Industrial sensor electromagnetic sensitivity self-adapting calibration system

Abstract

The application relates to the technical field of sensor electromagnetic sensitivity calibration, in particular to an industrial sensor electromagnetic sensitivity self-adaptive calibration system, which comprises a data acquisition module, an electromagnetic sensitivity representation degree calculation module, a clustering effect calculation module, a clustering module and an electromagnetic sensitivity calibration module.The data acquisition module is used for acquiring the output values of each sensor under different test conditions, and then obtaining the corresponding data points of each sensor and the electromagnetic sensitivity of each sensor.The electromagnetic sensitivity representation degree calculation module is used for calculating the electromagnetic sensitivity representation degree of each dimension.The clustering effect calculation module is used for calculating the clustering effect of each dimension.The clustering module is used for obtaining the weight of the dimension, and clustering all the data points to obtain different multi-dimensional clustering clusters.The electromagnetic sensitivity calibration module is used for obtaining the electromagnetic sensitivity of each multi-dimensional clustering cluster, and then obtaining the multi-dimensional clustering cluster to which the data points corresponding to the current sensor belong and the electromagnetic sensitivity of the current sensor.The application can guarantee the efficiency and accuracy of sensor electromagnetic sensitivity calibration.

Description

Technical Field

[0001] This invention relates to the field of sensor electromagnetic sensitivity calibration technology, and specifically to an adaptive calibration system for the electromagnetic sensitivity of industrial sensors. Background Technology

[0002] Electromagnetic susceptibility calibration of industrial sensors refers to the process of quantifying and calibrating the performance response capabilities of industrial sensors under different electromagnetic environments. It aims to evaluate the sensor's anti-interference capability and measurement stability when subjected to electromagnetic interference, thereby ensuring its accurate and reliable operation in complex industrial environments. Because industrial sensors are often used in high-voltage, high-current, high-frequency, or strong magnetic field environments, electromagnetic interference can cause sensor data fluctuations or distortions in these environments, thus affecting the safety and accuracy of industrial control systems. Therefore, electromagnetic susceptibility calibration of industrial sensors is of great significance.

[0003] Current technologies for calibrating the electromagnetic susceptibility of sensors typically employ manual calibration based on test data. This method is time-consuming, labor-intensive, and prone to errors, and it cannot meet the demands of automated system production. Therefore, there is an urgent need for an automated adaptive calibration system to improve the efficiency of electromagnetic susceptibility calibration for industrial sensors. Summary of the Invention

[0004] To address the aforementioned technical problems, the present invention aims to provide an adaptive calibration system for the electromagnetic susceptibility of industrial sensors, the specific technical solution of which is as follows:

[0005] One embodiment of the present invention provides an adaptive calibration system for the electromagnetic susceptibility of industrial sensors, the system comprising:

[0006] The data acquisition module is used to acquire the output values ​​of each sensor under different test conditions and treat each test condition as a dimension; it combines the output values ​​of a sensor under each dimension into a data point and obtains the electromagnetic sensitivity of each sensor;

[0007] The electromagnetic susceptibility characterization degree calculation module is used to calculate the electromagnetic susceptibility characterization degree of a dimension based on the distribution of output values ​​of different data points in a dimension.

[0008] The clustering effect calculation module is used to cluster the output values ​​under one dimension to obtain different one-dimensional clusters; and to obtain the clustering effect of that dimension based on the distribution of each one-dimensional cluster and the distribution of the output values ​​within each one-dimensional cluster.

[0009] The clustering module is used to obtain the weight of a dimension based on the degree of electromagnetic susceptibility representation and the clustering effect; and to combine the weights of each dimension to cluster all data points to obtain different multidimensional clusters.

[0010] The electromagnetic susceptibility calibration module is used to obtain the electromagnetic susceptibility of each multidimensional cluster based on the electromagnetic susceptibility of the sensor corresponding to each data point within each multidimensional cluster; and to determine the multidimensional cluster to which the data point corresponding to the current sensor belongs, wherein the electromagnetic susceptibility of the multidimensional cluster to which it belongs is the electromagnetic susceptibility of the current sensor.

