A method for optimizing the collection and analysis of gas cabinet sensor data

Abstract

The present application relates to the technical field of digital data processing, and proposes a method for optimizing the collection and analysis of gas cabinet sensor data, including: confirming dimensional data corresponding to at least two dimensions based on the time series data corresponding to the odor data; confirming the importance corresponding to the important data segments of each dimensional data based on the amplitude and slope of each dimensional data, clustering the importance corresponding to the important data segments, confirming the discreteness of the important data segments of each dimensional data in the corresponding clustering cluster, confirming the convergence degree corresponding to the important data segments of each dimensional data based on the importance and discreteness corresponding to the important data segments of each dimensional data, and performing dimensionality reduction processing on the time series data corresponding to the odor data based on the convergence degree and a preset dimensionality reduction algorithm to confirm the odor type corresponding to the odor data. Constraining the odor data by the convergence degree can improve the dimensionality reduction effect, thereby improving the accuracy of judging the odor type.

Description

Technical Field

[0001] The present application relates to the technical field of digital data processing, and in particular to a method for optimizing the collection and analysis of gas cabinet sensor data. Background Art

[0002] In modern industrial and environmental monitoring, gas cabinet sensors are used to monitor gas concentration, temperature, humidity and other parameters in real time, and are widely used in petrochemical, environmental protection, smart home and other fields.

[0003] The data obtained when the optical gas sensor detects odors is high-dimensional data, containing multiple types of odor data, and each type of odor data contains multiple dimensions of data. Therefore, it is necessary to reduce the dimensionality of the obtained data to analyze the odor types contained in the cabinet based on the obtained odor data. Because the data obtained when the optical gas sensor detects odors is too high in dimension, the subsequent data dimensionality reduction effect is poor, which in turn affects the accuracy of odor type judgment. Summary of the invention

[0004] In view of the above, it is necessary to provide a method for optimizing the collection and analysis of gas cabinet sensor data, which improves the accuracy of odor type judgment and further reduces the working cost compared with the traditional method of judging the odor type of the gas cabinet.

[0005] The first aspect of the present application provides a method for optimizing the collection and analysis of gas cabinet sensor data, which is applied to the field of odor data analysis. The method comprises:

[0006] Collecting time series data corresponding to the odor data in the gas cabinet to confirm dimensional data corresponding to at least two dimensions according to the time series data;

[0007] Based on the amplitude and slope of each dimensional data, determine the importance of the important data segment of each dimensional data, wherein the important data segment refers to the data segment composed of data points on both sides of the amplitude maximum point corresponding to each dimensional data and greater than the average amplitude;

[0008] Clustering the importance of the important data segments of at least two dimensional data to determine the degree of dispersion of the important data segments of each dimensional data in the corresponding clustering cluster;

[0009] According to the importance and discreteness corresponding to the important data segments of each dimensional data, the convergence degree corresponding to the important data segments of each dimensional data is determined;

[0010] Based on the convergence degree corresponding to the important data segments of each dimensional data and the preset dimensionality reduction algorithm, the time series data corresponding to the odor data is subjected to dimensionality reduction processing to confirm the odor type corresponding to the odor data.

[0011] In one embodiment, based on the amplitude and slope of each dimensional data, the importance of the important data segment of each dimensional data is confirmed, wherein the important data segment refers to a data segment composed of data points on both sides of the amplitude maximum point corresponding to each dimensional data and greater than the average amplitude, specifically including:

[0012] Count the amplitude of each data point in each dimension of data and calculate the average amplitude of each dimension of data;

[0013] Based on the amplitude of the maximum point of each dimensional data, the average amplitude and the slope corresponding to each data point in the important data segment, the importance of the important data segment of each dimensional data is calculated.

[0014] In one embodiment, the importance of the important data segment of each dimensional data is calculated based on the amplitude of the maximum point of each dimensional data, the average amplitude and the slope corresponding to each data point in the important data segment, and specifically includes:

[0015]

[0016] in, For the Dimensional data The importance of the important data segments corresponding to the maximum value points, For the The first dimension of data The amplitude of the maximum point, For the The average amplitude of the dimensional data, For the The number of data points in the important data segment corresponding to the maximum value point, For the The important data segment corresponding to the maximum value point The slope corresponding to the data point.

[0017] In one embodiment, clustering the importance levels of the important data segments of at least two dimensional data to determine the degree of dispersion of the important data segments of each dimensional data in the corresponding clustering clusters specifically includes:

[0018] Clustering the importance levels corresponding to the important data segments of at least two dimensional data to determine a preset number of clusters;

[0019] According to the cluster clusters, counting the number and average amplitude of important data segments in the cluster clusters corresponding to the target important data segments, and the cluster distance between the target important data segments and the cluster center of the corresponding cluster clusters;

[0020] The number and average amplitude of important data segments in the cluster corresponding to the target important data segment, as well as the cluster distance between the target important data segment and the cluster center of the corresponding cluster are input into the discrete degree calculation formula to calculate the discrete degree of the target important data segment in the corresponding cluster cluster.

