Industrial internet of things temperature transmitter data low power transmission system

Abstract

The present application relates to the technical fields of industrial internet of things and industrial automation monitoring, in particular to an industrial internet of things temperature transmitter data low-power transmission system, comprising: a local trend feature extraction unit for maintaining a sliding time window, calculating the change rate and change acceleration of the current physical process, generating current model parameters; a virtual shadow prediction unit for obtaining a virtual prediction value; a dynamic confidence interval decision unit for generating a dynamic confidence threshold, comparing the deviation with the dynamic confidence threshold, and generating a model failure trigger signal when the deviation is greater than the dynamic confidence threshold; a model parameterization resynchronization unit for constructing the current model parameters into a data frame, sending the data frame through a wireless module, and writing the current model parameters back to the virtual shadow prediction unit to update the sent parameters; the present application effectively solves the problem of key process detail loss caused by discretization in traditional dead zone schemes, and improves the analysis value of industrial monitoring data.

Description

Technical Field

[0001] This invention relates to the fields of industrial Internet of Things (IoT) and industrial automation monitoring technology, specifically to a low-power data transmission system for industrial IoT temperature transmitters. Background Technology

[0002] With the widespread application of industrial IoT technology, the demand for remote real-time monitoring of equipment operating status, especially temperature parameters, is becoming increasingly significant. As a key sensing node, industrial wireless temperature transmitters are usually limited by battery power capacity. Therefore, the energy efficiency of data transmission strategies directly determines the service life and maintenance costs of the equipment.

[0003] Currently, traditional data transmission and monitoring methods mainly rely on periodic reporting or dead-zone transmission mechanisms based on static thresholds. In periodic reporting mode, the transmitter mechanically transmits discrete sampling points containing a large amount of redundant information, leading to frequent activation of the radio frequency module and causing serious battery energy consumption. While technologies using fixed dead-zone transmission can filter out some invalid data, their decision logic is based solely on a single numerical difference, ignoring the dynamic characteristics of continuous changes in the physical process. This approach results in the receiver only acquiring discrete values, losing key process details characterizing the rate of temperature change and trend reversals. Furthermore, fixed decision thresholds cannot adapt to the changing industrial environment: in environments with strong electromagnetic interference or thermal noise, random fluctuations can easily trigger erroneous wake-ups and transmissions, leading to increased unnecessary power consumption; while under stable operating conditions, fixed thresholds may fail to detect subtle trend changes due to insufficient sensitivity. Therefore, how to avoid the loss of process details while strictly limiting power consumption, and how to resolve the contradiction between noise immunity and sensitivity that fixed thresholds cannot balance, in order to achieve high-fidelity monitoring of physical processes, has become an urgent problem to be solved in this field. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a low-power data transmission system for industrial IoT temperature transmitters. Specifically, the technical solution of this invention includes:

[0005] A low-power data transmission system for industrial IoT temperature transmitters includes:

[0006] A local trend feature extraction unit is used to maintain a sliding time window, which stores the most recent sampling points; perform differential or least squares operations on the original data stream within the sliding time window to calculate the rate of change and acceleration of the current physical process; generate current model parameters based on the rate of change and acceleration; the current model parameters include the current baseline value, the first-order slope coefficient, and the second-order acceleration coefficient.

[0007] A virtual shadow prediction unit is used to store the transmitted parameters; the transmitted parameters are used to deduce the theoretical temperature value at the current moment to obtain the virtual prediction value;

[0008] The dynamic confidence interval decision unit is used to receive the real sampled value and the virtual predicted value; calculate the absolute value of the difference between the real sampled value and the virtual predicted value to obtain the deviation; generate a dynamic confidence threshold based on the historical fluctuation variance of the device in steady state and a preset accuracy requirement; compare the deviation with the dynamic confidence threshold; and generate a model failure trigger signal when the deviation is greater than the dynamic confidence threshold.