[0011] Preferably, acquiring the electromagnetic sensitivity of each sensor includes:

[0012] The electromagnetic sensitivity of each sensor is determined by manual evaluation based on its output value under each test condition.

[0013] Preferably, the electromagnetic susceptibility characterization level of a dimension is calculated based on the distribution of output values ​​at different data points in that dimension, including:

[0014] Arrange the output values ​​of different data points in one dimension in ascending order to obtain the output value sequence corresponding to that dimension; obtain the absolute value of the difference between every two adjacent output values ​​in the output value sequence, and record it as the data interval between every two adjacent output values; calculate the difference between the maximum and minimum values ​​in the output value sequence and multiply it by the reciprocal of the standard deviation of the data interval between every two adjacent output values ​​to obtain the electromagnetic susceptibility characterization degree of that dimension.

[0015] Preferably, the clustering effect of a dimension is obtained based on the distribution of each one-dimensional cluster and the distribution of output values ​​within each one-dimensional cluster, including:

[0016] One-dimensional clusters corresponding to a given dimension are grouped into a set of one-dimensional clusters, which are arranged according to their output values. The absolute value of the difference between the maximum value in a given one-dimensional cluster and the minimum value in the next nearest one-dimensional cluster is recorded as the interval distance. The Euclidean distance between an output value and its nearest neighbor within a one-dimensional cluster is calculated and recorded as the minimum distance. The mean of the minimum distances of all output values ​​in the given one-dimensional cluster is then calculated and recorded as the mean minimum distance of the given one-dimensional cluster. A negative correlation mapping is performed using an exponential function with the natural constant as the base of the exponential function on the mean of the mean minimum distances of all one-dimensional clusters corresponding to that dimension to obtain the mapping result. This mapping result is multiplied by the mean of the interval distances between any two adjacent one-dimensional clusters in the set corresponding to that dimension to obtain the clustering effect for that dimension.

[0017] Preferably, the weight of a dimension is obtained based on the degree of electromagnetic susceptibility characterization and clustering effect, including:

[0018] The weight of a dimension is obtained by averaging the normalized value of the electromagnetic susceptibility representation of a dimension with the normalized value of the clustering effect.

[0019] Preferably, the electromagnetic susceptibility of each multidimensional cluster is obtained based on the electromagnetic susceptibility of the sensor corresponding to each data point within each multidimensional cluster, including:

[0020] The mode of the electromagnetic susceptibility of the sensor corresponding to each data point within a multidimensional cluster is taken as the electromagnetic susceptibility of that multidimensional cluster.

[0021] Preferably, determining the multidimensional cluster to which the data point corresponding to the current sensor belongs includes:

[0022] The reciprocal of the absolute value of the difference between the output value of a data point in one dimension within a multidimensional cluster and the output value of the corresponding data point in the same dimension of the current sensor is used as the similarity of that data point in that dimension. The average of the similarities of all data points in that dimension within the multidimensional cluster is calculated and denoted as the dimensional similarity. The sum of the dimensional similarities of all dimensions is denoted as the overall similarity. The reciprocal of the distance between the data point corresponding to the current sensor and the cluster center of the multidimensional cluster is normalized and added to the normalized value of the overall similarity to obtain the probability of belonging to the multidimensional cluster. The multidimensional cluster with the highest probability of belonging is the multidimensional cluster to which the data point corresponding to the current sensor belongs.

[0023] The embodiments of this invention have at least the following beneficial effects: This application obtains data points corresponding to multiple sensors and their electromagnetic susceptibility by taking the output values ​​of each sensor under different historical test conditions as a dimension, and combining the output values ​​of a sensor under different dimensions into a data point; then, it obtains the electromagnetic susceptibility characterization degree and clustering effect of each dimension, combines the electromagnetic susceptibility characterization degree and clustering effect to obtain the weight of each dimension, and finally combines the weight of each dimension to cluster all data points to obtain different multidimensional clusters. The stronger the influence of the dimension itself, the higher its influence in the multidimensional data point clustering process, so that the obtained multidimensional clusters provide a more accurate division of electromagnetic susceptibility; finally, it obtains the electromagnetic susceptibility of each multidimensional cluster, and then determines the multidimensional cluster to which the data point of the current sensor belongs. The electromagnetic susceptibility of the multidimensional cluster to which the current sensor belongs is the electromagnetic susceptibility of the current sensor, realizing automated calibration of the electromagnetic susceptibility of the sensor and improving calibration efficiency. Attached Figure Description

