A method for detecting activated carbon filtration efficiency based on big data analysis

Abstract

The present application relates to the technical field of filtration efficiency detection, and specifically to a method for detecting activated carbon filtration efficiency based on big data analysis, the method comprising: obtaining data related to each filtration process of the activated carbon equipment, and corresponding each filtration process to a high-dimensional point; dividing the high-dimensional points based on the overall change of the filtration detection index of each high-dimensional point; comprehensively analyzing the changes in the filtration detection index at different stages of the high-dimensional point, obtaining the adsorption characteristic degree of the filtration detection index, and classifying the high-dimensional points; combining the data differences between the point to be filtered and all high-dimensional points, predicting and judging the activated carbon filtration efficiency. The present application aims to solve the limitations faced by traditional distance metrics when classifying high-dimensional historical filtration data, so as to improve the accuracy of filtration efficiency prediction and enhance the accurate judgment of filtration efficiency.

Description

Technical Field

[0001] The present application relates to the technical field of filtration efficiency detection, and specifically to a method for detecting activated carbon filtration efficiency based on big data analysis. Background Art

[0002] Activated carbon filtration is a process of water treatment using activated carbon as filter material. Activated carbon is a black solid material with rich pore structure and carbon as the main skeleton after gasification (carbonization, activation) treatment. Activated carbon adsorbs pollutants in water or air onto its surface through adsorption. Over time, the adsorption sites on the surface of activated carbon will gradually become saturated, reducing the adsorption efficiency. Therefore, in order to maintain the efficient filtration performance of the activated carbon filter, it is necessary to predict the filtration efficiency of activated carbon.

[0003] When predicting the filtration efficiency of activated carbon equipment, historical data is usually relied on for analysis and model building. For this reason, clustering algorithms are often used to classify historical data, which usually involves using Euclidean distance as a distance metric. However, when faced with high-dimensional historical data, the distances between data points become sparse, and the distances between most data points become very similar, making it difficult to accurately classify the data related to the filtration efficiency of activated carbon, affecting the accuracy of the judgment of the filtration efficiency of activated carbon. Summary of the invention

[0004] In view of the above, it is necessary to provide an activated carbon filtration efficiency detection method based on big data analysis to solve the limitations of traditional distance metrics when classifying high-dimensional historical filtration data, so as to improve the accuracy of filtration efficiency prediction and enhance the accurate judgment of filtration efficiency.

[0005] An embodiment of the present application provides a method for detecting activated carbon filtration efficiency based on big data analysis, the method comprising:

[0006] The filter detection index of the preset type at all times of each filtration treatment of the activated carbon equipment is taken as a high-dimensional point;

[0007] Based on the overall changes of various filtration detection indicators at each high-dimensional point of the activated carbon equipment, each filtration treatment of activated carbon is divided into multiple stages;

[0008] Comprehensively analyze the changes of various filtration detection indicators at different stages of each high-dimensional point of the activated carbon equipment to obtain the adsorption characteristics of various filtration detection indicators;

[0009] Based on the difference in the starting time data of all kinds of filtration detection indicators at different high-dimensional points of the activated carbon equipment, combined with the adsorption characteristic degree, the high-dimensional points are classified;

[0010] Based on the classification results of high-dimensional points and the data difference between the point to be filtered and all high-dimensional points, the activated carbon efficiency of the point to be filtered is predicted and judged; among them, the high-dimensional point composed of the data at the starting time of various filtration detection indicators of the activated carbon equipment to be filtered is recorded as the point to be filtered.

[0011] The step of dividing each filtration process of activated carbon into multiple stages includes:

[0012] For any filtering detection index of each high-dimensional point of the activated carbon equipment, mutation point detection is performed on the data at all times in the corresponding time period, and the mutation points are selected as segmentation points to obtain all stages of any filtering detection index of each high-dimensional point.

[0013] The specific process of screening mutation points is as follows:

[0014] If the number of mutation points is equal to the preset number, the time periods corresponding to each mutation point and the adjacent mutation points are taken as each stage;

[0015] If the number of mutation points is greater than the preset number, the average value of each mutation point data and the starting time data is calculated, and the mutation point with the largest average value is used as the first segmentation point; the average value of each mutation point after the first segmentation point and the ending time data is calculated, and the mutation point with the smallest average value is used as the third segmentation point;

[0016] For each mutation point between the first division point and the third division point, the data difference between each mutation point and the first division point at the corresponding time, and the inverse proportional result of the corresponding time interval are used as the first speed change value of each mutation point. Based on the third division point, the second speed change value of each mutation point is obtained by the same calculation method as the first speed change value; the mutation point with the largest difference between the first speed change value and the second speed change value is used as the second division point.

