upper dead-zone limits based on this adjustment factor to obtain the dynamic dead-zone threshold for data compression at the current moment. Specifically, this includes: Calculate the sum of the sensitivity adjustment coefficient and the local nonstationarity index, and then calculate the ratio obtained by dividing the sensitivity adjustment coefficient by the sum. The adjustment factor is obtained by subtracting the ratio from the numerical value of 1. Calculate the difference between the upper and lower dead zones, and multiply the difference by an adjustment factor to obtain the dynamic increment; Add the lower dead zone limit to the dynamic increment to obtain the dynamic dead zone threshold used for data compression at the current moment.

4. The PLC remote monitoring data processing method based on edge computing according to claim 3, characterized in that, After uploading the measured data at the current moment as valid data, the process also includes: Update the data points uploaded in the previous moment to the measured data in the current moment; Set a maximum forced upload interval. When the time since the last data upload exceeds the maximum forced upload interval, force the upload of the current measured data.

5. The PLC remote monitoring data processing method based on edge computing according to claim 1, characterized in that, The rated reference constant is taken as the rated parameter of the object being measured, and the rated reference constant is greater than zero.

6. The PLC remote monitoring data processing method based on edge computing according to claim 3, characterized in that, The sensitivity adjustment coefficient is a positive real number, used to adjust the response speed of the dynamic dead zone threshold as the local nonstationarity index changes.

7. The PLC remote monitoring data processing method based on edge computing according to claim 1, characterized in that, After the valid data is uploaded to the cloud platform, it also includes: After receiving valid data with non-uniform spacing, the cloud platform performs linear interpolation based on the data's timestamp to reconstruct a continuous monitoring curve.

8. The PLC remote monitoring data processing method based on edge computing according to any one of claims 1 to 7, characterized in that, The key timing data includes voltage, current, and temperature data of the motor or frequency converter; the edge gateway reads the key timing data via Modbus TCP protocol or OPC UA protocol.

9. A PLC remote monitoring data processing system based on edge computing, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the PLC remote monitoring data processing method based on edge computing according to any one of claims 1 to 8.

Images