[0024] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a system block diagram of an adaptive calibration system for the electromagnetic sensitivity of an industrial sensor, provided as an embodiment of the present invention. Detailed Implementation

[0026] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an industrial sensor electromagnetic sensitivity adaptive calibration system proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0027] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0028] The following description, in conjunction with the accompanying drawings, details a specific scheme for an adaptive calibration system for the electromagnetic sensitivity of an industrial sensor provided by the present invention.

[0029] Example: The main application scenario of this invention is: when calibrating the electromagnetic sensitivity of industrial sensors, it is necessary to set up an electromagnetic interference environment and calibrate the electromagnetic sensitivity of the sensor based on the sensor data collected under various interference environments. The manual calibration process is repetitive and redundant, with low calibration efficiency, and is not suitable for the overall system automation. Therefore, it is necessary to calibrate the electromagnetic sensitivity of the sensor automatically.

[0030] Please see Figure 1 The diagram illustrates a system block diagram of an adaptive calibration system for the electromagnetic sensitivity of an industrial sensor, provided by an embodiment of the present invention. The system includes the following modules:

[0031] The data acquisition module is used to collect the output values ​​of each sensor under different test conditions and treat each test condition as a dimension; it combines the output values ​​of a sensor under each dimension into a data point and obtains the electromagnetic sensitivity of each sensor.

[0032] Electromagnetic susceptibility testing of industrial sensors involves setting up an electromagnetic interference (EMI) test environment. An radio frequency (RF) signal source is used to directionally radiate interference signals into the sensor at different frequencies and field strengths. The sensor's output value under the corresponding interference signals is then monitored and collected. The EMI test environment includes various test conditions, as shown in the table below:

[0033] Table 1

[0034]

[0035] Among them, A1 and B1, A2 and B2, A3 and B3, A4 and B4, and A5 and B5 represent five different test conditions, while C1, C2, C3, C4, and C5 represent the output values ​​of a sensor under these five test conditions. In historical tests, the electromagnetic susceptibility of the sensor was obtained by manually evaluating multiple test results using the sensor's output values ​​under different frequency bands and field strengths (different test conditions).

[0036] Because subsequent cluster analysis is required for each sensor to group sensors with similar electromagnetic susceptibility from historical tests together, and because the degree of interference to sensors varies under different test conditions (some sensors are particularly sensitive to specific frequencies or field strengths), it is necessary to differentiate the output values ​​of the same sensor under each test condition. Here, each test condition is used as a dimension; for example, frequency band A1 and field strength B1 correspond to one dimension, and the data values ​​under this dimension are the sensor's output values. Frequency band A2 and field strength B2 correspond to a second dimension, and so on, thus obtaining multiple dimensions composed of all test conditions. In this space composed of multiple dimensions, a data point consists of the output value of a sensor under each dimension (each test condition), with one data point corresponding to one sensor. All historical data points corresponding to all sensors are then input into this space for subsequent analysis.

[0037] The electromagnetic susceptibility characterization degree calculation module is used to calculate the electromagnetic susceptibility characterization degree of a dimension based on the distribution of output values ​​of different data points in a dimension.

[0038] Because the distribution of sensor sensitivity to electromagnetic interference varies across dimensions, some dimensions may have a higher representation of electromagnetic sensitivity. Therefore, it is necessary to increase the weight of such dimensions in the clustering. This results in a more accurate classification of sensors with different electromagnetic sensitivities based on the clustering results of all data points obtained.

[0039] This analysis examines the distribution of output values ​​from multiple sensors within each dimension. A larger distribution range of output values ​​from multiple sensors within a dimension indicates greater susceptibility to fluctuations in that dimension's data. This suggests that the sensors are generally more susceptible to electromagnetic interference in that dimension, making it a more volatile and electromagnetically sensitive dimension, thus providing a better representation of electromagnetic sensitivity. However, the distribution range may be expanded from individual data points and cannot reflect the overall general distribution within that dimension. Therefore, considering the uniformity of the distribution between output values ​​within that dimension, a more uniform distribution interval suggests a more likely uniform distribution across the entire range. Consequently, a larger distribution range coupled with a more uniform distribution interval indicates a better representation of the overall electromagnetic sensitivity of the output values ​​within the current dimension.