[0021] In one embodiment, the number and average amplitude of the important data segments in the cluster corresponding to the target important data segment, and the cluster distance between the target important data segment and the cluster center of the corresponding cluster are input into the discrete degree calculation formula to calculate the discrete degree of the target important data segment in the corresponding cluster, which specifically includes:

[0022]

[0023] in, For the The important data segments of the target in the clusters The degree of discreteness, For the The important data segments of the target in the clusters The importance of For the The number of important data segments in the clusters, Target important data segment Middle The amplitude of the data point, For the The average amplitude of important data segments in clusters, Target important data segment The number of data points, Target important data segment With The cluster distance of the cluster centers of the clusters, For is the base exponential function.

[0024] In one embodiment, determining the convergence degree corresponding to the important data segment of each dimensional data according to the importance degree and the discrete degree corresponding to the important data segment of each dimensional data specifically includes:

[0025] Count the sum of the amplitudes of the data points in the target important data segment and the sum of the amplitudes of the data points in the dimensional data corresponding to the target important data segment;

[0026] Calculate the constraint weight corresponding to the target important data segment based on the importance and discreteness corresponding to the target important data segment, the sum of the amplitudes of the data points in the target important data segment, and the sum of the amplitudes of the data points in the dimensional data corresponding to the target important data segment;

[0027] The constraint weight corresponding to the target important data segment and the amplitude mean of the dimensional data corresponding to the target important data segment are input into the convergence degree calculation formula to calculate the convergence degree corresponding to the target important data segment.

[0028] In one embodiment, the calculation of the constraint weight corresponding to the target important data segment based on the importance level, discreteness level, the sum of amplitudes of data points in the target important data segment, and the sum of amplitudes of data points in the dimensional data corresponding to the target important data segment specifically includes:

[0029]

[0030] in, For the The important data segments of the target in the clusters The constraint weight, For the The important data segments of the target in the clusters The degree of discreteness, For the The important data segments of the target in the clusters The importance of For the Target important data segments of dimensional data No. The amplitude of the data point, Target important data segment The number of data points, Target important data segment The corresponding Dimensional data The amplitude of the data point, Target important data segment The corresponding The number of data points for each dimension.

[0031] In one embodiment, the step of inputting the constraint weight corresponding to the target important data segment and the amplitude mean of the dimensional data corresponding to the target important data segment into a convergence degree calculation formula to calculate the convergence degree corresponding to the target important data segment specifically includes:

[0032]

[0033] in, For the The important data segments of the target in the clusters The degree of convergence, For the The important data segments of the target in the clusters The constraint weight, Target important data segment The corresponding The mean amplitude of the data in each dimension.

[0034] In one embodiment, based on the convergence degree corresponding to the important data segments of each dimensional data and the preset dimensionality reduction algorithm, the time series data corresponding to the odor data is subjected to dimensionality reduction processing to confirm the odor type corresponding to the odor data, specifically including:

[0035] Based on the convergence degree corresponding to the important data segments of each dimensional data, the time series data is converged to confirm the final time series data;

[0036] Performing dimensionality reduction processing on the final time series data by using a preset dimensionality reduction algorithm to confirm the dimensionality reduction data;

[0037] The odor type corresponding to the odor data is confirmed based on the dimension reduction data.

[0038] In one embodiment, the collecting of time series data corresponding to the odor data in the gas cabinet to confirm dimensional data corresponding to at least two dimensions according to the time series data specifically includes:

[0039] Collecting time series data corresponding to the odor data in the gas cabinet, denoising the time series data according to a preset denoising algorithm, and confirming the final time series data;

[0040] According to the final time series data, dimensional data corresponding to at least two dimensions are confirmed.

[0041] The embodiment of the present application first collects time series data corresponding to the odor data in the gas cabinet to confirm the dimensional data corresponding to at least two dimensions according to the time series data, and then confirms the importance corresponding to the important data segments of each dimensional data based on the amplitude and slope of each dimensional data, wherein the important data segment refers to the data segment composed of data points on both sides of the amplitude maximum point corresponding to each dimensional data and greater than the average amplitude, and then clusters the importance corresponding to the important data segments of at least two dimensional data to confirm the degree of discreteness of the important data segments of each dimensional data in the corresponding clustering cluster, and further confirms the degree of convergence corresponding to the important data segments of each dimensional data based on the importance and discreteness corresponding to the important data segments of each dimensional data, and finally, based on the convergence degree corresponding to the important data segments of each dimensional data and the preset dimensionality reduction algorithm, performs dimensionality reduction processing on the time series data corresponding to the odor data to reduce the influence of irrelevant data, so as to avoid the problem that the data dimension obtained by the traditional optical gas sensor when detecting gas is too high, resulting in poor subsequent data dimensionality reduction effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0043] Figure 1 It is a flow chart of a method for optimizing the collection and analysis of gas cabinet sensor data according to an embodiment of the present application;

[0044] Figure 2 This is a schematic diagram of a first sub-process of a method for optimizing the collection and analysis of gas cabinet sensor data according to an embodiment of the present application;

[0045] Figure 3 This is a schematic diagram of a second sub-process of a method for optimizing the collection and analysis of gas cabinet sensor data according to an embodiment of the present application;