[0009] The model parameterization resynchronization unit is used to respond to the model failure trigger signal; acquire the current model parameters; construct the current model parameters into a data frame; the data frame includes a new reference point, a new rate of change, a new acceleration coefficient, and an effective timestamp; transmit the data frame through a wireless module; and write the current model parameters back to the virtual shadow prediction unit to update the transmitted parameters.

[0010] Preferably, the current model parameters are polynomial coefficients that define the shape of the temperature change curve over the current short period of time.

[0011] Preferably, the virtual shadow prediction unit stores an empty model or an invalid state by default during the system power-on initialization phase, so that the deviation in the first sampling period will necessarily be greater than the dynamic confidence threshold, thereby triggering the first synchronization.

[0012] Preferably, the process of the dynamic confidence interval decision unit generating the dynamic confidence threshold includes: calling the dynamic noise benchmark evaluation submodule; analyzing the historical fluctuation variance of the device in steady state to determine the background noise level; and generating a tolerance interval that adapts to changes in environmental noise, which is used as the dynamic confidence threshold, in combination with the accuracy requirements set by the user.

[0013] Preferably, the write-back operation of the model parameterization resynchronization unit is used to make the model in the virtual shadow prediction unit consistent with the current physical process, so that the deviation calculated by the dynamic confidence interval decision unit at the next moment is zero, thereby suppressing subsequent communication requests.

[0014] Preferably, the rate of change corresponds to the first derivative, which characterizes how fast the temperature rises or falls; the acceleration of change corresponds to the second derivative, which characterizes the turning point of the trend.

[0015] Preferably, the dynamic confidence threshold is not a fixed value, but a tolerance range that adapts to the background noise level; the tolerance range represents the maximum reasonable error range that the system allows the model prediction to have.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] 1. This system constructs a positive coupling mechanism between the virtual shadow prediction unit and the physical process, and uses the transmitted parameters to deduce theoretical values. Communication is triggered only when the actual deviation exceeds the threshold. This on-demand transmission mode changes the energy waste caused by traditional periodic reporting and mechanical dead zone transmission. While ensuring monitoring accuracy, it greatly reduces the activation frequency of the radio frequency module, thereby significantly extending the battery life and achieving a balance between low power consumption and high fidelity.

[0018] 2. This system uses local trend feature extraction technology to transform discrete sampling points into model parameters that include the rate of change and acceleration of change, thus realizing the transmission of data in a higher dimension. The receiving end can reconstruct a continuous and smooth temperature curve based on the polynomial coefficients, fully preserving dynamic features such as the rate of temperature rise and trend reversal. This effectively solves the problem of loss of key process details due to discretization in traditional dead zone solutions, and enhances the analytical value of industrial monitoring data.

[0019] 3. This system introduces a dynamic confidence interval decision mechanism, which can assess the background noise level in real time based on the historical fluctuation variance of the equipment under steady state, and generate an adaptive tolerance interval accordingly. This mechanism automatically widens the threshold in a strong interference environment to filter out false triggers caused by random noise, and automatically narrows it in a stable environment to improve the sensitivity to capture small changes. It solves the contradiction between noise resistance and sensitivity that is difficult to balance with a fixed threshold, and enhances environmental adaptability.

[0020] 4. This system uses the write-back mechanism of the model parameterization resynchronization unit to force an update of the local shadow model after data transmission is completed, so that it immediately becomes consistent with the current physical process. This closed-loop calibration eliminates the continuous deviation caused by model lag, allowing the communication module to quickly return to the silent state after one calibration. This ensures real-time synchronization between the remote and local states, avoids redundant handshake protocols, and further improves channel utilization and system energy efficiency. Attached Figure Description

[0021] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0022] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0024] Example 1:

[0025] Please see Figure 1 A low-power data transmission system for industrial IoT temperature transmitters, including:

[0026] The local trend feature extraction unit is used to maintain a sliding time window, which stores the most recent sampling points. It performs differential or least squares operations on the original data stream within the sliding time window to calculate the rate of change and acceleration of the current physical process. Based on the rate of change and acceleration, it generates current model parameters, which include the current baseline value, the first-order slope coefficient, and the second-order acceleration coefficient.

