Current sensor data calibration method for industrial environment

Abstract

The invention discloses a current sensor data calibration method for an industrial environment, and the method comprises the steps: collecting the original current sensor data of target equipment in the industrial environment in real time, and carrying out the dynamic preprocessing of the original current sensor data, and generating a first calibration data set; based on the first calibration data set, time domain features and frequency domain features are extracted, and a second feature vector set is constructed; iteratively calculating an optimal calibration parameter set through a self-adaptive optimization algorithm in combination with historical calibration data and the second feature vector set; and dynamically correcting the current data of the current sensor according to the optimal calibration parameter set, and outputting and updating the calibrated current value. According to the invention, through dynamic preprocessing, comprehensive time domain and frequency domain feature extraction and combination of historical calibration data, parameter adaptive optimization updating is carried out, and the problem that current sensor data calibration in an industrial environment is difficult to accurately adapt to dynamic change environment conditions in real time is effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of current sensor data processing and calibration, and particularly to a method for calibrating current sensor data for industrial environments. Background Art

[0002] With the continuous improvement of the intelligent and automated level of industrial production, current sensors, as important basic components for industrial production and equipment operation status monitoring, the accuracy of their data directly affects the safety, stability and economic benefits of industrial systems. Currently, widely used current sensors mainly include Hall sensors, Rogowski coils, fluxgate sensors, etc. These sensors achieve real-time monitoring of current by measuring the magnetic field intensity or induced voltage around current-carrying conductors. However, in complex industrial environments, due to the combined effects of electromagnetic interference, equipment operation load fluctuations, temperature and humidity changes, and the aging of the sensor's own performance, the measurement data of current sensors has large drifts and deviations, seriously affecting the validity and accuracy of the data.

[0003] Existing technologies usually correct sensors through simple regular calibration or static calibration methods. These methods lack the ability to adapt to real-time environmental changes and are difficult to effectively address the dynamic fluctuations of sensor data under complex environmental conditions. Existing technologies correct sensors by presetting fixed calibration parameters or relying on standard reference data in laboratory environments. This method cannot compensate in real time for the dynamic effects of frequently fluctuating current signals in industrial fields and environmental condition changes on measurement data.

[0004] At the same time, the analysis of the characteristics of measurement data in existing technologies is limited to single time-domain or frequency-domain characteristics, and the two characteristic information cannot be fully integrated to comprehensively evaluate measurement errors. In addition, existing calibration methods generally lack an adaptive adjustment and optimization mechanism and are difficult to use historical calibration data and on-site real-time data to automatically iteratively optimize and update parameters, resulting in poor stability and insufficient generalization ability of calibration results, and reducing the practical application effect of calibration methods.

[0005] Therefore, the existing current sensor data calibration technologies for industrial environments have problems such as being difficult to dynamically adapt to complex industrial environments, low calibration accuracy, and poor real-time performance. The present invention proposes a method for parameter adaptive optimization and update through dynamic preprocessing, comprehensive time-domain and frequency-domain feature extraction, and combination with historical calibration data, aiming to effectively solve the problem that it is difficult to accurately and real-time adapt to dynamic changing environmental conditions in the calibration of current sensor data in industrial environments. Summary of the Invention

[0006] The purpose of this section is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. In this section, as well as in the abstract and title of the present application, some simplifications or omissions may be made to avoid obscuring the purpose of this section, the abstract, and the title. However, such simplifications or omissions shall not be used to limit the scope of the present invention.

[0007] In view of the above existing problems, the present invention is proposed.

[0008] To solve the above technical problems, the present invention provides the following technical solutions: Real-time collect the original current sensor data of target devices in the industrial environment, and perform dynamic preprocessing on the original current sensor data to generate a first calibration data set;

[0009] Based on the first calibration data set, extract time-domain features and frequency-domain features to construct a second feature vector set;

[0010] Through an adaptive optimization algorithm, combine historical calibration data with the second feature vector set to iteratively calculate an optimal calibration parameter set;

[0011] Dynamically correct the current sensor data according to the optimal calibration parameter set, and output and update the calibrated current value.

