A dust concentration collection and early warning method and system

Abstract

The present invention relates to the field of data processing technology, and more specifically, to a dust concentration collection and early warning method and system, the method comprising: taking a binary tree corresponding to a data subset where any data point in dust concentration data is located as a reference binary tree, calculating the credibility of any data point in any reference binary tree, obtaining logarithmic data with the prominence of the data point in the reference binary tree as the base, the ratio of the balance factor of the reference binary tree to the time complexity being a true number, correcting the credibility, obtaining a corrected value, taking the corrected value of the credibility of the data point in each reference binary tree as a weight, performing weighted summation on the complexity changes of all reference binary trees before and after the data point is inserted, obtaining the final abnormality of the data point, and giving an early warning of the dust concentration in the air based on the final abnormality of the current data point. The present invention can improve the accuracy of abnormality detection.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and more specifically, to a dust concentration collection and early warning method and system. Background Art

[0002] With the rapid development of industrial production, especially in industries involving powder spraying, such as metal spraying and plastic spraying, the risk of dust explosion is increasing. When the suspended dust in the powder spraying room reaches a certain concentration and encounters a fire source, it is very easy to cause an explosion, causing casualties and property losses. The Robust Random Cut Forest (RRCF) algorithm is an anomaly detection algorithm for dynamic data volume. It can be used for anomaly detection of dust concentration data in the powder spraying room, thereby realizing early warning of dust concentration.

[0003] In the related technology, such as the patent application document with publication number CN112181706A, a method for detecting anomalies in power dispatching data based on logarithmic interval isolation is disclosed. The method includes: constructing multiple logarithmic interval isolation subtrees and forming them into a logarithmic interval isolation forest anomaly detector, calculating the path length of each sample under each subtree, integrating the paths of the corresponding data in each subtree and taking the average value to obtain the anomaly score, and filtering out abnormal samples in the data set according to the anomaly rate threshold.

[0004] However, when determining the anomaly score of the data, the above scheme does not take into account the differences in the degree of anomaly determined for the same data in different subtrees, and does not take into account the impact on the final anomaly score. It only takes the average value. If there is a situation where the degree of anomaly of any data in a subtree is large, while the degree of anomaly in the remaining subtrees is small, the determined anomaly score will be too large, and thus the data may not be accurately detected for anomalies, resulting in low accuracy of the determined anomaly detection results. Summary of the invention

[0005] In order to solve the problem of being unable to accurately detect anomalies in data, the present invention provides a dust concentration collection and early warning method and system.

[0006] According to a first aspect of an embodiment of the present invention, a dust concentration collection and early warning method is provided, comprising:

[0007] Obtain a dust concentration data set in the air, randomly select multiple data subsets from the dust concentration data set, and construct a binary tree of each data subset;

[0008] The binary tree corresponding to the data subset where any data point in the dust concentration data set is located is used as the reference binary tree, and the average value of the difference between the prominence of any data point in any reference binary tree and the prominence of any data point in the remaining reference binary trees is obtained, and the credibility of any data point in any reference binary tree is calculated, and the credibility is negatively correlated with the average value;

[0009] Obtain logarithmic data in which the ratio of the balance factor of any reference binary tree to the time complexity is a true number, taking the prominence of any data point in any reference binary tree as the base, and taking the sum of the normalized value of the logarithmic data and the credibility as the corrected value of the credibility;

[0010] The corrected value of the credibility of any data point in each reference binary tree is used as the weight, and the weighted sum of the complexity changes of all reference binary trees before and after the insertion of any data point is taken to obtain the final abnormality degree of any data point, so as to carry out early warning control of the dust concentration in the air based on the final abnormality degree of the current data point.