[0046] Figure 4 This is a schematic diagram of a third sub-process of a method for optimizing the collection and analysis of gas cabinet sensor data according to an embodiment of the present application;

[0047] Figure 5 This is a schematic diagram of a fourth sub-process of a method for optimizing the collection and analysis of gas cabinet sensor data according to an embodiment of the present application;

[0048] Figure 6 This is a schematic diagram of the fifth sub-process of a method for optimizing the collection and analysis of gas cabinet sensor data according to an embodiment of the present application. DETAILED DESCRIPTION

[0049] In the description of the embodiments of the present application, words such as "exemplary", "or", "for example" and the like are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "or", "for example" and the like is intended to present related concepts in a concrete manner.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field in this application. The terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. It should be understood that, unless otherwise specified in this application, " / " means or. For example, A / B can represent A or B. "And / or" in this application is only a description of the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. "At least one" means one or more. "Multiple" means two or more than two. For example, at least one of a, b or c can represent: a, b, c, a and b, a and c, b and c, a, b and c seven situations.

[0051] It should also be noted that the terms "first" and "second" in the specification and drawings of the present application are used to distinguish similar objects, rather than to describe a specific order or sequence. The method disclosed in the embodiments of the present application or the method shown in the flowchart includes one or more steps for implementing the method. Without departing from the scope of the specification, the execution order of multiple steps can be interchanged with each other, and some steps can also be deleted.

[0052] See also Figure 1 , which shows a flow chart of a method for optimizing the collection and analysis of gas cabinet sensor data provided by an embodiment of the present application, the method comprising the following steps:

[0053] Step S101, collecting time series data corresponding to the odor data in the gas cabinet, and confirming dimensional data corresponding to at least two dimensions according to the time series data.

[0054] The time series data corresponding to the odor data in the gas cabinet refers to a data sequence obtained by an optical gas sensor installed in the gas cabinet and arranged in chronological order, wherein each data point is associated with a specific time point or time period. Time series data is usually an observation value or measurement result collected within a continuous time range. The dimensional data corresponding to at least two dimensions confirmed according to the time series data refers to expressing multiple data meanings in different dimensions based on the specific relationship between data points in the time series data.

[0055] It should be noted that data dimension is one of the important indicators to measure the complexity and information content of a data set. The higher the dimension, the higher the complexity of the data set, and more features are needed to describe each sample. The size of the data dimension has an important impact on the performance and effect of data analysis and machine learning algorithms. High-dimensional data (high dimension) usually faces the following challenges: 1. Dimensional disaster: As the data dimension increases, the sample space grows exponentially, resulting in increased data sparsity and greater distances between samples, which poses challenges to the calculation and storage of the algorithm. 2. Feature selection and dimensionality reduction: There may be redundant or irrelevant features in high-dimensional data, which will affect the performance of the algorithm. Therefore, feature selection or dimensionality reduction is required to reduce the number of features and improve the efficiency and accuracy of the algorithm. 3. Visualization difficulties: High-dimensional data cannot be directly visualized because we can usually only observe data in two-dimensional or three-dimensional space. Therefore, dimensionality reduction techniques are needed to map high-dimensional data to low-dimensional space for easy visualization and understanding. Therefore. In data analysis, choosing the right data dimension is an important issue. Too high or too low a dimension may lead to information loss or redundancy, so it is necessary to determine the appropriate data dimension based on the specific problem and the characteristics of the data set. This case is a technical solution to adjust the high-dimensional odor data in the dimensionality reduction process, which leads to poor dimensionality reduction effect due to the high dimensionality, so as to improve the accuracy of odor type judgment.

[0056] For details, please refer to the attached Figure 2 , the time series data corresponding to the odor data in the gas cabinet is collected to confirm the dimensional data corresponding to at least two dimensions according to the time series data, specifically including:

[0057] S201, collecting time series data corresponding to the odor data in the gas cabinet, denoising the time series data according to a preset denoising algorithm, and confirming the final time series data;

[0058] S202. Confirm dimensional data corresponding to at least two dimensions according to the final time series data.

[0059] Among them, after obtaining the time series data corresponding to the odor data collected in the gas cabinet, denoising processing is first performed through a preset denoising algorithm to obtain final time series data, and then dimensional analysis is performed on the final time series data to confirm dimensional data corresponding to at least two dimensions.

[0060] It should be noted that the preset denoising algorithm may be a median filtering algorithm, which is a commonly used nonlinear filtering algorithm for removing noise from images or signals. Its principle is to replace the value of the current pixel with the median in the local area, thereby smoothing the image and removing noise. The advantage of the median filtering algorithm is that it can effectively remove outliers such as salt and pepper noise and impulse noise without blurring the details of the data. The specific denoising process is described in the prior art and is not further limited in this case.

[0061] S102. Based on the amplitude and slope of each dimensional data, confirm the importance corresponding to the important data segments of each dimensional data, wherein the important data segments refer to data segments composed of data points located on both sides of the amplitude maximum point corresponding to each dimensional data and greater than the average amplitude.