[0012] As a preferred solution of the current sensor data calibration method for the industrial environment according to the present invention, the original current sensor data includes current values in a time series and corresponding industrial environment parameters.

[0013] As a preferred solution of the current sensor data calibration method for the industrial environment according to the present invention, the dynamic preprocessing includes noise suppression, power frequency interference filtering, and environmental parameter compensation, where:

[0014] Perform filtering on the collected instantaneous current values. Under preset conditions, if large-amplitude random fluctuations occur in consecutive multiple sampling periods, enable a higher-order filtering strategy to further suppress the noise;

[0015] Identify the main frequency of the power grid in the industrial environment, and filter out the interference components corresponding to this frequency and its harmonic components. When it is detected that the actual frequency of the industrial power grid fluctuates, adaptively adjust the range of interference filtering;

[0016] According to the collected environmental parameter information such as temperature, humidity, vibration intensity, and electromagnetic interference, judge the deviation caused by the above environmental conditions to the current measurement result, and calculate the corresponding compensation amount through a pre-established corresponding relationship, and add this compensation amount in real time when correcting the original current value;

[0017] Unify and package the current values obtained after noise suppression, power frequency interference filtering, and environmental parameter compensation with the corresponding timestamps, environmental parameters, and equipment operating condition identification information. Eliminate extreme outliers that do not conform to the sampling pattern, and retain the data information within the expected range. Define the retained and processed data set as the first calibration data set.

[0018] As a preferred embodiment of the current sensor data calibration method for industrial environments according to the present invention, based on the first calibration data set, extract time-domain features and frequency-domain features, including:

[0019] Calculate multiple time-domain features in each sampling sequence, where the time-domain features at least include the average value, fluctuation range, variance, skewness, and kurtosis;

[0020] Perform spectral analysis on the current sequence recorded in the first calibration data set, identify the amplitude changes within the main frequency, harmonics, and bandwidth ranges, and obtain the frequency-domain features;

[0021] Further analyze the remaining higher frequency bands after filtering the power frequency, extract the relative intensities and correlations of each harmonic component to distinguish different types of operating conditions;

[0022] Summarize the obtained time-domain features and frequency-domain features according to the time index.

[0023] As a preferred embodiment of the current sensor data calibration method for industrial environments according to the present invention, the construction of the second feature vector set includes:

[0024] Evaluate the correlation between the time-domain features, frequency-domain features, and historical calibration results, and screen out the features with less influence on the calibration effect;

[0025] Integrate the selected features into the same vector structure in sequence according to the time index number;

[0026] Establish corresponding feature vectors in each operating condition stage, so that each feature vector can correspond one by one with the subsequent historical calibration data;

[0027] Execute a normalization processing strategy for features with different dimensions to make each feature within the same numerical range;

[0028] Integrate all the screened and processed feature vectors to construct a data set containing multiple feature vector records, which is the second feature vector set.

[0029] As a preferred embodiment of the current sensor data calibration method for industrial environments according to the present invention, through an adaptive optimization algorithm, combine historical calibration data with the second feature vector set, and iteratively calculate the optimal calibration parameter set, including:

[0030] Analyze historical calibration data, set evaluation metrics according to the difference between the current deviation value and the true value, and add a constraint condition for limiting parameter distortion to the evaluation metrics, and measure the quality of the calibration parameters through the evaluation metrics;

[0031] In the initialization stage, assign an initial value to the calibration parameters;

[0032] In each iteration, input the second feature vector set and the corresponding historical calibration data into the adaptive optimization algorithm together, and the algorithm fine-tunes the current calibration parameters according to the evaluation result of the objective function;

[0033] If the degree of deviation reduction in this round of iteration meets the iteration condition, enter the next round of iteration and continuously approach the optimal solution;

[0034] If the degree of deviation reduction in this round of iteration does not meet the iteration condition, adjust the learning rate according to the hyperparameter strategy;

[0035] When the number of iterations reaches the preset upper limit, the optimization algorithm stops;

[0036] Take the calibration parameters obtained at this time as the optimal calibration parameter set, and use it to correct the current sensor data in the real-time environment.