[0011] The present invention uses the correction value of the credibility of any data point in each reference binary tree as a weight, which can reduce the difference between the abnormality levels of the data point determined by each reference binary tree, and the impact on the final abnormality level of the data point, thereby ensuring the accuracy of the determined final abnormality level, thereby improving the accuracy of abnormality detection of dust concentration data, so that the dust concentration in the air can be accurately warned; the present invention comprehensively determines the corresponding credibility by taking the average value of the difference between the prominence levels of the same data in different reference binary trees, and can accurately reflect the difference of the same data in different reference binary trees, thereby ensuring the accuracy of the credibility, and based on the structure of the reference binary tree itself, the credibility is corrected, and the correction value is used as the weight, which can ensure the accuracy of the determined weight.

[0012] In an exemplary embodiment of the present invention, the credibility of any data point in any reference binary tree satisfies the following relationship:

[0013] ;

[0014] In the formula, For the The data point in Credibility in a reference binary tree; For the The data point in prominence in the reference binary tree; For the The data point in prominence in the reference binary tree; For the The maximum value of the prominence of a data point in all reference binary trees; For the The minimum prominence of a data point in all reference binary trees; is the number of reference binary trees; for the summation symbol; is the absolute value symbol.

[0015] In an exemplary embodiment of the present invention, a method for obtaining a protrusion degree includes:

[0016] The data subset corresponding to any reference binary tree is used as the reference subset, the reference subset is clustered, the cluster center of the reference subset is obtained, and the median of the distance between each data point in the reference subset and the cluster center is determined;

[0017] The distance between any data point and the cluster center is taken as the reference distance, and the ratio of the reference distance to the median is calculated to obtain the prominence of any data point in any reference binary tree.

[0018] The present invention uses the median as a measurement standard, which can reduce the influence of outliers, thereby ensuring the accuracy of the determined prominence degree.

[0019] In an exemplary embodiment of the present invention, the modified value of the credibility satisfies the following relationship:

[0020] ;

[0021] In the formula, For the The data point in The corrected value of the credibility in the reference binary tree; For the The data point in Credibility in a reference binary tree; For the The balance factor of a reference binary tree, For the The number of nodes in the reference binary tree; Notation for time complexity; For the The data point in prominence in the reference binary tree; is the sign of the logarithmic function; is the normalization function.

[0022] The calculation formula of the correction value of the credibility provided by the present invention can limit the value of the correction degree to 0-1, reduce the difficulty of calculating the correction value of the credibility, and to a certain extent ensure the accuracy of the determined correction value of the credibility.

[0023] In an exemplary embodiment of the present invention, the balancing factor is a comprehensive balancing factor, and a method for obtaining the comprehensive balancing factor includes:

[0024] Calculate the difference between the number of nodes in the left subtree and the number of nodes in the right subtree of any reference binary tree to obtain a first difference;

[0025] Calculate the difference between the variance of the data points corresponding to all nodes in the left subtree and the variance of the data points corresponding to all nodes in the right subtree to obtain a second difference;

[0026] A sum operation is performed on the first difference and the second difference, and the reciprocal of the result of the sum operation is used as the comprehensive balance factor of any parameter binary tree.

[0027] The present invention determines the comprehensive balance factor of the reference binary tree by integrating multiple data, thereby ensuring the accuracy of the determined comprehensive balance factor.

[0028] In an exemplary embodiment of the present invention, the final abnormality degree of any data point satisfies the following relationship:

[0029] ;

[0030] In the formula, For the The final abnormality level of the data point; For the The data point in The corrected credibility in the reference binary tree; For the The reference binary tree is inserted into The change in complexity before and after the data point; is the number of reference binary trees; is the summation symbol.

[0031] In an exemplary embodiment of the present invention, early warning control of dust concentration in the air based on the final abnormality degree of the current data point includes:

[0032] If the final abnormality level of the current data is greater than or equal to the preset threshold, an alarm will be issued to promptly remind the operator that the dust concentration in the air at the current moment is greater than the safe value.

[0033] The present invention can give an early warning of the dust concentration in the air in real time and accurately.

[0034] According to a second aspect of an embodiment of the present invention, a dust concentration collection and early warning system is provided. The dust concentration collection and early warning system includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the first aspect of the embodiment of the present invention.