[0040] Therefore, the electromagnetic susceptibility characterization level of a dimension is calculated based on the distribution of output values ​​for different data points in a single dimension. Specifically, the output values ​​for a single dimension of different data points are arranged in ascending order to obtain the output value sequence corresponding to that dimension; the absolute value of the difference between every two adjacent output values ​​in the output value sequence is obtained and denoted as the data interval between every two adjacent output values; the difference between the maximum and minimum values ​​in the output value sequence is calculated and multiplied by the reciprocal of the standard deviation of the data interval between every two adjacent output values ​​to obtain the electromagnetic susceptibility characterization level of that dimension.

[0041] The specific calculation model for the degree of electromagnetic susceptibility characterization is as follows:

[0042] ,

[0043] in, This represents the degree of electromagnetic susceptibility characterization in the q-th dimension; and These represent the maximum and minimum values ​​of the output values ​​for different data points (different sensors) in this dimension, that is, the maximum and minimum values ​​in the output value sequence. The larger the data distribution range in this dimension, the more likely the output value of this dimension is to fluctuate. This indicates that the sensor is generally more susceptible to electromagnetic interference in this dimension, making it a variable dimension of electromagnetic sensitivity, and the better the characterization of electromagnetic sensitivity. This represents the set of absolute differences between any two adjacent output values ​​in the output value sequence; that is, the set of data intervals between any two adjacent output values. The standard deviation of all data intervals is denoted as . The smaller the standard deviation, the more uniform the overall distribution of the data within the distribution range. On the basis of a larger distribution range, this represents a better degree of representation of the overall electromagnetic susceptibility of the output values ​​in the current dimension.

[0044] This allows us to obtain the degree of electromagnetic susceptibility characterization for each dimension.

[0045] The clustering effect calculation module is used to cluster the output values ​​under one dimension to obtain different one-dimensional clusters; and to obtain the clustering effect of that dimension based on the distribution of each one-dimensional cluster and the distribution of the output values ​​within each one-dimensional cluster.

[0046] The above obtained the electromagnetic susceptibility characterization of each dimension. However, when the output values ​​are distributed in a concentrated block pattern across the entire range, the uniformity of the data may decrease, leading to a decrease in the obtained electromagnetic susceptibility characterization. In fact, the concentrated block distribution pattern does not affect the overall distribution of the output values ​​within the distribution range, and it is easier to divide the output values ​​into clusters in multidimensional clustering. Therefore, we further combine the concentrated block distribution characteristics of the output to obtain the clustering effect under the current dimension.

[0047] One-dimensional clustering is performed on all output values ​​in one dimension using the K-means clustering algorithm to obtain clusters, which are denoted as one-dimensional clusters. After one-dimensional clustering, the clusters are arranged according to the numerical value of the output values, thus obtaining a set of one-dimensional clusters. The larger the distribution interval between any two adjacent one-dimensional clusters, the more dispersed the distribution of the clusters, indicating a more obvious blocky distribution characteristic and a better clustering effect. On the other hand, the concentration of data distribution within each one-dimensional cluster is also obtained. The more concentrated the data distribution within each one-dimensional cluster, the more obvious the blocky distribution characteristic of the current one-dimensional cluster, the better the one-dimensional clustering effect of that dimension, and the more concentrated and similar the electromagnetic susceptibility within the cluster, the higher the reliability of the electromagnetic susceptibility within the cluster.

[0048] One-dimensional clusters corresponding to a given dimension are grouped into a set of one-dimensional clusters, which are arranged according to their output values. The absolute value of the difference between the maximum value in a given one-dimensional cluster and the minimum value in the next nearest one-dimensional cluster is recorded as the interval distance. The Euclidean distance between an output value and its nearest neighbor within a one-dimensional cluster is calculated and recorded as the minimum distance. The mean of the minimum distances of all output values ​​in the given one-dimensional cluster is then calculated and recorded as the mean minimum distance of the given one-dimensional cluster. A negative correlation mapping is performed using an exponential function with the natural constant as the base of the exponential function on the mean of the mean minimum distances of all one-dimensional clusters corresponding to that dimension to obtain the mapping result. This mapping result is multiplied by the mean of the interval distances between any two adjacent one-dimensional clusters in the set corresponding to that dimension to obtain the clustering effect for that dimension.