[0017] The steps for obtaining the adsorption characteristics of the various filtration detection indicators are as follows:

[0018] For each high-dimensional point, the difference between the data of various filtering detection indicators at each moment and the adjacent moments in any stage is taken as the adsorption rate at each moment in any stage;

[0019] Based on the overall change of the adsorption speed of various filtration detection indicators in the first stage, the adsorption characteristic values ​​of various filtration detection indicators in the first stage are obtained;

[0020] Obtain the data difference between each moment of the first stage and the end moment of various filtering detection indicators, and record it as the overall change value of various filtering detection indicators in the first stage; calculate the difference between the overall change values ​​of various filtering detection indicators in the first stage and the second stage, and obtain the change difference value of various filtering detection indicators in the first stage and the second stage; calculate the difference in adsorption speed of various filtering detection indicators at each moment and adjacent moments in the second stage, and fuse it with the adsorption speed at each moment; perform inverse proportional mapping of the change difference value of the first stage and the second stage with the fusion result of all moments in the second stage, and obtain the adsorption characteristic value of various filtering detection indicators in the second stage;

[0021] The result of fusing the adsorption characteristic values ​​of various filtration detection indicators in the first stage and the second stage is used as the adsorption characteristic degree of various filtration detection indicators.

[0022] The method of obtaining the adsorption characteristic values ​​of various filtration detection indicators in the first stage is specifically as follows:

[0023] For the first stage of each high-dimensional point, the difference in adsorption speed between each moment and the adjacent moment is calculated as the speed change at each moment; the time interval between each moment and the end moment is obtained; the cumulative sum of the inverse proportional mapping results of the speed change at all moments and the corresponding time intervals is calculated to obtain the relative speed change in the first stage; the inverse proportional mapping results of the relative speed change of various filtering detection indicators in the first stage and the overall change value are used as the adsorption characteristic values ​​of various filtering detection indicators in the first stage.

[0024] The process of classifying high-dimensional points is specifically as follows:

[0025] Based on the starting time data of any two high-dimensional points, combined with the adsorption characteristic degree, a characteristic distance between the any two high-dimensional points is obtained;

[0026] The preset characteristic distances between any high-dimensional point and the remaining high-dimensional points are used as the candidate radii of the any high-dimensional point, and the candidate radii are used as the neighborhood radius of the any high-dimensional point in turn; the total number of points of the remaining high-dimensional points in the neighborhood corresponding to any candidate radius and the degree of disorder of the corresponding characteristic distances are obtained; the frequency of the total number of points in the neighborhood corresponding to any candidate radius of the any high-dimensional point in all neighborhoods of all high-dimensional points is counted; the maximum value of the frequency of the any high-dimensional point and the fusion result of the degree of disorder of all neighborhoods are used as the preferred value of the any high-dimensional point based on the any candidate radius;

[0027] The candidate radius with the largest preferred value is used as the neighborhood radius of the clustering algorithm, and the feature distance is used as the metric distance between high-dimensional points. The high-dimensional points are classified to obtain all categories.

[0028] The step of obtaining the characteristic distance between any two high-dimensional points is specifically:

[0029] The starting time data of various filtration detection indicators at each high-dimensional point of the activated carbon equipment and the corresponding adsorption characteristic degree are fused and recorded as the adsorption distance of various filtration detection indicators; the differences in all the adsorption distances of any two high-dimensional points are fused to obtain the characteristic distance of the any two high-dimensional points.

[0030] Wherein, the step of predicting and judging the efficiency of activated carbon at the filtering treatment point includes:

[0031] Based on the data similarity between the to-be-filtered processing point and various filtering detection data indicators of each category of high-dimensional points, the predicted category of the to-be-filtered processing point is obtained from all categories for classifying the high-dimensional points;

[0032] The overall changes of various filtration detection indicators of all high-dimensional points in the comprehensive prediction category are combined with the adsorption characteristic degree to obtain the predicted activated carbon filtration efficiency of the point to be filtered;

[0033] When the predicted activated carbon filtration efficiency of the filter point to be detected is less than the preset filtration efficiency judgment threshold, it is judged that there is a problem with the filtration efficiency of the activated carbon equipment; otherwise, it is judged that there is no problem with the filtration efficiency of the activated carbon equipment.

[0034] The step of obtaining the predicted category of the point to be filtered is as follows:

[0035] The difference between the starting time data of all kinds of filtering detection indicators of the point to be filtered and each high-dimensional point is merged to determine the initial distance between the point to be filtered and each high-dimensional point;

[0036] For each category after high-dimensional points are classified, an inverse proportional mapping of the average value of the initial distance between the point to be filtered and all high-dimensional points in each category is obtained, which is recorded as the predicted optimal value of each category. The category with the largest predicted optimal value is selected as the predicted category of the point to be filtered.

[0037] The step of obtaining the predicted activated carbon filtration efficiency of the point to be filtered includes:

[0038] Calculate the data difference of various filtering detection indicators at the start and end time of each high-dimensional point, and record it as the total change value of various filtering detection indicators at each high-dimensional point; calculate the total change value of all filtering detection indicators at each high-dimensional point and the inverse proportional mapping of the corresponding standard value to merge them as the filtering efficiency of each high-dimensional point;

[0039] The normalized value of the inverse proportional mapping result between the adsorption characteristic degree of each high-dimensional point in the prediction category and the corresponding initial distance is obtained, and recorded as the reference weight of each high-dimensional point in the prediction category; the reference weight of all high-dimensional points in the prediction category is merged with the filtration efficiency to obtain the predicted activated carbon filtration efficiency of the point to be filtered.