[0037] As a preferred solution of the current sensor data calibration method for industrial environment according to the present invention, dynamically correct the current current sensor data according to the optimal calibration parameter set, including:

[0038] In the industrial control system, load the calibration model generated by using the optimal calibration parameter set;

[0039] Input the currently collected current sensor data together with the corresponding environmental information and the device working mode identifier into the calibration model;

[0040] The calibration model corrects the deviation of the input current value according to the correction strategy given by the optimal calibration parameter set;

[0041] Compare the corrected current value with the result corrected in the previous time period. If there is an abnormal jump in the short term, downgrade the correction model.

[0042] As a preferred solution of the current sensor data calibration method for industrial environment according to the present invention, output and update the calibrated current value, including:

[0043] Unify and store the corrected current value each time together with the time information, environmental information and the identification number of the optimal calibration parameters;

[0044] Regularly trigger the inspection of calibration parameters to determine whether it is necessary to restart the adaptive optimization algorithm for iterative update;

[0045] When the device undergoes large-scale maintenance, the operation and maintenance personnel manually trigger the system for recalibration;

[0046] When the corrected current value shows a large anomaly and still cannot be automatically corrected after repeated attempts, an alarm signal is issued and manual intervention is allowed.

[0047] Advantages of the present invention:

[0048] 1. By eliminating high-frequency noise interference, power frequency interference, and errors caused by environmental factors such as temperature, humidity, and electromagnetic interference in the data, it ensures that the data basis used in the subsequent calibration process is more real and reliable, provides a high-quality data source for subsequent feature extraction, and improves the overall data quality and credibility;

[0049] 2. Using time-domain features to reflect the trend, stability, and volatility of the current signal in the time dimension, and using frequency-domain features to reflect the characteristics of different frequency components in the signal and the distribution of harmonic components. The two features are fused and complementary, and can comprehensively depict the characteristics of the current signal under different working conditions and load states of industrial equipment, providing a more accurate and effective feature basis for subsequent adaptive parameter optimization, significantly improving the feature expression ability and distinguishing the characteristics of current data under different working conditions;

[0050] 3. By iteratively calculating and autonomously adjusting the calibration parameters, it realizes the real-time update of the calibration parameters with the change of the industrial environment and tends to the optimal value, ensuring that the calibration model can continuously and actively adapt to the dynamic changes of the industrial environment, improving the adaptability and stability of the calibration model, and effectively avoiding the limitations of traditional static parameter models;

[0051] 4. By using the optimized calibration parameters to real-time correct the collected current data, it significantly reduces the impact of the dynamic changes in the industrial environment on the current measurement accuracy, makes the real-time data output by the current sensor closer to the real current state, provides more accurate and reliable current measurement data input for industrial control systems and condition monitoring systems, improves the measurement accuracy and stability of the current sensor in industrial field applications, and effectively ensures the safe operation of industrial equipment and the accuracy of system control. Description of the Drawings

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. Among them:

[0053] Figure 1 This is a schematic flowchart of the current sensor data calibration method for industrial environments shown in the present invention. Specific embodiments

[0054] To make the above objects, features, and advantages of the present invention more apparent and understandable, the following will describe in detail the specific embodiments of the present invention with reference to the accompanying drawings of the specification. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention.

[0055] All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0056] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0057] According to an embodiment of the present invention, in combination with Figure 1 the flowchart shown, a current sensor data calibration method for industrial environments specifically includes the following steps:

[0058] S1. Real-time collect the original current sensor data of the target device in the industrial environment, and perform dynamic preprocessing on the original current sensor data to generate a first calibration data set. Herein, it should be noted in this step:

[0059] Obtain the time series of current sampling values from the output end of the current sensor of the target device in the industrial environment, and record the sampling time point corresponding to each sampling value;

[0060] While collecting, obtain the environmental parameter information of the industrial device, such as temperature, humidity, vibration intensity, and electromagnetic interference level;

[0061] If the industrial device has different working modes, while collecting the current value, store the working condition identification of the sampling data;

[0062] As an example, the original current sensor data at least includes instantaneous current measurement values, corresponding timestamps, environmental parameters, device working modes, and sensor ranges;

[0063] Furthermore, for the collected instantaneous current values, remove high-frequency random noise through a filter. If a large random fluctuation exceeding the threshold is detected in multiple consecutive sampling periods, automatically enable a higher-order filtering strategy to further enhance the noise suppression strength, so as to effectively reduce the influence of common transient pulse interference and high-frequency random noise in the industrial field on the subsequent calibration data.