[0049] The specific calculation model for the clustering effect of one dimension is as follows:

[0050] ,

[0051] in, This represents the clustering effect along the q-th dimension. This indicates the number of one-dimensional clusters obtained after performing one-dimensional clustering in this dimension. Let be the distance between the i-th one-dimensional cluster and its adjacent next one-dimensional cluster, and let be the absolute value of the difference between the maximum output value in the i-th one-dimensional cluster and the maximum output value in its adjacent next one-dimensional cluster. This is the mean distance between any two adjacent one-dimensional clusters in the set of one-dimensional clusters corresponding to this dimension. The larger the mean distance, the more discrete the distribution of the one-dimensional clusters corresponding to this dimension is, the more obvious the block distribution characteristics of the current dimension are, and the better the cluster classification effect of the clustering result.

[0052] This represents the number of output values ​​in the i-th one-dimensional cluster. Let represent the Euclidean distance between the j-th output value in the i-th one-dimensional cluster and its nearest neighbor, which is also the minimum distance of the j-th output value. Let be the minimum distance mean of the i-th one-dimensional cluster. The smaller this mean, the more concentrated the output values ​​are within the one-dimensional cluster, and the more obvious the blocky distribution characteristics of the current dimension. is the mean of the minimum distances of all one-dimensional clusters. The smaller the mean, the stronger the concentration of intra-cluster distribution of all one-dimensional clusters in this dimension, and the better the clustering effect in this dimension. exp represents the exponential function with the natural constant as the base, which is used for inverse proportional normalization, that is, negative correlation mapping.

[0053] This allows us to obtain the clustering results for each dimension.

[0054] The clustering module is used to obtain the weight of a dimension based on the degree of electromagnetic susceptibility representation and the clustering effect; and to combine the weights of each dimension to cluster all data points to obtain different multidimensional clusters.

[0055] The above steps obtain the electromagnetic susceptibility characterization level and clustering effect for each dimension. Combining these results yields the weight for each dimension. Specifically, the weight for a dimension is obtained by averaging the normalized value of its electromagnetic susceptibility characterization level and the normalized value of its clustering effect. A stronger electromagnetic susceptibility characterization level results in a better clustering effect, making that dimension more suitable for obtaining a higher weight in multidimensional clustering. This leads to more accurate clustering results for classifying sensors with different electromagnetic susceptibility levels.

[0056] The specific formula for calculating the weight is as follows:

[0057] ,

[0058] in, This represents the weight of the q-th dimension. This indicates that the range of values ​​is controlled within [0,1] by linear normalization. Furthermore, by combining the weights of each dimension, all data points are clustered to obtain different multidimensional clusters. Specifically, when calculating the Euclidean distance between any two data points, the weights of different dimensions are multiplied by the calculated components of different dimensions to obtain the final Euclidean distance. This makes the obtained clustering results more accurate in classifying sensors with different electromagnetic susceptibility. For example, for two-dimensional data points with weights a and b in the two dimensions, and two data points (x1, y1) and (x2, y2), the Euclidean distance between these two data points is... This completes the multi-dimensional clustering of data points corresponding to each sensor, using the K-means clustering algorithm. This concludes the clustering of all sensors that participated in the testing throughout history.

[0059] The electromagnetic susceptibility calibration module is used to obtain the electromagnetic susceptibility of each multidimensional cluster based on the electromagnetic susceptibility of the sensor corresponding to each data point within each multidimensional cluster; and to determine the multidimensional cluster to which the data point corresponding to the current sensor belongs, wherein the electromagnetic susceptibility of the multidimensional cluster to which it belongs is the electromagnetic susceptibility of the current sensor.