[0035] The present invention has the following effects:

[0036] The present invention uses the correction value of the credibility of any data point in the binary tree where it is located as a weight, reduces the influence of the difference between the abnormality degrees of the same data point determined by different binary trees, improves the accuracy of abnormality detection on dust concentration data, and thus can reduce the false alarm rate of the early warning system; when determining the credibility of any data point in the binary tree where it is located, the difference between the prominence degrees of the data point in all data subsets corresponding to the binary tree where it is located is used, and the influence of data fluctuations on the data point can be accurately evaluated, thereby ensuring the accuracy of the credibility, and based on the structure of the binary tree itself and the logarithmic data determined by the prominence degree of the data point in the binary tree where it is located, the credibility can be corrected, and multiple data can be integrated to correct the credibility, thereby ensuring the accuracy of the corrected value of the credibility, thereby ensuring the accuracy of the determined weight. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0038] Figure 1 The present invention is a schematic diagram of the steps of a dust concentration collection and early warning method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0040] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0041] Reference Figure 1 , a dust concentration collection and early warning method, including steps S1 to S4, specifically as follows:

[0042] S1: Obtain a dust concentration data set in the air, randomly select multiple data subsets from the dust concentration data set, and construct a binary tree for each data subset.

[0043] In an exemplary embodiment of the present invention, a dust concentration data set refers to a collection of dust concentration data in the air collected within a certain period of time. For example, the dust concentration data set can be obtained by collecting readings from a dust concentration sensor within a certain period of time at a certain collection frequency such as 1 Hz.

[0044] Optionally, a dust concentration sensor based on the principle of light scattering can be used to collect dust concentration data during the powder spraying process. It should be noted that the dust concentration sensor based on this principle includes a dilution structure, which can dilute the dust-laden gas so that the concentration of the diluted dust-laden gas is within the measurement range of the sensor. The measurement data of the sensor is then read by the controller, and the dilution multiple of the dilution structure is amplified to obtain the dust concentration of the original dust-laden gas, thereby completing the collection of dust concentration data in the air.

[0045] Among them, multiple data subsets are randomly selected from the dust concentration data set, and binary trees of each data subset are constructed. This is a method of constructing binary trees using the Robust Random Cut Forest (RRCF) algorithm. After using the algorithm to construct multiple binary trees of the dust concentration data set, the algorithm can weight the detection results of any data point in the dust concentration data set in multiple binary trees, so as to determine whether the data point is abnormal based on the size of the weighted value. It should be noted that the process of constructing a binary tree using the algorithm is a prior art, and this embodiment will not be described in detail here.

[0046] S2: Take the binary tree corresponding to the data subset where any data point in the dust concentration data set is located as the reference binary tree, obtain the average value of the difference between the prominence of any data point in any reference binary tree and the prominence in the remaining reference binary trees, and calculate the credibility of any data point in any reference binary tree. The credibility is negatively correlated with the average value.

[0047] In an exemplary embodiment of the present invention, prominence refers to data used to reflect the degree to which any data point is affected by data fluctuations. For example, when the prominence of any data point in any reference binary tree is greater, it means that the data point is more affected by data fluctuations. Conversely, when the prominence of any data point in any reference binary tree is smaller, it means that the data point is less affected by data fluctuations.

[0048] Optionally, when determining the prominence of any data point in any reference binary tree, the data subset for constructing the reference binary tree can be determined first, and then the prominence of the data point in the data subset can be calculated, thereby using the prominence of the data point in the data subset as the prominence of the data point in the reference binary tree.

[0049] Credibility refers to the data used to measure the accuracy of the degree of anomaly of any data point determined by each reference binary tree when using the RRCF algorithm to detect anomalies of any data point. For example, when the credibility of any data point in any reference binary tree is large, it means that the accuracy of the degree of anomaly of the data point determined by the reference binary tree is high. Therefore, when weighting the degree of anomaly of the data point determined by each reference binary tree, the weight of the degree of anomaly determined by the corresponding reference binary tree with higher credibility can be increased, so as to accurately detect anomalies of the data point.