[0064] Identify the main frequency of the power grid in the industrial environment, such as 60 Hz, and filter out the interference components corresponding to this frequency and its harmonic components. When it is detected that the actual frequency of the industrial power grid deviates, the filtering range is adjusted adaptively to avoid filtering deviation caused by grid instability;

[0065] Read the temperature, humidity, vibration intensity and electromagnetic interference environment parameter information collected by the target device, judge the deviation that these environmental conditions may cause to the current measurement, and calculate the corresponding compensation amount according to the pre-established relationship between environmental impact and current sensor response. When correcting the original current value, the calculated compensation amount is superimposed on the current value in real time, so that the corrected current measurement value can better reflect the actual power consumption load condition;

[0066] After the noise suppression, power frequency interference filtering and environmental compensation are all completed, the generated corrected current data is uniformly reviewed. If it is found that the values of some sampling points completely do not conform to the conventional sampling rules (such as extreme jumps), such data will be excluded;

[0067] Package the retained and processed current measurement values together with the corresponding timestamps, environmental parameters and working condition information to form a first calibration data set for subsequent steps to call;

[0068] Exemplarily, the expression form of the first calibration data set is:

[0069] First calibration data set = {(timestamp, corrected current value, environmental parameters, device working condition identifier,...),...}

[0070] Among them, each record contains the final current value obtained after noise suppression, power frequency interference filtering and environmental compensation, as well as the spatio-temporal attributes and device status descriptions corresponding to this value.

[0071] Preferably, through the above implementation process, the present invention realizes continuous acquisition and dynamic preprocessing of the original current sensor data in the industrial environment, and finally outputs a first calibration data set that can truly reflect the operation of the target device and the influence of its environmental factors, laying a solid foundation for the accuracy and applicability of the subsequent calibration algorithm.

[0072] S2. Based on the first calibration data set, extract time-domain features and frequency-domain features to construct a second feature vector set. Among them, it should be noted in this step that:

[0073] According to the first calibration data set output in step S1, segment the current sampling values and the corresponding timestamps to obtain multiple continuous sampling sequences;

[0074] For each sampling sequence segment, calculate several time-domain features, which at least include:

[0075] Average value: used to describe the overall level of the current signal during this period;

[0076] Fluctuation range: used to characterize the difference between the maximum and minimum values of the current signal during this period, reflecting the rapid jump situation;

[0077] Variance: used to quantify the degree of dispersion of the sampling sequence during this period;

[0078] Skewness: used to describe whether the distribution shape has a deviation;

[0079] Kurtosis: used to measure whether the distribution shape has a sharp peak;

[0080] Match the above time-domain features with the corresponding paragraph index information to initially form several time-domain feature records for subsequent combination with frequency-domain features;

[0081] After extracting the time-domain features of the current sequences of the same batch and the same paragraph, further perform spectral analysis on this sequence to identify the amplitude changes within the main frequency, harmonics, and bandwidth ranges;

[0082] Perform spectral splitting on the residual frequency band after filtering out the power frequency, extract the relative intensities of each harmonic component and the degree of correlation with other components to distinguish different industrial equipment operating conditions;

[0083] Finally, obtain several frequency-domain feature indicators, such as the main harmonic amplitude, total harmonic content, and energy distribution ratio within the bandwidth range, and associate them with the corresponding time-domain paragraph index to form frequency-domain feature records.