[0062] Among them, the amplitude of the dimensional data refers to the fluctuation or amplitude of the data point of the dimensional data, which is used to describe the strength or size of the data. The slope of each dimensional data refers to the rate of change or speed of change at a certain data point in the dimensional data. The slope can be used to describe the change trend and speed of the data. The important data segment refers to the data segment composed of data points located on both sides of the amplitude maximum point corresponding to each dimensional data and greater than the average amplitude. It should be noted that when each dimensional data includes multiple maximum points, each dimensional data includes multiple important data segments, and the important data segment can be a data segment or a data point (that is, when the maximum point is equal to the average amplitude of the current dimensional data). The amplitude maximum point corresponding to each dimensional data can be obtained according to the existing technical means, and this case further defines this. It should be noted that the importance of data refers to the degree of influence of data on problems or decisions, which can be evaluated through target relevance, credibility and accuracy, variability and volatility, scope of influence, amount of information and insight. The importance of the important data segment of each dimensional data refers to the influence of the current important data segment on the subsequent convergence degree calculation.

[0063] S103: Cluster the importance levels corresponding to the important data segments of at least two dimensional data, and confirm the degree of dispersion of the important data segments of each dimensional data in the corresponding clustering cluster.

[0064] Among them, after obtaining the importance corresponding to the important data segments of the at least two dimensional data, a cluster analysis calculation is performed on them to obtain a preset number of cluster clusters, and the cluster clusters may include the importance corresponding to the important data segments of multiple dimensional data. The cluster cluster is a set formed by grouping the importance corresponding to the important data segments according to the similarity of the importance corresponding to the important data segments in the cluster analysis. It needs to be explained that in the cluster analysis, the importance corresponding to the important data segments is assigned to different clusters, and each cluster represents the importance corresponding to a group of similar important data segments. The similarity between the importance corresponding to the important data segments within the cluster is high, while the similarity between the importance corresponding to the important data segments between different clusters is low. The formation of cluster clusters is achieved by calculating the distance or similarity between the importance corresponding to the important data segments. The discrete degree of the important data segments of each dimensional data in the corresponding cluster cluster refers to the degree of distribution or change of the important data segments of each dimensional data in the corresponding cluster cluster, which is used to measure the size of the interval or difference between the important data segments of each dimensional data.

[0065] S104. Confirm the convergence degree corresponding to the important data segment of each dimensional data according to the importance degree and the discrete degree corresponding to the important data segment of each dimensional data.

[0066] The convergence degree corresponding to the important data segment of each dimensional data refers to the degree to which the important data segment of each dimensional data needs to reach a predetermined target before dimensionality reduction. In other words, the important data segment of each dimensional data needs to converge to the corresponding convergence degree to avoid high-dimensional problems. After obtaining the importance degree corresponding to the important data segment of each dimensional data and the degree of dispersion of the importance degree corresponding to the important data segment in the corresponding cluster cluster, the convergence degree corresponding to the important data segment is calculated.

[0067] S105. Based on the convergence degree corresponding to the important data segments of each dimensional data and a preset dimensionality reduction algorithm, the time series data corresponding to the odor data is subjected to dimensionality reduction processing to confirm the odor type corresponding to the odor data.

[0068] Among them, after obtaining the convergence degree corresponding to the important data segments of each dimensional data, the important data segments of each dimensional data are converged to obtain a new time series data, and then the new time series data is subjected to dimensionality reduction processing through a preset dimensionality reduction algorithm to obtain the data after dimensionality reduction, and finally the odor type corresponding to the odor data is confirmed based on the data after dimensionality reduction.

[0069] For details, please refer to the attached Figure 3, based on the convergence degree corresponding to the important data segments of each dimensional data and the preset dimensionality reduction algorithm, the time series data corresponding to the odor data is subjected to dimensionality reduction processing to confirm the odor type corresponding to the odor data, specifically including:

[0070] S301, based on the convergence degree corresponding to the important data segments of each dimensional data, performing convergence processing on the time series data to confirm the final time series data;

[0071] S302, performing dimensionality reduction processing on the final time series data by using a preset dimensionality reduction algorithm to confirm the dimensionality reduction data;

[0072] S303: confirming the odor type corresponding to the odor data according to the dimension reduction data.

[0073] Among them, the final time series data refers to the time series data after convergence processing based on the convergence degree corresponding to the important data segment, and the dimensionality reduction algorithm can be a linear discriminant analysis algorithm. After obtaining the final time series data, the final time series data is subjected to dimensionality reduction processing by the linear discriminant analysis algorithm to obtain dimensionality reduction data, so as to confirm the odor type corresponding to the odor data based on the dimensionality reduction data.

[0074] It should be noted that the general principle of Linear Discriminant Analysis (LDA) algorithm for dimensionality reduction is as follows:

[0075] 1. Calculate the intra-class scatter matrix Sw: For each class, calculate the scatter of samples within the class. The specific method is to calculate the covariance matrix of samples within each class, and then weighted sum the covariance matrices of all classes.

[0076] 2. Calculate the inter-class divergence matrix Sb: For each class, calculate the difference between this class and other classes. This is done by calculating the mean vector for each class and then calculating the difference between the mean vectors.