[0027] The virtual shadow prediction unit is used to store the transmitted parameters and use the transmitted parameters to deduce the theoretical temperature value at the current moment to obtain the virtual prediction value.

[0028] The dynamic confidence interval decision unit is used to receive the real sampled value and the virtual predicted value; calculate the absolute value of the difference between the real sampled value and the virtual predicted value to obtain the deviation; generate a dynamic confidence threshold based on the historical fluctuation variance of the device in steady state and the preset accuracy requirement; compare the deviation with the dynamic confidence threshold; and generate a model failure trigger signal when the deviation is greater than the dynamic confidence threshold.

[0029] The model parameterization resynchronization unit is used to respond to the model failure trigger signal, obtain the current model parameters, and construct the current model parameters into a data frame. The data frame includes a new reference point, a new rate of change, a new acceleration coefficient, and an effective timestamp. The data frame is sent through a wireless module to write the current model parameters back to the virtual shadow prediction unit to update the sent parameters.

[0030] The local trend feature extraction unit is configured as the core sensing and computing module of the system, responsible for extracting continuous physical evolution patterns from discrete raw sampling sequences. This unit continuously ingests raw data streams generated by temperature sensors through a hardware interface. To achieve accurate trend capture, this unit allocates and maintains a sliding time window in memory that follows the first-in, first-out principle. The size of this window is set to cover a number of sampling points sufficient to reflect the minimum characteristic period of the current physical process. For example, it is set to a number of sampling points greater than 3 to 5 times the thermal response time constant of the measured object, to ensure that the window data has statistical significance. In each clock cycle of data update, this unit performs differential operations or least squares fitting operations on the data matrix within the window. During this operation, the system constructs a functional relationship between time and temperature to calculate a set of key indicators characterizing the dynamic characteristics of the current physical process: namely, the rate of change representing the speed of temperature change and the acceleration representing the turning point of the change trend. After the calculation is completed, this unit encapsulates the calculated data into a set of structured current model parameters and passes them to the next-level unit. This parameter set explicitly includes the intercept, linear coefficient, and quadratic coefficient of the fitted curve at the current moment.

[0031] The virtual shadow prediction unit constructs a computational model locally on the transmitter that is completely synchronized with the state of the remote server, using this as a logical reference for communication decisions. The input data of this unit comes directly from the parameters sent by the model parameterization resynchronization unit after communication is completed. In terms of operation, the unit performs polynomial extrapolation calculations based on the currently stored parameter set and the real-time system clock to deduce the temperature value that should theoretically be observed at the current moment, assuming that the physical laws have not changed. This calculation product is defined as the virtual prediction value and is sent to the decision unit. The shadow model referred to here is essentially a mathematical state machine running in the local microcontroller, and its state variables are strictly controlled by each wireless transmission action, ensuring the real-time consistency of the local and remote perceptions of the state of the measured object.

[0032] The dynamic confidence interval decision unit, as the system's logical gating hub, undertakes the decision-making task of balancing data fidelity and energy consumption. This unit receives real sampled values ​​from sensors and virtual predicted values ​​from shadow units in parallel. When the operation logic starts, this unit calculates the absolute value of the difference between the two to obtain the instantaneous deviation. More importantly, this unit abandons static threshold determination and instead generates a dynamic confidence threshold in real time through a weighted algorithm based on the historical fluctuation variance of the equipment during steady-state operation and the accuracy index pre-configured by the user. In the determination phase, this unit compares the instantaneous deviation with the dynamic threshold: only when the calculated deviation value significantly exceeds the reasonable fluctuation range defined by the threshold in a statistically significant manner is it determined that the current physical process has deviated from the original model trajectory, thereby generating and latching a model failure trigger signal.