[0084] In an optional implementation, summarize the obtained time-domain features and frequency-domain features according to the time index to ensure that each time period has a corresponding set of time-domain and frequency-domain features. To improve the efficiency and accuracy of subsequent calibration, perform a correlation assessment on the obtained features with the previous calibration results. By analyzing the contribution degree of each feature in the existing calibration data, screen out the features that have less impact on the calibration effect to avoid introducing too many redundant dimensions;

[0085] Arrange the time-domain features and frequency-domain features retained through the above screening based on the time index number and integrate them into the same vector structure in sequence;

[0086] If the numerical dimensions of different features vary greatly, perform normalization processing to make each feature within a relatively consistent numerical range and reduce the deviation caused by inconsistent dimensions;

[0087] Unify and collect all the feature vectors that have completed screening and processing, record them in a data set, and form the second feature vector set;

[0088] Exemplarily, the mathematical expression form of the second feature vector set is:

[0089] Ω2 = {v1, v2, …, v I}

[0090] where Ω2 represents the second feature vector set, I represents the total number of feature vectors,

[0091] each v i is a feature vector in the i-th paragraph:

[0092] v i = (T 1,i , T 2,i , …, T a,i , F 1,i , F 2,i , …, F b,i )

[0093] where T 1,i , T 2,i , …, T a,i respectively represent multiple different eigenvalue calculated in the time domain of this paragraph, and F 1,i , F 2,i , …, F b,i respectively represent different eigenvalues obtained in the frequency domain analysis of this paragraph;

[0094] After normalization processing, the above-mentioned time domain features and frequency domain features will be in a unified numerical interval, which is convenient for iterative optimization in subsequent steps.

[0095] It should be noted that through the implementation of this step, multi-dimensional feature extraction and comprehensive processing of the current information in the first calibration data set are realized, and finally each feature is orderly summarized in the form of the second feature vector set. This second feature vector set provides a more accurate and rich input basis for the subsequent adaptive optimization algorithm, so as to achieve high accuracy and high adaptability in calibrating current data in a complex industrial environment.

[0096] S3. Through the adaptive optimization algorithm, combine the historical calibration data with the second feature vector set, and iteratively calculate the optimal calibration parameter set. Among them, it should be noted in this step that:

[0097] When analyzing the historical calibration data, obtain the deviation between the true current value corresponding to each data (such as measured by a standard device) and the predicted current value, and perform square sum processing on this deviation to measure the fitting quality of the current calibration parameter to the true value;

[0098] For example, the difference between the predicted value and the true value for each historical data record is accumulated and defined as a total error to characterize the overall calibration accuracy;

[0099] Furthermore, to prevent the calibration parameters from being distorted during the iteration process, a constraint term for the parameters themselves needs to be added to the error measurement index. For example, if the parameter change amplitude is too large compared to the previous iteration, a penalty is imposed on the excess part in the objective function to ensure the smoothness of parameter update;

[0100] This constraint term is introduced into the objective function by multiplying the regularization coefficient with the corresponding penalty function to meet the requirement of the smoothness of the calibration parameters;

[0101] Exemplarily, the objective function is defined as:

[0102]

[0103] where, Θ represents the calibration parameter vector, I true,i represents the true current value corresponding to this historical data, f(v i , Θ) represents the current parameter Θ acting on the feature vector v i to obtain the predicted current output, λ represents the regularization coefficient used to balance the importance of the error term and the constraint term, and Ω(Θ) is the parameter update amplitude penalty function used to suppress the excessive parameters;

[0104] Before starting the iteration, an initial value needs to be determined for the calibration parameter vector Θ. Since the initial parameters will affect the iteration convergence speed and the final calibration accuracy, if the parameters deviate too much from the actual values, the learning rate needs to be moderately adjusted in the early iterations to ensure that it can quickly approach a reasonable range;

[0105] During each iteration, all the feature records in the second feature vector set are retrieved, and the corresponding historical true current values are obtained;

[0106] Thus, the optimized batch data input for this iteration is formed, and the adaptive optimization algorithm will update the parameters accordingly;

[0107] After evaluating the objective function value corresponding to the current calibration parameter Θ k , the parameter update amount for this iteration is determined according to the error direction and magnitude:

[0108]

[0109] where, Θ k is the parameter vector at the k - th iteration, α k is the learning rate for this iteration, represents the objective function at Θ kThe gradient at [location] is used to indicate the direction and magnitude of parameter update;

[0110] If an adaptive optimization method is used, then α k is dynamically adjusted during the iteration process according to the cumulative gradient information to balance the convergence speed and the global search ability;

[0111] After calculating the gradient, it is checked in real time whether the upper limit of the parameter change amount set for Ω(Θ) is triggered. Once it is detected that the parameter exceeds the threshold interval, a penalty value is additionally added to the objective function to enhance the penalty for the default situation;