[0077] 3. Solve the generalized eigenvalue problem: Find the best projection direction by solving the eigenvalue and eigenvector of Sw^(-1)*Sb. The larger the eigenvalue corresponding to the eigenvector, the greater the contribution of the eigenvector to the classification.

[0078] 4. Select the most important eigenvectors: According to the size of the eigenvalue, select the first k most important eigenvectors to form the projection matrix W. The larger the eigenvalues ​​corresponding to these eigenvectors, the more category information they contain.

[0079] 5. Dimensionality reduction: Multiply the original data set X by the projection matrix W to obtain the reduced-dimensional data set Y. Y = X * W.

[0080] Through the above steps, the LDA algorithm can map high-dimensional data into a low-dimensional space while retaining the most important category information. This can achieve the purpose of dimensionality reduction, reduce the dimension of features, and achieve better classification results in the space after dimensionality reduction.

[0081] The embodiment of the present application first collects time series data corresponding to the odor data in the gas cabinet to confirm the dimensional data corresponding to at least two dimensions based on the time series data, and then confirms the importance corresponding to the important data segments of each dimensional data based on the amplitude and slope of each dimensional data, wherein the important data segment refers to the data segment composed of data points on both sides of the amplitude maximum point corresponding to each dimensional data and greater than the average amplitude, and then clusters the importance corresponding to the important data segments of at least two dimensional data to confirm the degree of discreteness of the important data segments of each dimensional data in the corresponding clustering cluster, and further confirms the degree of convergence corresponding to the important data segments of each dimensional data based on the importance and discreteness corresponding to the important data segments of each dimensional data, and finally, based on the convergence degree corresponding to the important data segments of each dimensional data and the preset dimensionality reduction algorithm, performs dimensionality reduction processing on the time series data corresponding to the odor data to confirm the odor type corresponding to the odor data. The dimensionality of the time series data corresponding to the odor data is constrained by the degree of convergence corresponding to the important data segments of each dimensional data, and then dimensionality reduction processing is performed to confirm the odor type corresponding to the odor data. This can avoid the problem that the data dimension obtained when the optical gas sensor detects odors is too high, resulting in poor subsequent data dimensionality reduction effect, further improve the accuracy of odor type judgment, and reduce work costs.

[0082] In one embodiment of the present application, see Figure 4 , the step S102: based on the amplitude and slope of each dimensional data, confirming the importance of the important data segment of each dimensional data, wherein the important data segment refers to a data segment consisting of data points located on both sides of the amplitude maximum point corresponding to each dimensional data and greater than the average amplitude, specifically including:

[0083] S401, counting the amplitude of each data point in each dimension of data, and calculating the average amplitude of each dimension of data.

[0084] Each dimensional data includes a preset number of data points, each data point corresponds to an amplitude, the amplitude of each data point in each dimensional data is counted, and then based on the amplitude of each data point, the average amplitude of each dimensional data is calculated.

[0085] S402: Calculate the importance of the important data segments of each dimensional data based on the amplitude of the maximum point of each dimensional data, the average amplitude, and the slope corresponding to each data point in the important data segment.

[0086] After obtaining the amplitude of the maximum point of each dimensional data, the average amplitude and the slope corresponding to each data point in the important data segment, the three are used as calculation parameters of the importance corresponding to the important data segment of the current dimensional data.

[0087] Specifically, the importance of the important data segment of each dimensional data is calculated based on the amplitude of the maximum point of each dimensional data, the average amplitude, and the slope corresponding to each data point in the important data segment, and specifically includes:

[0088]

[0089] in, For the The first dimension of data The importance of the important data segments corresponding to the maximum value points, For the The first dimension of data The amplitude of the maximum point, For the The average amplitude of the dimensional data, For the The number of data points in the important data segment corresponding to the maximum value point, For the The important data segment corresponding to the maximum value point The slope corresponding to the data point.

[0090] It should be noted that when calculating the importance of the important data segments of each dimension data, when the change trend or change degree of a certain data interval segment in each dimension data is large, it proves that the data interval contains more useful information. Therefore, the amplitude change size and slope change size corresponding to the important data segment are selected as the parameters of the change trend to further calculate the importance of the important data segment. The larger it is, the larger the amplitude span of the important data segment is, and the greater the corresponding importance needs to be. It is the sum of the slopes of all data points in the important data segment. The larger the slope, the greater the corresponding importance should be.

[0091] The embodiment of the present application obtains the sum of the amplitude span of the important data segment and the slope of the data points in the important data segment, and uses it as a calculation parameter to further obtain the importance corresponding to the important data segment, which is used as a calculation parameter for subsequent dimensional constraints to make subsequent dimensionality reduction more accurate and improve the accuracy of odor type judgment.

[0092] In one embodiment of the present application, refer to the attached Figure 5Step S103: clustering the importance levels of the important data segments of at least two dimensional data to confirm the degree of dispersion of the important data segments of each dimensional data in the corresponding cluster cluster, specifically including:

[0093] S501: Cluster the importance levels corresponding to important data segments of at least two dimensional data, and determine a preset number of clusters.