[0040] This application has at least the following beneficial effects:

[0041] The present application first maps the data of each historical filtration treatment of the activated carbon device to high-dimensional points, providing a data basis for subsequent data classification and prediction of the activated carbon filtration efficiency; divides each activated carbon filtration treatment into stages, effectively measures the different changes in the filtration detection indicators during each activated carbon filtration treatment, and obtains the adsorption characteristic degree based on the change characteristics of different stages, characterizing the adsorption characteristics of the activated carbon on the substance and the historical filtration efficiency at different stages; by introducing the adsorption characteristic degree and combining the data difference at the starting time to characterize the distance between high-dimensional points, the classification of high-dimensional points can be effectively completed, overcoming the limitations of traditional distance metrics when classifying high-dimensional data, and can eliminate abnormal data and improve the accuracy of classification; combined with the data difference between high-dimensional points in similar categories to the points to be filtered, the filtration efficiency of the current activated carbon filtration equipment is predicted and judged, the filtration efficiency prediction accuracy is improved, and the accuracy of filtration efficiency judgment is enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A flow chart of an activated carbon filtration efficiency detection method based on big data analysis provided in this application;

[0043] Figure 2 A flow chart for obtaining adsorption characteristics provided in this application;

[0044] Figure 3 A flowchart for classifying high-dimensional points provided for this application;

[0045] Figure 4 A flow chart for predicting and judging the efficiency of activated carbon at a filtering treatment point provided in this application. DETAILED DESCRIPTION

[0046] 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.

[0047] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art in the present application. The terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application.

[0048] It should also be noted that the terms "first" and "second" in this application and its drawings 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 flow chart includes one or more steps for implementing the method. Without departing from the scope of protection of this application, the execution order of multiple steps can be interchanged with each other, and some steps can also be deleted.

[0049] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0050] This application first proposes a method for detecting activated carbon filtration efficiency based on big data analysis, which is applied to the technical field of filtration efficiency detection. Figure 1 , the method comprises the following steps:

[0051] S1: The preset type of filtration detection index at all times of each filtration treatment of the activated carbon equipment to be tested is taken as a high-dimensional point.

[0052] This application mainly analyzes and predicts the filtration efficiency of activated carbon by analyzing the filtration detection indicators of each filtration treatment process of the activated carbon equipment, so as to determine whether the activated carbon needs to be regenerated or replaced.

[0053] Based on this, the filtration detection index of each filtration treatment of the activated carbon equipment is obtained;

[0054] It should be noted that the filtration process is a continuous process that can filter water and gas. The data collected during the filtration process includes various filtration detection indicators at all times from the beginning to the end of the filtration process. Among them, the filtration detection indicators include but are not limited to turbidity, organic matter concentration, heavy metal ion concentration, ammonia nitrogen content, dissolved oxygen content, pH value, etc. The implementer can select appropriate filtration detection indicators for analysis, and this application does not limit this.

[0055] Some embodiments of the present application mainly detect the filtration efficiency of activated carbon equipment on water. Specifically, a turbidity sensor is used to obtain the turbidity of the filtration process in real time, a total organic carbon (TOC) analyzer is used to obtain the concentration of organic matter in the filtration process in real time, a heavy metal analyzer is used to obtain the concentration of heavy metal ions in the filtration process in real time, and an ammonia nitrogen sensor is used to obtain the ammonia nitrogen content in the filtration process in real time.

[0056] All kinds of filtration detection indicators at all times in the corresponding time period obtained by each filtration treatment of the activated carbon device are regarded as a high-dimensional point, that is, one high-dimensional point corresponds to a complete filtration process of the activated carbon device.

[0057] In some embodiments of the present application, for each high-dimensional point, there are four corresponding filtration detection indicators: turbidity, organic matter concentration, heavy metal ion concentration, and ammonia nitrogen concentration, and there is data for the corresponding time period under each filtration detection indicator.

[0058] S2: Based on the overall changes of various filtration detection indicators at each high-dimensional point of the activated carbon equipment, each filtration treatment of activated carbon is divided into multiple stages.

[0059] Since the initial distance of the high-dimensional points reflects the data changes at the starting time of water filtration, and in the actual treatment process, the saturation degree of the activated carbon adsorption points and environmental parameters are different during different water filtration treatments, the activated carbon filtration efficiency corresponding to the high-dimensional points will also change accordingly. Therefore, the initial distance cannot effectively reflect the distance characteristics between the high-dimensional points.

[0060] Based on this, each filtration process of activated carbon is divided into:

[0061] For any filtering detection index of each high-dimensional point of the activated carbon equipment, mutation point detection is performed on the data at all times in the corresponding time period, and the mutation points are selected as segmentation points to obtain all stages of any filtering detection index of each high-dimensional point.

[0062] It should be noted that the number of mutation points detected by mutation point detection is greater than or equal to 3. In each filtration process of activated carbon, the activated carbon does not start to adsorb in the initial stage, so the changes in the filtration detection indicators generally have three stages. The first stage is the rapid adsorption stage. At this time, the activated carbon begins to contact the pollutants. Since the adsorption sites on the surface of the activated carbon are not saturated, the adsorption rate is fast, and the filtration detection indicators drop rapidly during this stage. The second stage is the stable adsorption stage. As the pollutants are absorbed, the adsorption sites on the surface of the activated carbon gradually become saturated, and the adsorption rate will gradually slow down; in this stage, the rate of decrease of the pollutant concentration continues to slow down, showing a gentle downward trend, and the changes in the filtration detection indicators are no longer drastic. The third stage is the saturation stage. At this time, the pollutant concentration reaches when the adsorption sites on the surface of the activated carbon are saturated or the pollutant concentration is too low. The diffusion rate of the pollutant molecules and the surface of the activated carbon is relatively slow. At this time, the pollutant concentration no longer decreases, and the filtration detection indicators remain in a stable state.