[0112] When judging whether the degree of deviation reduction in the current round of iteration reaches the predetermined condition, the following strategy is used for processing:

[0113] When the deviation in the current round of iteration drops significantly and the parameters do not default or fluctuate too much, the original learning rate is maintained to accelerate the convergence speed;

[0114] When the deviation reduction is not obvious or the volatility becomes larger, the learning rate is moderately reduced to ensure that the optimal solution can be approximated more stably in subsequent iterations;

[0115] When the number of iterations reaches the preset maximum threshold, it can be determined that the stop condition is satisfied;

[0116] The parameter vector obtained after the last iteration is denoted as Θ opt , and it is regarded as the optimal calibration parameter set output by this process; Exemplarily expressed as:

[0117]

[0118] where k * represents the round index when the iteration stops;

[0119] After obtaining the optimal calibration parameter set Θ opt , it is loaded into the subsequent step S4 to correct the current data collected in real time.

[0120] Preferably, in the embodiment of the present invention, through an adaptive optimization algorithm, combining historical calibration data and the second feature vector set, the objective function formed by the evaluation index and the regularization constraint term is used to iteratively update the calibration parameters, so as to effectively prevent the parameters from overfitting or distortion, and under the preset constraint conditions and learning rate regulation strategy, finally obtain Θ opt as the optimal calibration parameter set, and this optimal calibration parameter set can make flexible calibrations for complex situations such as multiple working conditions and multiple feature dimensions in the industrial environment, which is beneficial to realizing high-precision dynamic correction of current sensor data in the subsequent step S4.

[0121] S4. Dynamically correct the current sensor data according to the optimal calibration parameter set, and output and update the calibrated current value. It should be noted in this step that:

[0122] After the end of step S3, the optimal calibration parameter set obtained through the adaptive optimization algorithm has been saved in the industrial control system. Each time the system starts, this parameter set will be automatically read and loaded into the initialization module of the calibration model to ensure that the subsequent dynamic correction steps are based on the latest and optimal parameter information;

[0123] In an alternative implementation, the system sets a running mode identifier. For example, it automatically switches to the preheating mode during the startup phase, and then switches to the online correction mode after confirming that the acquisition and correction functions are normal. Through this mode switch, meaningless correction operations are avoided in the short time just after the system starts;

[0124] When the target device is running, the current sensor continuously outputs the original current measurement value. At the same time, the corresponding environmental parameters (such as temperature, humidity, electromagnetic interference level) and device operating condition identifiers (such as full load, no load, standby mode) at this moment are synchronously obtained;

[0125] Pack the data collected in real time into an input vector and input it into the calibration model for subsequent correction calculations;

[0126] Furthermore, the calibration model performs deviation correction operations on the input vector based on the loaded optimal calibration parameter set;

[0127] If an environmental mutation is detected (such as a sudden increase in temperature or a sudden increase in interference), the model automatically retrieves redundant features to reduce the impact of instantaneous extreme values on the correction result;

[0128] If the device operating condition identifier switches from no load to full load, the model calls the correction logic applicable to the high-load state in the optimal calibration parameter set to amplify or compensate the current value to follow the actual current response requirements;

[0129] For example, when a large motor starts, the sensor may measure a high impact current, and the environmental temperature has not changed. The calibration model uses the correction strategy corresponding to the startup condition in the optimal parameter set to identify and correct the instantaneous spike to obtain a more stable and relatively real startup current measurement value;

[0130] Exemplarily, the calibration model is:

[0131]

[0132] where a is the lower integration time point, b is the upper integration time point, p is the feature combination index from 1 to P, and P is the total number of multiple time-frequency feature combinations involved in the calibration model, is a comprehensive function used to describe the synergistic effect between environmental parameters and real-time current data. η represents the environmental information collected at time τ, ξ represents the original data of the current sensor at time τ, and Θ opt represents the optimal calibration parameter set finally obtained through the adaptive optimization algorithm in step S3, and γ p is the correction weight factor for the p-th characteristic function, δ(u) is the environmental interference intensity function within the corresponding time period of the integration variable u, τ is the outer integration variable in this model for traversing the time period from a to b, u is the inner integration variable for calculating the cumulative noise suppression within the range of 0 to τ, and I orr represents the finally output calibrated current value;