[0094] Among them, after obtaining the importance levels corresponding to the important data segments of the at least two dimensional data, the importance levels corresponding to the important data segments of the at least two dimensional data are clustered according to a preset clustering algorithm. The preset clustering algorithm can be a DBSCAN density clustering algorithm to confirm a preset number of cluster clusters.

[0095] It should be noted that the DBSCAN density clustering algorithm determines the formation of clusters by defining density and neighborhood. A core point is a point that contains at least MinPts data points in the neighborhood, a boundary point is a point that contains less than MinPts data points in the neighborhood but has a core point in the neighborhood, and a noise point is neither a core point nor a boundary point. DBSCAN forms clusters by expanding the neighborhood of core points and assigning boundary points to the same cluster. Noise points will be marked as noise or discarded. The specific clustering method is not further limited in this case, and reference can be made to the prior art.

[0096] S502: According to the cluster clusters, count the number and average amplitude of important data segments in the cluster cluster corresponding to the target important data segment, and the cluster distance between the target important data segment and the cluster center of the corresponding cluster cluster.

[0097] Among them, after obtaining a preset number of clusters through clustering, the number of all important data segments of the target important data segment in the cluster where it is located, as well as the average amplitude of all important data segments in the cluster can be counted. At the same time, the clustering distance between the target important data segment and the cluster center of the cluster where it is located is calculated, so as to subsequently calculate the degree of discreteness of the target important data segment in the corresponding cluster.

[0098] S503, inputting the number and average amplitude of the important data segments in the cluster corresponding to the target important data segment, and the cluster distance between the target important data segment and the cluster center of the corresponding cluster into a discrete degree calculation formula, and calculating the discrete degree of the target important data segment in the corresponding cluster.

[0099] Among them, after obtaining the number and average amplitude of important data segments in the cluster cluster corresponding to the target important data segment, as well as the cluster distance between the target important data segment and the cluster center of the corresponding cluster cluster, they are used as calculation parameters and input into the discrete degree calculation formula to calculate the discrete degree of the target important data segment in the corresponding cluster cluster.

[0100] Specifically, the number and average amplitude of the important data segments in the cluster cluster corresponding to the target important data segment, and the cluster distance between the target important data segment and the cluster center of the corresponding cluster cluster are input into the discrete degree calculation formula to calculate the discrete degree of the target important data segment in the corresponding cluster cluster, specifically including:

[0101]

[0102] in, For the The important data segments of the target in the clusters The degree of discreteness, For the The important data segments of the target in the clusters The importance of For the The number of important data segments in the clusters, Target important data segment Middle The amplitude of the data point, For the The average amplitude of important data segments in clusters, Target important data segment The number of data points, Target important data segment With The cluster distance of the cluster centers of the clusters, For is the base exponential function.

[0103] It should be noted that Refers to the target important data segment The importance of an important data segment accounts for the proportion of its importance in the importance of the cluster to which it belongs. When calculating the importance of an important data segment, it is obtained based on the change in amplitude in the important data segment and the distribution of data points in the important data segment. Therefore, the ratio of the importance of the target important data segment to the importance of the cluster to which it belongs can reflect the proportion of the importance of the target important data segment in the cluster to which it belongs. That is, the smaller the ratio is, the different the changing trend of the data point is from that of other data intervals, so the data has a larger discreteness. The target data segment Compared with the variance of the cluster to which it belongs, the larger the variance is, the greater the difference between the data segment and other data is, and therefore the greater the degree of dispersion is. Refers to the target important data segment With The clustering distance of the cluster centers of the cluster clusters, when the target important data segment The farther away from the cluster center, the smaller the target data segment is in the multidimensional data of the cluster. The change in is different from the change in all the data in the cluster, so the degree of dispersion is greater.

[0104] In one embodiment of the present application, refer to the attached Figure 6 Step S104, confirming the convergence degree corresponding to the important data segment of each dimensional data according to the importance degree and discrete degree corresponding to the important data segment of each dimensional data, specifically includes:

[0105] S601, calculating the sum of the amplitudes of the data points in the target important data segment and the sum of the amplitudes of the data points in the dimensional data corresponding to the target important data segment;

[0106] S602. Calculate the constraint weight corresponding to the target important data segment based on the importance and discreteness corresponding to the target important data segment, the sum of the amplitudes of the data points in the target important data segment, and the sum of the amplitudes of the data points in the dimensional data corresponding to the target important data segment.

[0107] The constraint weight corresponding to the target important data segment refers to certain constraints imposed on the target important data segment in the optimization problem. These constraints can be linear constraints, nonlinear constraints or other forms of constraints, which are used to limit the value range of weights or parameters to meet the specific requirements of the problem.

[0108] Specifically, based on the importance and discreteness corresponding to the target important data segment, the sum of the amplitudes of the data points in the target important data segment, and the sum of the amplitudes of the data points in the dimensional data corresponding to the target important data segment, the constraint weight corresponding to the target important data segment is calculated, specifically including:

[0109]

[0110] in, For the The important data segments of the target in the clusters The constraint weight, For the The important data segments of the target in the clusters The degree of discreteness, For the The important data segments of the target in the clusters The importance of For the Target important data segments of dimensional data No. The amplitude of the data point, Target important data segment The number of data points, Target important data segment The corresponding Dimensional data The amplitude of the data point, Target important data segment The corresponding The number of data points for each dimension.