[0063] Based on this, the specific process of screening mutation points is:

[0064] If the number of mutation points is equal to the preset number, the time period corresponding to each mutation point and the adjacent mutation point is taken as each stage; if the number of mutation points is greater than the preset number, the average value of the data of each mutation point and the data at the starting time is calculated, and the mutation point with the largest average value is taken as the first segmentation point; the average value of the data of each mutation point after the first segmentation point and the data at the ending time is calculated, and the mutation point with the smallest average value is taken as the third segmentation point; for each mutation point between the first segmentation point and the third segmentation point, the difference between the data at the corresponding time of each mutation point and the first segmentation point, and the inverse proportional result of the corresponding time interval are taken as the first speed change value of each mutation point, and based on the third segmentation point, the second speed change value of each mutation point is obtained by using the same calculation method as the first speed change value; the mutation point with the largest difference between the first speed change value and the second speed change value is taken as the second segmentation point.

[0065] It should be noted that mutation point detection is used to identify significant change points in time series data. Implementers can choose Manner-Kendall mutation point detection method, Bernaola Galvan segmentation algorithm, Pettitt mutation detection method, Pelt (Pruned Exact Linear Time) mutation point detection algorithm, etc. to perform mutation point detection on the sequence; implementers can select appropriate mutation point detection methods according to actual conditions, and this application does not impose any restrictions on this.

[0066] In some embodiments of the present application, since the processing methods of various filtering detection indicators for the time period corresponding to each high-dimensional point are the same, the turbidity indicator is taken as an example, and the mutation points of all data for the time period corresponding to the turbidity indicator are obtained by the Pelt mutation point detection algorithm. The preset number is 3. If the detected mutation points are 3, they are recorded as the first, second, and third segmentation points in the order of collection; if the detected mutation points are greater than 3, the mutation points are further screened, and the mutation point with the largest average value of the turbidity indicator between the start time and the start time is taken as the first segmentation point; starting from the first segmentation point, the mutation point with the smallest average value of the turbidity indicator between the end time and the end time is taken as the third segmentation point; for each mutation point between the first segmentation point and the third segmentation point, the ratio of the absolute value of the difference between each mutation point and the turbidity indicator at the time corresponding to the first segmentation point to the corresponding time interval is taken as the first speed change value of each mutation point, and based on the third segmentation point, the second speed change value of each mutation point is obtained by the same calculation method as the first speed change value; the mutation point with the largest absolute value of the difference between the first speed change value and the second speed change value is taken as the second segmentation point. The time period from the turbidity index at the starting moment to the first mutation point is recorded as the starting stage, and the time period from the first division point to the second division point is recorded as the first stage. Correspondingly, the second stage and the third stage are obtained.

[0067] S3: Comprehensively analyze the changes of various filtration detection indicators at different stages of each high-dimensional point of the activated carbon equipment to obtain the adsorption characteristics of various filtration detection indicators.

[0068] For each high-dimensional point, the difference between the data of various filtering detection indicators at each moment in any stage and the data of adjacent moments is taken as the adsorption speed at each moment in any stage.

[0069] Based on the overall change of the adsorption speed of various filtration detection indicators in the first stage, the adsorption characteristic values ​​of various filtration detection indicators in the first stage are obtained:

[0070] For the first stage of each high-dimensional point, the difference in adsorption speed between each moment and the adjacent moment is calculated as the speed change at each moment; the time interval between each moment and the end moment is obtained; the cumulative sum of the inverse proportional mapping results of the speed change at all moments and the corresponding time intervals is calculated to obtain the relative speed change in the first stage; the data difference between each moment and the end moment of the first stage of various filtering detection indicators is obtained, and recorded as the overall change value of various filtering detection indicators in the first stage; the inverse proportional mapping results of the relative speed change of various filtering detection indicators in the first stage and the overall change value are used as the adsorption characteristic values ​​of various filtering detection indicators in the first stage.

[0071] It should be noted that the difference represents the numerical difference between variables, specifically: difference, absolute value of difference, ratio, etc.; generally, the result of subtraction or division is used to represent the inverse proportional relationship between two variables; this application does not impose any restrictions on this.

[0072] Since the adsorption characteristics of the first stage are that the adsorption speed is large and the adsorption speed is constantly decaying, and since the analysis methods for various filtration detection indicators are consistent, in some embodiments of the present application, the turbidity indicator is taken as an example to construct the adsorption characteristic value of the turbidity indicator in the first stage Among them, ND qs ND zz Respectively represent the turbidity index data at the start and end of the first stage, and the absolute value of the difference |ND qs -ND zz | represents the overall change value of the turbidity index in the first stage; N represents the total number of turbidity index data in the first stage, V n 、V n+1 They represent the adsorption rates at the nth and n+1th moments in the first stage, respectively. n Represents the time interval between the nth moment in the first stage and the end moment of the first stage.