[0133] When the system generates the current corrected value, it will simultaneously obtain the correction result of the previous time period (or the previous sampling period) for comparative analysis. If there is an abnormal jump in a very short time, for example, the correction value suddenly soars from low to high, but no sudden increase in the load is detected, the model is determined to be mis-corrected and the system enters the degradation mode;

[0134] Exemplarily, the degradation mode specifically includes:

[0135] Temporarily use the calibration parameters of the previous time period to smooth the correction process;

[0136] Mark and save the abnormal data points;

[0137] Through this degradation process, unnecessary interference to the device operation is avoided, and false triggering of the downstream control system is reduced;

[0138] In an alternative embodiment, when the calibration model completes the dynamic correction of the current value, the correction result is output as the finally calibrated current value to the industrial control system;

[0139] Exemplarily, when outputting the calibrated current value each time, this value will be uniformly stored together with the corresponding timestamp, environmental information, and identification number of the optimal calibration parameters; for example, a data record is represented as follows:

[0140] {

[0141] "Timestamp": T_current,

[0142] "Calibrated current value": I_corrected,

[0143] "Temperature": Env_temp,

[0144] "Interference level": Env_emc,

[0145] "Load status": Load_mode,

[0146] "Optimal parameter ID":Param_opt_ID

[0147] }

[0148] The above storage facilitates retrospective analysis, that is, when a failure or fluctuation occurs, past data can be retrieved for comparison;

[0149] In an optional embodiment, when a significant change in the equipment operating conditions and environmental conditions is detected, the adaptability of the current optimal calibration parameters is automatically checked, wherein:

[0150] If the correction effect of the current parameters on the deviation continues to decrease over time, the adaptive optimization process in step S3 can be triggered again to iteratively update the parameters;

[0151] When industrial equipment undergoes large-scale maintenance or component replacement, maintenance personnel manually trigger the recalibration process to re-identify environmental characteristics and parameter matching;

[0152] When repeated corrections are still unable to suppress obvious deviations, the system will issue an alarm signal and transfer control to manual on-site inspection, and then resume normal calibration operation after the fault is eliminated.

[0153] It should also be noted that in this embodiment, the real-time dynamic correction of the current sensor data in the industrial environment is achieved through the above-mentioned implementation method, which not only can output and update the calibrated current value in time, but also provides complete data storage, anomaly detection and re-optimization triggering capabilities, thereby ensuring that a high current measurement accuracy and reliability are continuously maintained in a complex production environment.

[0154] The above-mentioned current sensor data preprocessing and feature extraction method can be performed using methods and means in the prior art, and will not be described in detail in this example.

[0155] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A current sensor data calibration method for industrial environments, characterized in that, Including: Collecting the original current sensor data of target devices in the industrial environment in real time, and dynamically preprocessing the original current sensor data to generate a first calibration data set; Extracting time-domain features and frequency-domain features based on the first calibration data set to construct a second feature vector set; Through an adaptive optimization algorithm, combining historical calibration data with the second feature vector set, iteratively calculating an optimal calibration parameter set; Dynamically correcting the current sensor data according to the optimal calibration parameter set, and outputting and updating the calibrated current value.

2. The method for calibrating current sensor data for an industrial environment according to claim 1, wherein The original current sensor data includes current values in a time series and corresponding industrial environment parameters.

3. The current sensor data calibration method for industrial environment according to claim 1 or 2, characterized in that, The dynamic preprocessing includes noise suppression, power frequency interference filtering, and environmental parameter compensation, where: Performing filtering processing on the collected instantaneous current values. Under preset conditions, if large-amplitude random fluctuations occur in consecutive sampling periods, a higher-order filtering strategy is enabled to further suppress the noise; Identifying the main frequency of the power grid in the industrial environment, and filtering out the interference components corresponding to this frequency and its harmonic components. When it is detected that the actual frequency of the industrial power grid fluctuates, the range of interference filtering is adaptively adjusted; According to the collected temperature, humidity, vibration intensity, and electromagnetic interference environment parameter information, judging the deviation of the above environmental conditions on the current measurement result, and calculating the corresponding compensation amount through the pre-established corresponding relationship, and adding this compensation amount in real time when correcting the original current value; Uniformly encapsulating the current values obtained after noise suppression, power frequency interference filtering, and environmental parameter compensation with the corresponding timestamps, environmental parameters, and equipment working condition identification information, removing extreme outliers that do not conform to the sampling rules, and retaining the data information within the expected range. The retained and processed data set is defined as the first calibration data set.