[0033] The model parameterization resynchronization unit is used to perform the system's wireless communication tasks and complete the state closure. This unit is activated when the rising edge of the model failure trigger signal is detected. The execution process includes: directly retrieving the latest current model parameters from the feature extraction unit and packaging them with the current timestamp information to construct a dedicated data frame containing the new reference point, new rate of change, new acceleration coefficient, and effective timestamp; this unit drives the wireless radio frequency module to transmit the data frame to the remote server; after completing the transmission action, this unit immediately performs the critical write-back operation, forcibly overwriting the latest parameter set into the register of the virtual shadow prediction unit.

[0034] This system utilizes the thermodynamic inertia principle that the temperature changes of large-mass objects inevitably follow to construct a positive coupling mechanism between physical processes and communication behavior. Compared with existing technologies, this solution no longer mechanically transmits redundant data points, nor does it lose process details due to dead zones. Instead, it achieves on-demand communication: energy is consumed only for transmission when physical laws undergo substantial changes. This mechanism, while ensuring microampere-level power consumption constraints, achieves high-fidelity reconstruction and synchronization of process curve morphology, significantly extending battery life and enhancing data value.

[0035] Example 2:

[0036] The current model parameters are polynomial coefficients that define the shape of the temperature change curve over the current short period of time.

[0037] The current model parameters output by the local trend feature extraction unit are precisely defined mathematically. To achieve the goal of describing a continuous physical process with minimal data load, the system uses a second-order polynomial model to fit the temperature sequence within the sliding window. Under this definition, the current model parameters are specified as the set of coefficients of the polynomial. The computation unit uses least squares regression analysis with endpoint constraints, that is, introducing Lagrange multipliers or setting the current time as a fixed constraint point in the fitting operation, forcing the intercept c of the fitted curve to be strictly equal to the true sampled value at the current time, so as to ensure the numerical continuity when switching between the old and new models. Under this constraint, the best fitting curve is determined. ,in As the benchmark value corresponding to the constant term The slope corresponds to the coefficient of the linear term. The curvature corresponds to the coefficient of the quadratic term;

[0038] Using polynomial coefficients as the transmission carrier essentially increases the dimensionality of the transmitted content. The receiving end only needs to use the three received coefficients and timestamps to reconstruct a smooth and continuous temperature change curve through integration. This technique significantly reduces the amount of data transmitted while fully preserving the dynamic trend characteristics of temperature changes, effectively solving the technical problem of losing key process information due to data discretization in traditional dead-zone schemes.

[0039] Example 3:

[0040] The virtual shadow prediction unit stores an empty model or an invalid state by default during the system power-on initialization phase, so that the deviation in the first sampling period will necessarily be greater than the dynamic confidence threshold, thereby triggering the first synchronization.

[0041] The initialization logic of the virtual shadow prediction unit during the cold start phase is specified in detail. To ensure that the system can quickly establish a reference synchronization after power-on, the register of this unit is preset to an empty model or a specific invalid flag bit in the initialization program. Under this configuration, when the system collects the first real temperature data and sends it to the decision unit, since the output of the shadow model is invalid or zero, the calculated deviation value will be mathematically a maximum value or logically infinite. This value will inevitably exceed any dynamic confidence threshold set based on physical meaning, thereby forcibly triggering the model failure trigger signal.

[0042] This initialization strategy cleverly utilizes the system's error correction mechanism to complete the initial data synchronization by pre-setting a judgment condition that will inevitably fail. This design eliminates the data blind spots that may occur in the early stages of system startup due to the lack of a benchmark model, ensures seamless connection of the monitoring process, and avoids the need to write additional complex handshake protocol code.

[0043] Example 4:

[0044] The process of generating a dynamic confidence threshold by the dynamic confidence interval decision unit includes: calling the dynamic noise benchmark evaluation submodule; analyzing the historical fluctuation variance of the device under steady state to determine the background noise level; and generating a tolerance interval that adapts to changes in environmental noise, which is used as the dynamic confidence threshold, in combination with the accuracy requirements set by the user.