[0050] In an exemplary embodiment of the present invention, the determination of the prominence of any data point in any reference binary tree can be achieved by the following steps:

[0051] (1) Take the data subset corresponding to any reference binary tree as the reference subset, cluster the reference subset, obtain the cluster center of the reference subset, and determine the median of the distance between each data point in the reference subset and the cluster center;

[0052] Optionally, k-means clustering can be performed on each data point in the reference subset, and the k value is set to 1, so that the cluster center of the reference subset can be obtained, and then the distances between each data point in the reference subset and the cluster center are arranged in ascending order, so that the median of the distances between each data point in the reference subset and the cluster center can be determined.

[0053] Optionally, other clustering algorithms may be used to cluster the data points in the reference subset, such as the DBSCAN clustering algorithm, the Bisecting K-Means clustering algorithm, and the like.

[0054] (2) Take the distance between any data point and the cluster center as the reference distance, calculate the ratio of the reference distance to the median, and obtain the prominence of any data point in any reference binary tree.

[0055] Specifically, the prominence of any data point in any reference binary tree satisfies the following relationship:

[0056]

[0057] In the formula, For the The data point in prominence in the reference binary tree; For the The cluster center of the reference subset corresponding to the reference binary tree; is the first data points; is a way of expressing distance. It is a representation of the median distance between each data point in the reference subset and the cluster center.

[0058] Furthermore, after determining the prominence of any data point in all reference binary trees, the credibility of the data point in any reference binary tree can be calculated. Specifically, the credibility of any data point in any reference binary tree satisfies the following relationship:

[0059] ;

[0060] In the formula, For the The data point in Credibility in a reference binary tree; For the The data point in prominence in the reference binary tree; For the The data point in prominence in the reference binary tree; For the The maximum value of the prominence of a data point in all reference binary trees; For the The minimum value of the prominence of a data point in all reference binary trees. It should be noted that It is the role of maximum and minimum normalization; is the number of reference binary trees; for the summation symbol; is the absolute value symbol.

[0061] in, Reflects the The data point in The average of the differences between the prominence in the first reference binary tree and the prominence in the remaining reference binary trees. The larger the value, the more likely the data point is in the The probability that the prominence degree in the reference binary tree is abnormal is greater, and the data point is more affected by data fluctuations. The lower the credibility in the reference binary tree.

[0062] In another embodiment, the credibility of any data point in any reference binary tree may also be determined by other calculation formulas, such as by taking the opposite number instead of the proportional relationship calculation formula.

[0063] S3: Obtain logarithmic data with the prominence of any data point in any reference binary tree as the base and the ratio of the balance factor of any reference binary tree to the time complexity as a true number, and take the sum of the normalized value of the logarithmic data and the credibility as the correction value of the credibility.

[0064] It should be noted that the structure of the binary tree itself will also affect the results of anomaly detection. Therefore, the balance factor and time complexity that can reflect the structure of the binary tree itself can be used to evaluate the construction effect of the corresponding binary tree.

[0065] Among them, the time complexity of any reference binary tree can be expressed as , For the The number of nodes in a reference binary tree. It should be noted that the structure of the binary tree built based on the RRCF algorithm is similar to a balanced tree. The time complexity of each operation on the binary tree is related to the height of the tree. The height of the tree is kept at a logarithmic level during random segmentation and updating. Therefore, based on this feature, the time complexity of any reference binary tree can be determined.

[0066] In an exemplary embodiment of the present invention, the balance factor is a comprehensive balance factor, and the comprehensive balance factor of any reference binary tree can be determined by the following steps:

[0067] The difference between the number of nodes in the left subtree and the number of nodes in the right subtree of any reference binary tree is calculated to obtain a first difference; the difference between the variance of the data points corresponding to all the nodes in the left subtree and the variance of the data points corresponding to all the nodes in the right subtree is calculated to obtain a second difference; the first difference and the second difference are summed, and the inverse of the result of the summation operation is used as the comprehensive balance factor of any parameter binary tree.