4. The current sensor data calibration method for industrial environment according to claim 3, characterized in that Based on the first calibration data set, extracting time-domain features and frequency-domain features, including: Calculating multiple time-domain features in each sampling sequence, and the time-domain features at least include average value, fluctuation range, variance, skewness, and kurtosis; Performing spectrum analysis on the current sequence recorded in the first calibration data set, identifying the amplitude changes within the main frequency, harmonics, and bandwidth ranges to obtain frequency-domain features; Performing further analysis on the remaining higher frequency bands after filtering out the power frequency, and extracting the relative intensity and correlation of each harmonic component to distinguish different types of working conditions; Summarizing the obtained time-domain features and frequency-domain features according to the time index.

5. The current sensor data calibration method for industrial environment according to claim 4, characterized in that, The constructing of the second feature vector set includes: Performing a correlation evaluation on the time-domain features, frequency-domain features, and historical calibration results, and screening out the features with less influence on the calibration effect; Integrating the screened features into the same vector structure in sequence according to the time index numbering; Respectively establishing corresponding feature vectors in each working condition stage, so that each feature vector can correspond one by one to the subsequent historical calibration data; Executing a normalization processing strategy on features with different dimensions, so that each feature is within the same numerical range; Integrating all the screened and processed feature vectors to construct a data set containing multiple feature vector records, which is the second feature vector set.

6. The current sensor data calibration method for industrial environment according to claim 1, characterized in that, Through an adaptive optimization algorithm, combining historical calibration data with the second set of feature vectors, iteratively calculate the optimal set of calibration parameters, including: Analyze historical calibration data, set evaluation metrics according to the difference between the current deviation value and the true value, and add a constraint condition for restricting parameter distortion to the evaluation metrics, and measure the quality of calibration parameters through the evaluation metrics; In the initialization stage, assign an initial value to the calibration parameters; In each iteration, input the second set of feature vectors and the corresponding historical calibration data into the adaptive optimization algorithm together, and the algorithm fine-tunes the current calibration parameters according to the evaluation result of the objective function; If the degree of deviation reduction in this round of iteration meets the iteration condition, enter the next round of iteration and continuously approach the optimal solution; If the degree of deviation reduction in this round of iteration does not meet the iteration condition, adjust the learning rate according to the hyperparameter strategy; When the number of iterations reaches the preset upper limit, the optimization algorithm stops; Use the calibration parameters obtained at this time as the optimal set of calibration parameters to correct the current sensor data in a real-time environment.

7. The method for calibrating current sensor data for industrial environments according to claim 6, wherein Dynamically correct the current sensor data according to the optimal set of calibration parameters, including: In an industrial control system, load the calibration model generated using the optimal set of calibration parameters; Input the currently collected current sensor data together with the corresponding environmental information and the device working mode identifier into the calibration model; The calibration model corrects the deviation of the input current value according to the correction strategy given by the optimal set of calibration parameters; Compare the corrected current value with the result corrected in the previous time period. If there is an abnormal jump in the short term, downgrade the correction model.

8. The method for calibrating current sensor data for an industrial environment according to claim 7, wherein Output and update the calibrated current value, including: Uniformly store the corrected current value each time together with the time information, environmental information, and the identification number of the optimal calibration parameters; Regularly trigger the inspection of calibration parameters to determine whether it is necessary to restart the adaptive optimization algorithm for iterative update; If the device undergoes large-scale maintenance, the operation and maintenance personnel manually trigger the system for re-calibration; When the corrected current value shows a large anomaly and still cannot be automatically corrected after repeated occurrences, send an alarm signal and allow manual intervention.

Images