[0045] The dynamic noise benchmark evaluation submodule embedded in the dynamic confidence interval decision unit executes a strict threshold calculation process. This submodule uses a rolling statistical algorithm to analyze the sampled data of the device in real time within the most recent steady-state period. It performs a difference operation on the data sequence that matches the current model order. For example, for a second-order model, it uses second-order difference to eliminate the influence of the acceleration term or high-pass filtering to remove the low-frequency trend component of temperature change over time. It only calculates the variance or standard deviation of the high-frequency residual sequence after removal to quantify the current background noise level. The specific generation logic of the threshold follows the following principle: multiply the calculated background noise level by a preset confidence coefficient, and then add the absolute accuracy requirement value set by the user to finally synthesize a floating dynamic confidence threshold. Among them, the background noise level reflects the intensity of environmental electromagnetic interference and the thermal noise of the sensor itself, while the dynamic confidence threshold defines a tolerance range that adapts and expands with environmental conditions.

[0046] This adaptive threshold mechanism solves the adaptability problem of fixed threshold schemes under complex operating conditions. In a noisy environment, the threshold is automatically widened to filter out false triggers caused by random interference and prevent unnecessary consumption of battery energy. In a stable environment with low noise, the threshold is automatically narrowed to improve the sensitivity to capture subtle trend changes. This dynamic adjustment mechanism ensures that the system is always at the optimal balance point between detection performance and energy efficiency under different operating conditions.

[0047] Example 5:

[0048] The write-back operation of the model parameterization resynchronization unit is used to make the model in the virtual shadow prediction unit consistent with the current physical process, so that the deviation calculated by the dynamic confidence interval decision unit at the next moment is zero, thereby suppressing subsequent communication requests.

[0049] The paper elaborates on the closed-loop effect of the write-back operation performed by the model parameterization resynchronization unit. When the latest model parameters are written into the virtual shadow prediction unit, the local prediction model is immediately corrected to a state that perfectly matches the current real physical process. In the following sampling period, when the dynamic confidence interval decision unit performs the comparison, its input virtual prediction value will be generated by the new model. This value is mathematically very close to the real sampled value, and the difference between the two will quickly converge to zero or contain only a small random noise component. This deviation value must fall within the range allowed by the dynamic confidence threshold.

[0050] This write-back mechanism builds the system's expected digital reshaping capability; through a one-time parameter calibration, the system automatically eliminates a series of subsequent transmission demands that may arise due to model lag, allowing the communication module to quickly return to a silent state; this one-time calibration and long-term silent working mode greatly improves channel utilization and the overall energy efficiency of the system, embodying the advanced monitoring concept that silence is control.

[0051] Example 6:

[0052] The rate of change corresponds to the first derivative, which characterizes how fast the temperature rises or falls; the acceleration of change corresponds to the second derivative, which characterizes the turning point of the trend.

[0053] The physical and mathematical correspondence between the parameters calculated by the local trend feature extraction unit is clarified; the rate of change is strictly mathematically the first derivative of temperature with respect to time, and its physical meaning is to quantify the heating or cooling rate of the measured medium at the current moment; the acceleration of change corresponds to the second derivative of temperature with respect to time, and its physical meaning is to describe the curvature or turning characteristics of the temperature change trend, that is, to reflect whether the temperature change process is accelerating or tending to level off.

[0054] By establishing a clear physical-mathematical mapping, this invention ensures that the transmitted data has clear physical process indicators; this enables the receiving system to directly use these parameters for process diagnosis based on the rate of change, making up for the shortcomings of existing technologies that only transmit discrete values ​​and lose process dynamic information, and significantly improving the usability and analytical value of industrial monitoring data.

[0055] Example 7:

[0056] The dynamic confidence threshold is not a fixed value, but rather a tolerance range that adapts to the background noise level; the tolerance range represents the maximum reasonable error range that the system allows for model predictions.