[0073] It should be understood that the smaller the relative speed change of the turbidity index in the first stage, the more the change in the adsorption rate conforms to the adsorption characteristics of the first stage of activated carbon adsorption. The larger the corresponding adsorption characteristic value, the less likely the data collected in the first stage is to be interfered with, and the greater the credibility of the data.

[0074] Based on the overall change of the adsorption speed of various filtration detection indicators in the second stage, the adsorption characteristic values ​​of various filtration detection indicators in the second stage are obtained:

[0075] The difference between the overall change values ​​of various filtering detection indicators in the first stage and the second stage is calculated to obtain the change difference values ​​of various filtering detection indicators in the first stage and the second stage; the difference in adsorption speed of various filtering detection indicators at each moment and adjacent moments in the second stage is calculated, and the difference is fused with the adsorption speed at each moment; the change difference value between the first stage and the second stage is inversely mapped with the fusion result of all moments in the second stage to obtain the adsorption characteristic values ​​of various filtering detection indicators in the second stage.

[0076] It should be noted that the specific operations of fusion are addition, multiplication and mixed operations of addition and multiplication, and this application does not impose any restrictions on this.

[0077] In some embodiments of the present application, taking the turbidity index as an example, the adsorption characteristic value of the second stage turbidity index is constructed. Among them, C1 and C2 represent the overall change values ​​of turbidity index in the first stage and the second stage respectively. The larger the difference value of |C1-C2|, the smaller the change of pollutant concentration in the second stage compared with the first stage; U represents the total number of turbidity index data in the second stage, V u 、V u+1 They represent the adsorption speeds at the u-th and u+1-th moments in the second stage respectively. ε represents a parameter adjustment factor that is preset to be greater than zero to prevent the denominator from being 0, and its value is 0.001.

[0078] It should be noted that the absolute value of the difference in adsorption rate is the adsorption rate change, and multiplying it by V u This is to reflect the characteristic of lower adsorption rate at each moment in the second stage.

[0079] The result of fusing the adsorption characteristic values ​​of various filtration detection indicators in the first stage and the second stage is used as the adsorption characteristic degree of various filtration detection indicators.

[0080] In some embodiments of the present application, the sum of the adsorption characteristic values ​​of various filtration detection indicators in the first stage and the second stage is used as the adsorption characteristic degree of the various filtration detection indicators.

[0081] Among them, the flow chart for obtaining adsorption characteristics is as follows Figure 2 shown.

[0082] It should be noted that each high-dimensional point of the activated carbon equipment has an adsorption characteristic under each filtration detection index.

[0083] S4: Based on the difference in the starting time data of all kinds of filtration detection indicators at different high-dimensional points of the activated carbon equipment, combined with the adsorption characteristic degree, the high-dimensional points are classified.

[0084] The starting time data of various filtration detection indicators at each high-dimensional point of the activated carbon equipment and the corresponding adsorption characteristic degree are fused and recorded as the adsorption distance of various filtration detection indicators; the differences in all the adsorption distances of any two high-dimensional points are fused to obtain the characteristic distance of the any two high-dimensional points.

[0085] The preset characteristic distances between any high-dimensional point and the remaining high-dimensional points are used as the respective candidate radii of the any high-dimensional point, and the respective candidate radii are used in turn as the neighborhood radius of the any high-dimensional point; the total number of points of the remaining high-dimensional points in the neighborhood corresponding to any candidate radius and the degree of confusion of the corresponding characteristic distances are obtained; the frequency of the total number of points in the neighborhood corresponding to any candidate radius of the any high-dimensional point in all neighborhoods of all high-dimensional points is counted; and the maximum value of the frequency of the any high-dimensional point and the fusion result of the degree of confusion of all neighborhoods are used as the preferred value of the any high-dimensional point based on the any candidate radius.

[0086] It should be noted that the degree of chaos reflects the irregularity of the overall data distribution, including: Gini impurity, information entropy, variance, etc.; this application does not impose any restrictions on this.

[0087] In some embodiments of the present application, for various filtration detection indicators at each high-dimensional point of the activated carbon equipment, the product of the starting time data and the corresponding adsorption characteristic degree is calculated and recorded as the adsorption distance of the various filtration detection indicators; the absolute values ​​of the differences of all the adsorption distances of any two high-dimensional points are added to obtain the characteristic distance of the any two high-dimensional points.

[0088] Furthermore, every two high-dimensional points have a corresponding characteristic distance. Taking the high-dimensional point q as an example, obtain the K high-dimensional points with the smallest characteristic distance from it. In this application, K=5 is set, and these K characteristic distances are used as the selected radii in turn. Each high-dimensional point can obtain K selected radii. Here, taking the sth selected radius as an example, the high-dimensional point q is taken as the center, and the sth selected radius is taken as the neighborhood radius to obtain the sth neighborhood of the high-dimensional point q, and the total number of points of the remaining high-dimensional points in the sth neighborhood of the high-dimensional point q, as well as the characteristic distances between the remaining high-dimensional points in the sth neighborhood and the high-dimensional point q are counted; construct the preferred value of the sth selected radius Among them, f max represents the maximum probability of the other high-dimensional points in the sth neighborhood of the high-dimensional point q, S represents the total number of high-dimensional points, P sRepresents the feature distance permutation entropy of the s-th neighborhood of the high-dimensional point q.