[0060] The process involves obtaining clustering results for historically tested sensors. These clusters categorize sensors with different electromagnetic susceptibility characteristics into different types based on various multidimensional clusters. For newly tested sensors, output values ​​under multiple dimensions (multiple test conditions) are obtained. The newly tested sensors are then assigned to clusters based on the similarity between their output values ​​and those within the multidimensional clusters. The most recently tested sensor is designated as the current sensor. Additionally, the electromagnetic susceptibility of each multidimensional cluster is obtained. Specifically, the mode of the electromagnetic susceptibility of the sensors corresponding to each data point within a multidimensional cluster is used as the electromagnetic susceptibility of that cluster.

[0061] The algorithm obtains the multi-dimensional similarity between the output values ​​of the data points corresponding to the current sensor in multiple dimensions and the multi-dimensional similarity between the data points in each multi-dimensional cluster. The higher the similarity between the output values ​​of the data points corresponding to the current sensor in each dimension and the output values ​​in each dimension of the multi-dimensional cluster, the stronger the probability that the data points belong to that multi-dimensional cluster. On the other hand, the algorithm obtains the Euclidean distance between the data points corresponding to the current sensor and the cluster centers of each multi-dimensional cluster in the sample space. The closer the data points are to the cluster centers, the greater the probability that they belong to that multi-dimensional cluster.

[0062] Specifically, the reciprocal of the absolute value of the difference between the output value of a data point in one dimension within a multidimensional cluster and the output value of the corresponding data point in the same dimension is obtained as the similarity of that data point in that dimension. The average of the similarities of all data points in that dimension within the multidimensional cluster is calculated and denoted as the dimensional similarity. The sum of the dimensional similarities of all dimensions is calculated and denoted as the overall similarity. The reciprocal of the distance between the data point corresponding to the current sensor and the cluster center of the multidimensional cluster is normalized and added to the normalized value of the overall similarity to obtain the probability of belonging to the multidimensional cluster. The multidimensional cluster with the highest probability of belonging is the multidimensional cluster to which the data point corresponding to the current sensor belongs.

[0063] The specific calculation model for the probability of affiliation to a multidimensional cluster is as follows:

[0064] ,

[0065] in, This represents the probability of belonging to the z-th multidimensional cluster, which is the probability that the data point corresponding to the current sensor belongs to the z-th multidimensional cluster. For the number of dimensions, This represents the number of data points in the z-th multidimensional cluster; This represents the output value of the j-th data point in the i-th dimension. This represents the output value of the data point corresponding to the current sensor in the i-th dimension. This value represents the similarity between the j-th data point and the i-th dimension in the z-th multidimensional cluster. The larger the value, the more similar the output value of the current sensor's data point in the i-th dimension is to the output value of the j-th data point in the z-th multidimensional cluster in the i-th dimension. Let represent the dimensional similarity of the i-th dimension. The stronger the dimensional similarity, the more similar the output value of the data point corresponding to the current sensor in the i-th dimension is to the output value in the i-th dimension of the multidimensional cluster. Then, the dimensional similarity of each dimension is summed to obtain the overall similarity. The larger this value is, the greater the likelihood that the data point corresponding to the current sensor belongs to the multidimensional cluster.

[0066] This represents the distance between the data point corresponding to the current sensor and the cluster center of the z-th multidimensional cluster. This distance is the Euclidean distance. The smaller the distance, the greater the probability that the data point corresponding to the current sensor belongs to the z-th multidimensional cluster. norm represents the normalization operation.

[0067] The electromagnetic sensitivity of the multidimensional cluster to which the data point corresponding to the current sensor belongs is the electromagnetic sensitivity of the current sensor, thereby achieving highly efficient adaptive calibration of the electromagnetic sensitivity of automated industrial sensors; in addition, the clustering results of the multidimensional cluster can be updated in real time by setting the update interval.