[0057] The non-steady-state characteristics of the dynamic confidence threshold are further defined; this threshold is no longer a hard-coded constant, but a variable that is positively correlated with the background noise level; logically, the tolerance range defined by this threshold represents the system's acceptance boundary for prediction errors: any deviation falling within this range is judged by the system as a reasonable fluctuation caused by environmental noise or measurement uncertainty, and no communication needs to be triggered; otherwise, it is considered that a substantial change has occurred in the physical process.

[0058] This embodiment emphasizes the robustness of the technical solution in non-ideal environments; by introducing a tolerance range that fluctuates with the noise level, the system can effectively avoid false change reporting caused by environmental interference, while ensuring keen perception of real physical changes when the noise background is low; this design gives the sensor a biological-like adaptive adjustment capability, enabling it to adapt to a wide range of application scenarios from the laboratory to harsh industrial sites.

[0059] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An industrial internet of things temperature transmitter data low power transmission system, characterized in that, The application relates to a temperature prediction system, comprising: a local trend feature extraction unit for maintaining a sliding time window storing a plurality of recent sampling points; performing differential operation or least square operation on the original data stream in the sliding time window to calculate the change rate and change acceleration of the current physical process, generating current model parameters based on the change rate and change acceleration, wherein the current model parameters include a current reference value, a first-order slope coefficient and a second-order acceleration coefficient; a virtual shadow prediction unit for storing transmitted parameters and deducing a theoretical temperature value at the current time to obtain a virtual prediction value by using the transmitted parameters; a dynamic confidence interval decision unit for receiving a real sampling value and the virtual prediction value; calculating the absolute value of the difference between the real sampling value and the virtual prediction value to obtain a deviation; generating a dynamic confidence threshold based on the historical fluctuation variance of the device in a steady state and a preset accuracy requirement, wherein the specific process comprises: calling a dynamic noise reference evaluation submodule; analyzing the historical fluctuation variance of the device in a steady state to determine the background noise level; combining the accuracy requirement set by the user to generate a tolerance interval that changes adaptively with the environmental noise as the dynamic confidence threshold; comparing the deviation with the dynamic confidence threshold; and generating a model failure trigger signal when the deviation is greater than the dynamic confidence threshold; a model parameterization resynchronization unit for responding to the model failure trigger signal; acquiring the current model parameters; constructing the current model parameters into a data frame; the data frame includes a new reference point, a new change rate, a new acceleration coefficient and an effective time stamp; sending the data frame through a wireless module; and writing the current model parameters back to the virtual shadow prediction unit to update the transmitted parameters.

2. The industrial internet of things temperature transmitter data low power transmission system of claim 1, wherein, The current model parameters are polynomial coefficients defining the shape of the temperature change curve in the current short time.

3. The industrial internet of things temperature transmitter data low power transmission system of claim 1, wherein, The virtual shadow prediction unit stores an empty model or an invalid state by default in the system power-on initialization stage, so that the deviation is necessarily greater than the dynamic confidence threshold in the first sampling period, thereby triggering the first synchronization.

4. The industrial internet of things temperature transmitter data low power transmission system of claim 1, wherein, The write-back operation of the model parameterization resynchronization unit is used to make the model in the virtual shadow prediction unit consistent with the current physical process, so that the deviation calculated by the dynamic confidence interval decision unit is zero at the next time, thereby suppressing the subsequent communication request.

5. The industrial internet of things temperature transmitter data low power transmission system of claim 1, wherein, The change rate corresponds to the first-order derivative, representing the speed of temperature rise or fall; and the change acceleration corresponds to the second-order derivative, representing the turning situation of the trend.

6. The industrial internet of things temperature transmitter data low power transmission system of claim 1, wherein, The dynamic confidence threshold is not a fixed value, but the tolerance interval that changes adaptively with the background noise level; and the tolerance interval represents the maximum reasonable error range allowed by the system for model prediction.

Images