[0089] It should be noted that the neighborhood of each high-dimensional point has the corresponding total number of other high-dimensional points. The frequency of the total number of points corresponding to each neighborhood of each high-dimensional point is obtained, and the maximum frequency is f max The feature distance permutation entropy is specifically the permutation entropy of the feature distances of the high-dimensional point and the other high-dimensional points in the corresponding neighborhood.

[0090] The candidate radius with the largest preferred value is used as the neighborhood radius of the clustering algorithm, and the feature distance is used as the metric distance between high-dimensional points. The high-dimensional points are classified to obtain all categories.

[0091] In some embodiments of the present application, each selected radius of a high-dimensional point has a corresponding preferred value, and the selected radius with the largest preferred value is used as the neighborhood radius in the DBSCAN algorithm, and minPts is set to 5 to cluster all high-dimensional points. It should be noted that the settings of the clustering algorithm parameters can be adjusted by the implementer. The DBSCAN clustering process is a well-known technology and will not be repeated here.

[0092] Among them, the flowchart for classifying high-dimensional points is as follows Figure 3 shown.

[0093] S5: Based on the classification results of high-dimensional points and the data difference between the point to be filtered and all high-dimensional points, the activated carbon efficiency of the point to be filtered is predicted and judged.

[0094] The data of the starting time of various filtering detection indicators of the activated carbon equipment to be filtered are obtained and recorded as the filtering treatment point; the difference between the starting time data of all kinds of filtering detection indicators of the filtering treatment point and each high-dimensional point is merged to determine the initial distance between the filtering treatment point and each high-dimensional point.

[0095] For each category after high-dimensional points are classified, an inverse proportional mapping of the average value of the initial distance between the point to be filtered and all high-dimensional points in each category is obtained, which is recorded as the predicted optimal value of each category. The category with the largest predicted optimal value is selected as the predicted category of the point to be filtered.

[0096] In some embodiments of the present application, data of the starting time of various filtering detection indicators of the activated carbon equipment to be filtered are obtained and recorded as the filtering treatment point; the absolute value of the difference between the filtering treatment point to be filtered and all kinds of filtering detection indicators of each high-dimensional point at the starting time is calculated, and the cumulative sum of all the absolute values ​​of the difference at the starting time is taken as the initial distance between any two high-dimensional points.

[0097] It should be understood that the initial distance measures the degree of difference between each filtration detection index at the starting moment of the two high-dimensional point processes of the activated carbon device. The greater the degree of difference, the greater the initial distance.

[0098] It should be noted that the points to be filtered can be regarded as high-dimensional points with only the starting time data of various filtering detection indicators; since the initial distance is obtained for the initial time data of the high-dimensional point, the initial distance between the points to be filtered and the high-dimensional point can be obtained.

[0099] Furthermore, based on the clustered categories, high-dimensional points that are more similar to the points to be filtered are screened out, and the inverse of the average value of the initial distance between all high-dimensional points in each category and the points to be filtered is obtained, which is recorded as the predicted optimal value of each category. Then each category can obtain the corresponding category prediction optimal value, and the category with the largest category prediction optimal value is used as the predicted category of the points to be filtered.

[0100] The data differences of various filtering detection indicators at the starting and ending times of each high-dimensional point are calculated, and recorded as the total change value of various filtering detection indicators at each high-dimensional point; the total change value of all filtering detection indicators at each high-dimensional point is calculated and fused with the inverse proportional mapping of the corresponding standard value, which is used as the filtering efficiency of each high-dimensional point.

[0101] In some embodiments of the present application, each high-dimensional point corresponds to data of a sewage treatment process, and the activated carbon filtration efficiency of the sewage treatment process is in Respectively represent the data of the rth filtration detection index at the start and end time of the sewage treatment; BZ r It represents the standard value of the rth filtration detection index when the water quality reaches the discharge standard; R represents the number of types of filtration detection indicators.

[0102] It should be noted that the standard values ​​of various filtration detection indicators when the water quality reaches the discharge standard can be obtained from prior knowledge.

[0103] Based on the activated carbon filtration efficiency corresponding to each high-dimensional point and combined with the prediction category, the current activated carbon filtration efficiency is predicted:

[0104] The normalized value of the inverse proportional mapping result between the adsorption characteristic degree of each high-dimensional point in the prediction category and the corresponding initial distance is obtained, and recorded as the reference weight of each high-dimensional point in the prediction category; the reference weight of all high-dimensional points in the prediction category is merged with the filtration efficiency to obtain the predicted activated carbon filtration efficiency of the point to be filtered.

[0105] In some embodiments of the present application, the predicted activated carbon filtration efficiency of the test point to be filtered is obtained by the predicted reference and filtration efficiency corresponding to the high-dimensional point in the predicted category of the test point to be filtered. Among them, CK i Represents the reference weight of the i-th high-dimensional point in the prediction category, represents the activated carbon filtration efficiency of the i-th high-dimensional point in the prediction category, and L represents the number of high-dimensional points in the prediction category; wherein, the reference weight of the i-th high-dimensional point is specifically the ratio of the adsorption characteristic degree of the i-th high-dimensional point to the corresponding initial distance.