[0068] Specifically, the comprehensive balance factor of any reference binary tree satisfies the following relationship:

[0069] ;

[0070] In the formula, For the The comprehensive balance factor of a reference binary tree; For the The number of nodes in the left subtree of the reference binary tree; For the The number of nodes in the right subtree of the reference binary tree; For the The variance of the data points corresponding to all nodes in the left subtree of the reference binary tree; For the The variance of the data points corresponding to all nodes in the right subtree of the reference binary tree; is the absolute value symbol.

[0071] in, is the first difference. The larger the value, the The greater the difference in the number of nodes in the left and right subtrees of a reference binary tree, the worse the overall balance factor of the corresponding reference binary tree; is the second difference. The larger the value, the The greater the difference between the data points corresponding to all nodes in the left and right subtrees of a reference binary tree, the worse the overall balance factor of the corresponding reference binary tree.

[0072] Furthermore, after determining the time complexity and comprehensive balance factor of any reference binary tree, the ratio of the time complexity to the comprehensive balance factor of any reference binary tree can be calculated, and the construction effect of the reference binary tree can be evaluated using the ratio.

[0073] In another embodiment, the height of the left subtree in any reference binary tree minus the height of the right subtree may be directly calculated to obtain the traditional balance factor of the reference binary tree, thereby evaluating the construction effect of the reference binary tree based on the ratio of the time complexity of the reference binary tree to the traditional balance factor.

[0074] Furthermore, since the prominence of any data point in any reference binary tree will affect the contribution of the data point to the reference binary tree, the credibility of the data point in the reference binary tree can be corrected based on the prominence of any data point in any reference binary tree and the construction effect of the reference binary tree, thereby obtaining a corrected value of the credibility.

[0075] Specifically, the corrected value of the credibility of any data point in any reference binary tree satisfies the following relationship:

[0076] ;

[0077] In the formula, For the The data point in The corrected value of the credibility in the reference binary tree; For the The data point in Credibility in a reference binary tree; For the The balance factor of a reference binary tree, For the The number of nodes in the reference binary tree; Notation for time complexity; For the The data point in prominence in the reference binary tree; is the sign of the logarithmic function; is the normalization function.

[0078] in, Reflects the The construction effect of the reference binary tree. The larger the value, the better the construction effect of the binary tree. The lower the correction amplitude of the credibility of the data point in the reference binary tree.

[0079] S4: Take the corrected value of the credibility of any data point in each reference binary tree as the weight, and perform weighted summation on the complexity changes of all reference binary trees before and after inserting any data point to obtain the final abnormality degree of any data point, so as to perform early warning control on the dust concentration in the air based on the final abnormality degree of the current data point.

[0080] Specifically, the final abnormality degree of any data point satisfies the following relationship:

[0081] ;

[0082] In the formula, For the The final abnormality level of the data point; For the The data point in The corrected credibility in the reference binary tree; For the The reference binary tree is inserted into The change in complexity before and after the data point; is the number of reference binary trees; is the summation symbol.

[0083] The complexity change refers to the difference between the sum of the depths of the nodes in any reference binary tree before inserting a new data point in the reference binary tree and the sum of the depths of the nodes in the reference binary tree after inserting the new data point. It should be noted that the process of calculating the depth of each node is a prior art and will not be described in detail in this embodiment.

[0084] In an exemplary embodiment of the present invention, early warning control of dust concentration in the air can be achieved by the following steps:

[0085] If the final abnormality level of the current data is greater than or equal to the preset threshold, an alarm will be issued to promptly remind the operator that the dust concentration in the air at the current moment is greater than the safe value.