[0111] It should be noted that For the Target important data segments of dimensional data The sum of the amplitudes of all data points, For the The sum of the amplitudes of all data points of the dimensional data, Refers to the target important data segment The ratio of the sum of the amplitudes of all data points in the dimension data to the sum of the amplitudes of all data points in the dimension data where it is located. The larger the ratio, the more likely it is that the target data segment For the The greater the impact of the dimensional data, the more it is necessary to Assign a larger constraint weight. At the same time, combining the importance and discreteness of the target important data segment as the calculation parameters of the constraint weight can make the constraint weight more accurate, so that the subsequent odor type judgment is more accurate.

[0112] S603. Input the constraint weight corresponding to the target important data segment and the amplitude mean of the dimensional data corresponding to the target important data segment into a convergence degree calculation formula to calculate the convergence degree corresponding to the target important data segment.

[0113] After obtaining the constraint weight corresponding to the target important data segment and the amplitude mean of the dimensional data corresponding to the target important data segment, the convergence degree corresponding to the target important data segment is calculated. The convergence degree corresponding to the target important data segment is the final state of the target important data segment that needs to be constrained.

[0114] Specifically, the step of inputting the constraint weight corresponding to the target important data segment and the amplitude mean of the dimensional data corresponding to the target important data segment into the convergence degree calculation formula to calculate the convergence degree corresponding to the target important data segment specifically includes:

[0115]

[0116] in, For the The important data segments of the target in the clusters The degree of convergence, For the The important data segments of the target in the clusters The constraint weight, Target important data segment The corresponding The mean amplitude of the data in each dimension.

[0117] The embodiment of the present application first collects time series data corresponding to the odor data in the gas cabinet to confirm the dimensional data corresponding to at least two dimensions based on the time series data, and then confirms the importance corresponding to the important data segments of each dimensional data based on the amplitude and slope of each dimensional data, wherein the important data segment refers to the data segment composed of data points on both sides of the amplitude maximum point corresponding to each dimensional data and greater than the average amplitude, and then clusters the importance corresponding to the important data segments of at least two dimensional data to confirm the degree of discreteness of the important data segments of each dimensional data in the corresponding clustering cluster, and further confirms the degree of convergence corresponding to the important data segments of each dimensional data based on the importance and discreteness corresponding to the important data segments of each dimensional data, and finally, based on the convergence degree corresponding to the important data segments of each dimensional data and the preset dimensionality reduction algorithm, performs dimensionality reduction processing on the time series data corresponding to the odor data to confirm the odor type corresponding to the odor data. The dimensionality of the time series data corresponding to the odor data is constrained by the degree of convergence corresponding to the important data segments of each dimensional data, and then dimensionality reduction processing is performed to confirm the odor type corresponding to the odor data. This can avoid the problem that the data dimension obtained when the optical gas sensor detects odors is too high, resulting in poor subsequent data dimensionality reduction effect, further improve the accuracy of odor type judgment, and reduce work costs.

[0118] The flowchart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to the embodiment of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowchart and the block diagram in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in a different order from the order disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

[0119] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for optimizing the collection and analysis of gas cabinet sensor data, applied in the field of odor data analysis, characterized in that: The method comprises: Collecting time series data corresponding to the odor data in the gas cabinet to confirm dimensional data corresponding to at least two dimensions according to the time series data; Based on the amplitude and slope of each dimensional data, determine the importance of the important data segment of each dimensional data, wherein the important data segment refers to a data segment consisting of data points located on both sides of the amplitude maximum point corresponding to each dimensional data and greater than the average amplitude; Clustering the importance of the important data segments of at least two dimensional data to determine the degree of dispersion of the important data segments of each dimensional data in the corresponding clustering cluster; According to the importance and discreteness corresponding to the important data segments of each dimensional data, the convergence degree corresponding to the important data segments of each dimensional data is determined; Based on the convergence degree corresponding to the important data segments of each dimensional data and the preset dimensionality reduction algorithm, the time series data corresponding to the odor data is subjected to dimensionality reduction processing to confirm the odor type corresponding to the odor data; Based on the convergence degree corresponding to the important data segments of each dimensional data and the preset dimensionality reduction algorithm, the time series data corresponding to the odor data is subjected to dimensionality reduction processing to confirm the odor type corresponding to the odor data, specifically including: Based on the convergence degree corresponding to the important data segments of each dimensional data, the time series data is converged to confirm the final time series data; Performing dimensionality reduction processing on the final time series data by using a preset dimensionality reduction algorithm to confirm the dimensionality reduction data; The odor type corresponding to the odor data is confirmed based on the dimension reduction data.