[0106] When the predicted activated carbon filtration efficiency of the filter point to be tested is less than the preset filtration efficiency judgment threshold, it is judged that there is a problem with the filtration efficiency of the activated carbon equipment, and the activated carbon equipment needs to be thermally regenerated or replaced before filtering treatment; otherwise, it is judged that there is no problem with the filtration efficiency of the activated carbon equipment and it can continue to be used.

[0107] Among them, the flowchart for predicting the efficiency of activated carbon at the filtration treatment point is as follows Figure 4 shown.

[0108] In some embodiments of the present application, the preset filtration efficiency judgment value is 0.8.

[0109] The embodiment of the present application provides an activated carbon filtration efficiency detection method based on big data analysis, the method comprising: firstly mapping the data of each historical filtration treatment of the activated carbon device to high-dimensional points, providing a data basis for subsequent data classification and prediction of the activated carbon filtration efficiency; dividing each activated carbon filtration treatment into stages, effectively measuring the different change processes of the filtration detection indicators during each activated carbon filtration treatment, and obtaining the adsorption characteristic degree based on the change characteristics of different stages, characterizing the adsorption characteristics of the activated carbon on the substance and the historical filtration efficiency at different stages; by introducing the adsorption characteristic degree and combining the data difference at the starting time to characterize the distance between high-dimensional points, the classification of high-dimensional points can be effectively completed, overcoming the limitations of traditional distance measurement when classifying high-dimensional data, and eliminating abnormal data, and improving the accuracy of classification; combining the data difference between high-dimensional points in similar categories to the points to be filtered, predicting and judging the filtration efficiency of the current activated carbon filtration device, improving the prediction accuracy of filtration efficiency, and enhancing the accuracy of filtration efficiency judgment.

[0110] 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 application. 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 function marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two continuous 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 continuous 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.

[0111] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for detecting activated carbon filtration efficiency based on big data analysis, characterized in that: The method comprises the following steps: The preset type of filtration detection index of the activated carbon equipment at all times during each filtration process is taken as a high-dimensional point; Based on the overall changes of various filtration detection indicators at each high-dimensional point of the activated carbon equipment, each filtration treatment of activated carbon is divided into multiple stages; Comprehensively analyze the changes of various filtration detection indicators at different stages of each high-dimensional point of the activated carbon equipment to obtain the adsorption characteristics of various filtration detection indicators; Based on the difference in the starting time data of all kinds of filtration detection indicators at different high-dimensional points of the activated carbon equipment, combined with the adsorption characteristic degree, the high-dimensional points are classified; Based on the classification results of high-dimensional points, combined with the data differences between the point to be filtered and all high-dimensional points, the activated carbon efficiency of the point to be filtered is predicted and judged; wherein, the high-dimensional point composed of the data of the starting time of various filtering detection indicators of the activated carbon equipment to be filtered is recorded as the point to be filtered; The filtration detection indicators include: turbidity, organic matter concentration, heavy metal ion concentration, ammonia nitrogen content, dissolved oxygen content, and pH value; The step of dividing each filtering process of the activated carbon into multiple stages comprises: For any filtering detection index at each high-dimensional point of the activated carbon equipment, mutation point detection is performed on the data at all times in the corresponding time period, and the mutation points are selected as segmentation points to obtain all stages of any filtering detection index at each high-dimensional point; The steps for obtaining the adsorption characteristic of the various filtration detection indicators are as follows: For each high-dimensional point, the difference between the data of various filtering detection indicators at each moment and the adjacent moments in any stage is taken as the adsorption rate at each moment in any stage; Based on the overall change of the adsorption speed of various filtration detection indicators in the first stage, the adsorption characteristic values ​​of various filtration detection indicators in the first stage are obtained; Obtain the data difference between each moment of the first stage and the end moment of various filtering detection indicators, and record it as the overall change value of various filtering detection indicators in the first stage; calculate the difference between the overall change values ​​of various filtering detection indicators in the first stage and the second stage, and obtain the change difference value of various filtering detection indicators in the first stage and the second stage; calculate the difference in adsorption speed of various filtering detection indicators at each moment and adjacent moments in the second stage, and fuse it with the adsorption speed at each moment; perform inverse proportional mapping of the change difference value of the first stage and the second stage with the fusion result of all moments in the second stage, and obtain the adsorption characteristic value of various filtering detection indicators in the second stage; The result of fusing the adsorption characteristic values ​​of various filtration detection indicators in the first stage and the second stage is used as the adsorption characteristic degree of various filtration detection indicators.