[0086] Optionally, the preset threshold can be set to 0.8. If the final abnormality degree of the current data point is greater than or equal to 0.8, the early warning mechanism is triggered and an alarm is issued, thereby promptly reminding the operator that the dust concentration in the air at the current moment is too high, so that the operator can stop the operation of the powder sprayer in time to ensure safety in the production process. This embodiment does not impose any special restrictions on the setting of the threshold, as long as it is within the commonly used setting range in the field of industrial monitoring.

[0087] The present invention also provides a dust concentration collection and early warning system, the system includes a memory and a processor, and a computer program is stored in the memory, the computer program integrates the function of a dust concentration collection and early warning method, when the computer program is executed, the accuracy of abnormality detection can be improved through a dust concentration collection and early warning method.

[0088] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0089] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A dust concentration collection and early warning method, characterized in that: include: Obtaining a dust concentration data set in the air, randomly selecting multiple data subsets from the dust concentration data set, and constructing a binary tree of each of the data subsets; Taking a binary tree corresponding to a data subset where any data point in the dust concentration data set is located as a reference binary tree, obtaining the prominence of any data point in any reference binary tree, including: taking a data subset corresponding to any reference binary tree as a reference subset, clustering the reference subset, obtaining the cluster center of the reference subset, and determining the median of the distance between each data point in the reference subset and the cluster center; taking the distance between any data point and the cluster center as a reference distance, calculating the ratio of the reference distance to the median, and obtaining the prominence of any data point in any reference binary tree; calculating the credibility of any data point in any reference binary tree: ; For the The data point in Credibility in a reference binary tree; For the The data point in prominence in the reference binary tree; For the The data point in prominence in the reference binary tree; For the The maximum value of the prominence of a data point in all reference binary trees; For the The minimum prominence of a data point in all reference binary trees; is the number of reference binary trees; is the absolute value symbol; Calculate the corrected value of the credibility: ; For the The data point in The corrected value of the credibility in the reference binary tree; For the The data point in Credibility in a reference binary tree; For the The balance factor of a reference binary tree, For the The number of nodes in the reference binary tree; Symbolize time complexity; is the sign of the logarithmic function; is the normalization function; The corrected value of the credibility of any data point in each reference binary tree is used as a weight, and the weighted sum of the complexity changes of all reference binary trees before and after the insertion of any data point is performed to obtain the final abnormality degree of any data point, so as to perform early warning control on the dust concentration in the air based on the final abnormality degree of the current data point.

2. A dust concentration collection and early warning method according to claim 1, characterized in that: The balance factor is a comprehensive balance factor, and a method for obtaining the comprehensive balance factor includes: Calculate the difference between the number of nodes in the left subtree and the number of nodes in the right subtree of any reference binary tree to obtain a first difference; Calculate the difference between the variance of the data points corresponding to all the nodes in the left subtree and the variance of the data points corresponding to all the nodes in the right subtree to obtain a second difference; A sum operation is performed on the first difference and the second difference, and the inverse of the result of the sum operation is used as the comprehensive balance factor of any reference binary tree.

3. A dust concentration collection and early warning method according to claim 1, characterized in that: The final abnormality degree of any data point satisfies the following relationship: ； In the formula, For the The final abnormality level of the data point; For the The data point in The corrected credibility in the reference binary tree; For the The reference binary tree is inserted into The change in complexity before and after the data point; is the number of reference binary trees; is the summation symbol.

4. A dust concentration collection and early warning method according to claim 1, characterized in that: The early warning control of dust concentration in the air based on the final abnormality degree of the current data point includes: If the final abnormality level of the current data is greater than or equal to the preset threshold, an alarm will be issued to promptly remind the operator that the dust concentration in the air at the current moment is greater than the safe value.

5. A dust concentration collection and early warning system, characterized in that: The dust concentration collection and early warning system comprises a memory and a processor, wherein a computer program is stored in the memory, and the processor executes the computer program to implement the steps of the method according to any one of claims 1 to 4.

Images