2. A method for optimizing the collection and analysis of gas cabinet sensor data according to claim 1, characterized in that: Based on the amplitude and slope of each dimensional data, the importance of the important data segment of each dimensional data is confirmed, wherein the important data segment refers to a data segment composed of data points located on both sides of the amplitude maximum point corresponding to each dimensional data and greater than the average amplitude, specifically including: Count the amplitude of each data point in each dimension of data and calculate the average amplitude of each dimension of data; Based on the amplitude of the maximum point of each dimensional data, the average amplitude and the slope corresponding to each data point in the important data segment, the importance of the important data segment of each dimensional data is calculated.

3. A method for optimizing the collection and analysis of gas cabinet sensor data according to claim 2, characterized in that: The step of calculating the importance of the important data segment of each dimensional data based on the amplitude of the maximum point of each dimensional data, the average amplitude and the slope corresponding to each data point in the important data segment specifically includes: in, For the Dimensional data The importance of the important data segments corresponding to the maximum value points, For the The first dimension of data The amplitude of the maximum point, For the The average amplitude of the dimensional data, For the The number of data points in the important data segment corresponding to the maximum value point, For the The important data segment corresponding to the maximum value point The slope corresponding to the data point.

4. A method for optimizing the collection and analysis of gas cabinet sensor data according to claim 3, characterized in that: The step of clustering the importance levels of the important data segments of at least two dimensional data and determining the degree of dispersion of the important data segments of each dimensional data in the corresponding cluster clusters specifically includes: Clustering the importance levels corresponding to the important data segments of at least two dimensional data to determine a preset number of clusters; According to the cluster clusters, counting the number and average amplitude of important data segments in the cluster clusters corresponding to the target important data segments, and the cluster distance between the target important data segments and the cluster center of the corresponding cluster clusters; The number and average amplitude of important data segments in the cluster corresponding to the target important data segment, as well as the cluster distance between the target important data segment and the cluster center of the corresponding cluster are input into the discrete degree calculation formula to calculate the discrete degree of the target important data segment in the corresponding cluster cluster.

5. A method for optimizing the collection and analysis of gas cabinet sensor data according to claim 4, characterized in that: The number and average amplitude of the important data segments in the cluster cluster corresponding to the target important data segment, as well as the cluster distance between the target important data segment and the cluster center of the corresponding cluster cluster are input into the discrete degree calculation formula to calculate the discrete degree of the target important data segment in the corresponding cluster cluster, specifically including: in, For the The important data segments of the target in the clusters The degree of discreteness, For the The important data segments of the target in the clusters The importance of For the The number of important data segments in the clusters, Target important data segment Middle The amplitude of the data point, For the The average amplitude of important data segments in clusters, Target important data segment The number of data points, Target important data segment With The cluster distance of the cluster centers of the clusters, For is the base exponential function.

6. A method for optimizing the collection and analysis of gas cabinet sensor data according to claim 5, characterized in that: The step of confirming the convergence degree corresponding to the important data segments of each dimensional data according to the importance degree and the discrete degree corresponding to the important data segments of each dimensional data specifically includes: Count the sum of the amplitudes of the data points in the target important data segment and the sum of the amplitudes of the data points in the dimensional data corresponding to the target important data segment; Calculate the constraint weight corresponding to the target important data segment based on the importance and discreteness corresponding to the target important data segment, the sum of the amplitudes of the data points in the target important data segment, and the sum of the amplitudes of the data points in the dimensional data corresponding to the target important data segment; The constraint weight corresponding to the target important data segment and the amplitude mean of the dimensional data corresponding to the target important data segment are input into the convergence degree calculation formula to calculate the convergence degree corresponding to the target important data segment.

7. A method for optimizing the collection and analysis of gas cabinet sensor data according to claim 6, characterized in that: The calculating of the constraint weight corresponding to the target important data segment based on the importance degree corresponding to the target important data segment, the discrete degree, the sum of the amplitudes of the data points in the target important data segment, and the sum of the amplitudes of the data points in the dimensional data corresponding to the target important data segment specifically includes: in, For the The important data segments of the target in the clusters The constraint weight, For the The important data segments of the target in the clusters The degree of discreteness, For the The important data segments of the target in the clusters The importance of For the Target important data segments of dimensional data No. The amplitude of the data point, Target important data segment The number of data points, Target important data segment The corresponding Dimensional data The amplitude of the data point, Target important data segment The corresponding The number of data points for each dimension.

8. A method for optimizing the collection and analysis of gas cabinet sensor data according to claim 7, characterized in that: The step of inputting the constraint weight corresponding to the target important data segment and the amplitude mean of the dimensional data corresponding to the target important data segment into a convergence degree calculation formula to calculate the convergence degree corresponding to the target important data segment specifically includes: in, For the The important data segments of the target in the clusters The degree of convergence, For the The important data segments of the target in the clusters The constraint weight, Target important data segment The corresponding The mean amplitude of the data in each dimension.

9. A method for optimizing the collection and analysis of gas cabinet sensor data according to any one of claims 1 to 8, characterized in that: The collecting of time series data corresponding to the odor data in the gas cabinet to confirm dimensional data corresponding to at least two dimensions according to the time series data specifically includes: Collecting time series data corresponding to the odor data in the gas cabinet, denoising the time series data according to a preset denoising algorithm, and confirming the final time series data; According to the final time series data, dimensional data corresponding to at least two dimensions are confirmed.

Images