2. The method for detecting activated carbon filtration efficiency based on big data analysis according to claim 1, characterized in that: The specific process of screening mutation points is as follows: If the number of mutation points is equal to the preset number, the time periods corresponding to each mutation point and the adjacent mutation points are taken as each stage; If the number of mutation points is greater than the preset number, the average value of each mutation point data and the starting time data is calculated, and the mutation point with the largest average value is used as the first segmentation point; the average value of each mutation point after the first segmentation point and the ending time data is calculated, and the mutation point with the smallest average value is used as the third segmentation point; For each mutation point between the first division point and the third division point, the data difference between each mutation point and the first division point at the corresponding time, and the inverse proportional result of the corresponding time interval are used as the first speed change value of each mutation point. Based on the third division point, the second speed change value of each mutation point is obtained by using the same calculation method as the first speed change value; the mutation point with the largest difference between the first speed change value and the second speed change value is used as the second division point.

3. The method for detecting activated carbon filtration efficiency based on big data analysis according to claim 1, characterized in that: The adsorption characteristic values ​​of various filtration detection indicators in the first stage are obtained as follows: For the first stage of each high-dimensional point, the difference in adsorption speed between each moment and the adjacent moment is calculated as the speed change at each moment; the time interval between each moment and the end moment is obtained; the cumulative sum of the inverse proportional mapping results of the speed change at all moments and the corresponding time intervals is calculated to obtain the relative speed change in the first stage; the inverse proportional mapping results of the relative speed change of various filtering detection indicators in the first stage and the overall change value are used as the adsorption characteristic values ​​of various filtering detection indicators in the first stage.

4. The method for detecting activated carbon filtration efficiency based on big data analysis according to claim 1, characterized in that: The process of classifying high-dimensional points is specifically as follows: Based on the starting time data of any two high-dimensional points, combined with the adsorption characteristic degree, a characteristic distance between the any two high-dimensional points is obtained; Taking the preset characteristic distances between any high-dimensional point and the other high-dimensional points as the candidate radii of the any high-dimensional point, and taking the candidate radii in turn as the neighborhood radius of the any high-dimensional point; obtaining the total number of the other high-dimensional points in the neighborhood corresponding to any candidate radius and the degree of confusion of the corresponding characteristic distances; Counting the frequency of the total number of points in the neighborhood corresponding to any selected radius of any high-dimensional point in all neighborhoods of all high-dimensional points; Taking the maximum value of the frequency of any high-dimensional point and the fusion result of the disorder degree of all neighborhoods as the preferred value of any high-dimensional point based on any selected radius; The candidate radius with the largest preferred value is used as the neighborhood radius of the clustering algorithm, and the feature distance is used as the metric distance between high-dimensional points. The high-dimensional points are classified to obtain all categories.

5. The method for detecting activated carbon filtration efficiency based on big data analysis according to claim 4, characterized in that: The obtaining of the characteristic distance between any two high-dimensional points is specifically: The starting time data of various filtration detection indicators at each high-dimensional point of the activated carbon equipment and the corresponding adsorption characteristic degree are fused and recorded as the adsorption distance of various filtration detection indicators; the differences in all the adsorption distances of any two high-dimensional points are fused to obtain the characteristic distance of the any two high-dimensional points.

6. The method for detecting activated carbon filtration efficiency based on big data analysis according to claim 1, characterized in that: The step of predicting and judging the efficiency of activated carbon at the point to be filtered includes: Based on the categories after high-dimensional point classification, the predicted category of the point to be filtered is obtained by combining the data similarity between the point to be filtered and various filtering detection data indicators of each category of high-dimensional points; The overall changes of various filtration detection indicators of all high-dimensional points in the comprehensive prediction category are combined with the adsorption characteristic degree to obtain the predicted activated carbon filtration efficiency of the point to be filtered; When the predicted activated carbon filtration efficiency of the filter point to be detected is less than the preset filtration efficiency judgment threshold, it is judged that there is a problem with the filtration efficiency of the activated carbon equipment; otherwise, it is judged that there is no problem with the filtration efficiency of the activated carbon equipment.

7. The method for detecting activated carbon filtration efficiency based on big data analysis according to claim 6, characterized in that: The method of obtaining the predicted category of the point to be filtered is specifically: The difference between the starting time data of all kinds of filtering detection indicators of the point to be filtered and each high-dimensional point is merged to determine the initial distance between the point to be filtered and each high-dimensional point; For each category after high-dimensional points are classified, an inverse proportional mapping of the average value of the initial distance between the point to be filtered and all high-dimensional points in each category is obtained, which is recorded as the predicted optimal value of each category. The category with the largest predicted optimal value is selected as the predicted category of the point to be filtered.

8. The method for detecting activated carbon filtration efficiency based on big data analysis according to claim 7, characterized in that: The specific steps of obtaining the predicted activated carbon filtration efficiency of the point to be filtered include: Calculate the data difference of various filtering detection indicators at the start and end time of each high-dimensional point, and record it as the total change value of various filtering detection indicators at each high-dimensional point; calculate the total change value of all filtering detection indicators at each high-dimensional point and the inverse proportional mapping of the corresponding standard value to merge them as the filtering efficiency of each high-dimensional point; The normalized value of the inverse proportional mapping result between the adsorption characteristic degree of each high-dimensional point in the prediction category and the corresponding initial distance is obtained, and recorded as the reference weight of each high-dimensional point in the prediction category; the reference weight of all high-dimensional points in the prediction category is merged with the filtration efficiency to obtain the predicted activated carbon filtration efficiency of the point to be filtered.

Images