[0068] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0069] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0070] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An industrial sensor electromagnetic sensitivity self-adaptive calibration system, characterized in that, The system includes: The data acquisition module is used to acquire the output values ​​of each sensor under different test conditions and treat each test condition as a dimension; it combines the output values ​​of a sensor under each dimension into a data point and obtains the electromagnetic sensitivity of each sensor; The electromagnetic susceptibility characterization degree calculation module is used to calculate the electromagnetic susceptibility characterization degree of a dimension based on the distribution of output values ​​of different data points in a dimension. The clustering effect calculation module is used to cluster the output values ​​under one dimension to obtain different one-dimensional clusters; and to obtain the clustering effect of that dimension based on the distribution of each one-dimensional cluster and the distribution of the output values ​​within each one-dimensional cluster. The clustering module is used to obtain the weight of a dimension based on the degree of electromagnetic susceptibility representation and the clustering effect. It combines the weight of each dimension to cluster all data points to obtain different multidimensional clusters. When calculating the Euclidean distance between any two data points, the calculated components of different dimensions are multiplied by the weights of different dimensions. The electromagnetic susceptibility calibration module is used to obtain the electromagnetic susceptibility of each multidimensional cluster based on the electromagnetic susceptibility of the sensors corresponding to each data point within each multidimensional cluster. The mode of the electromagnetic susceptibility of the sensors corresponding to each data point within a multidimensional cluster is taken as the electromagnetic susceptibility of that multidimensional cluster. The module also determines the multidimensional cluster to which the data point corresponding to the current sensor belongs, and the electromagnetic susceptibility of the multidimensional cluster to which it belongs is taken as the electromagnetic susceptibility of the current sensor. The calculation of the electromagnetic susceptibility characterization level of a dimension based on the distribution of output values ​​at different data points in a given dimension includes: Arrange the output values ​​of different data points in one dimension in ascending order to obtain the output value sequence corresponding to that dimension; obtain the absolute value of the difference between every two adjacent output values ​​in the output value sequence, and record it as the data interval between every two adjacent output values; calculate the difference between the maximum and minimum values ​​in the output value sequence and multiply it by the reciprocal of the standard deviation of the data interval between every two adjacent output values ​​to obtain the electromagnetic susceptibility characterization degree of that dimension. The step of obtaining the clustering effect of a dimension based on the distribution of each one-dimensional cluster and the distribution of output values ​​within each one-dimensional cluster includes: One-dimensional clusters corresponding to a given dimension are grouped into a set of one-dimensional clusters, which are arranged according to their output values. The absolute value of the difference between the maximum value in a given one-dimensional cluster and the minimum value in the next nearest one-dimensional cluster is recorded as the interval distance. Within a one-dimensional cluster, the Euclidean distance between an output value and its nearest neighbor is calculated, recorded as the minimum distance. The mean of the minimum distances of all output values ​​in the given one-dimensional cluster is then calculated, recorded as the mean minimum distance of the cluster. A negative correlation mapping is performed using an exponential function with the natural constant as the base of the exponential function on the mean of the mean minimum distances of all one-dimensional clusters corresponding to that dimension. This mapping result is then multiplied by the mean of the interval distances between any two adjacent one-dimensional clusters in the set corresponding to that dimension to obtain the clustering effect for that dimension. The process of obtaining the weight of a dimension based on the degree of electromagnetic susceptibility characterization and clustering effect includes: The weight of a dimension is obtained by averaging the normalized value of the electromagnetic susceptibility representation of a dimension with the normalized value of the clustering effect. The step of determining the multidimensional cluster to which the data point corresponding to the current sensor belongs includes: The reciprocal of the absolute value of the difference between the output value of a data point in one dimension within a multidimensional cluster and the output value of the corresponding data point in the same dimension of the current sensor is used as the similarity of that data point in that dimension. The average of the similarities of all data points in that dimension within the multidimensional cluster is calculated and denoted as the dimensional similarity. The sum of the dimensional similarities of all dimensions is denoted as the overall similarity. The reciprocal of the distance between the data point corresponding to the current sensor and the cluster center of the multidimensional cluster is normalized and added to the normalized value of the overall similarity to obtain the probability of belonging to that multidimensional cluster. The multidimensional cluster with the highest probability of belonging is the multidimensional cluster to which the data point corresponding to the current sensor belongs. Set an update interval to update the clustering results of multidimensional clusters in real time.

2. The industrial sensor electromagnetic sensitivity adaptive calibration system according to claim 1, characterized in that, The process of acquiring the electromagnetic sensitivity of each sensor includes: The electromagnetic sensitivity of each sensor is determined by manual evaluation based on its output value under each test condition.

Images