Method for detecting filtration performance of an agricultural field irrigation conduit filter

Abstract

The application relates to the technical field of machine learning, and discloses a filtering performance detection method for a farmland irrigation pipeline filter, which comprises the following steps: acquiring a time sequence of relevant parameters; acquiring a filtering feature sequence and adjacent data sets according to the time sequence of the relevant parameters; acquiring a clogging effect index according to the filtering feature sequence and the adjacent data sets; acquiring a joint clogging effect index according to the clogging effect index; acquiring a water quality feature sequence and a water quality change set according to the time sequence of the relevant parameters; acquiring a filtering water quality good index according to the water quality feature sequence and the water quality change set; acquiring a filtering performance anomaly index according to the joint clogging effect index and the filtering water quality good index; and acquiring a filtering performance detection data set according to the filtering performance anomaly index, and using an SVM classifier to acquire a performance detection result of the filter based on the filtering performance detection data set. The application improves the completeness of filtering performance evaluation and the accuracy of filtering performance detection.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and more specifically to a method for testing the filtration performance of filters used in farmland irrigation pipelines. Background Technology

[0002] With the modernization of agricultural production, water-saving irrigation methods have become a modern trend, with common methods including sprinkler irrigation and drip irrigation. To ensure the normal operation of irrigation equipment such as sprinkler and drip irrigation systems, farmland irrigation pipe filters have become an important component. However, due to the poor filtration performance of farmland irrigation pipe filters, blockages are easily caused in irrigation equipment, shortening its lifespan. Furthermore, when the filter's performance is poor, the filtered water contains more impurities, which can negatively impact the growth of crops.

[0003] Therefore, to improve the anti-clogging ability of irrigation equipment and the water purification capacity of filters, it is necessary to test the filtration performance of farmland irrigation pipeline filters in order to extend the service life of irrigation equipment and enable better growth of crops. Currently, most methods evaluate filter performance by measuring the uniformity and stability of the water flow rate after filtration. However, this does not fully demonstrate that changes in water flow rate are solely related to the filter; changes in electricity can also affect water flow rate. Therefore, evaluating filter performance solely based on water flow rate is inaccurate. Summary of the Invention

[0004] This invention provides a method for testing the filtration performance of filters used in farmland irrigation pipelines, in order to solve the problem of poor accuracy in filtration performance testing. The specific technical solution adopted is as follows:

[0005] One embodiment of the present invention provides a method for testing the filtration performance of a filter for farmland irrigation pipelines, the method comprising the following steps:

[0006] Obtain the time series of relevant parameters, including the time series of water flow velocity, water pressure, and water conductivity.

[0007] Based on the water flow velocity time series and water pressure time series, obtain the filtering feature sequence and adjacent data set of each data point at each acquisition time in the water flow velocity time series and water pressure time series, respectively; based on the filtering feature sequence and adjacent data set of each data point at each acquisition time in the water flow velocity time series and water pressure time series, obtain the clogging effect index of each data point at each acquisition time in the water flow velocity time series and water pressure time series, respectively; based on the clogging effect index of each data point at each acquisition time in the water flow velocity time series and water pressure time series, obtain the joint clogging effect index for each acquisition time.

[0008] Based on the water conductivity time series, obtain the water quality characteristic sequence and water quality change set of data points at each collection time in the water conductivity time series; obtain the filtration water quality excellence index at each collection time based on the water quality characteristic sequence and water quality change set of data points at each collection time in the water conductivity time series; obtain the filtration performance anomaly index at each collection time based on the filtration water quality excellence index and the combined clogging effect index at each collection time; obtain the filtration performance detection dataset based on the filtration performance anomaly index at all collection times.

[0009] An SVM classifier is used to obtain a detection model of filter performance based on a filter performance detection dataset, and the filter performance detection results are obtained based on the filter performance detection model.

[0010] Preferably, the method for obtaining the filtered feature sequence and adjacent data set of data points at each acquisition time in the water flow velocity time series and water pressure time series, respectively, based on the water flow velocity time series and water pressure time series, is as follows:

[0011] For each data point in the water flow velocity time series, the data point at the acquisition time is taken as the data point of the first target time. The sequence of the data points at the first preset parameter acquisition times closest to the first target time in ascending order of time is taken as the filtering feature sequence of the data points at the first target time. The data point at the acquisition time is taken as the data point of the first marked time. The set of the data points at the second preset parameter acquisition times closest to the first marked time is taken as the adjacent data set of the data points at the first marked time.

[0012] For each data point in the water pressure time series, the data point at the acquisition time is used as the data point of the second target time. The sequence of data points at the first preset parameter acquisition times closest to the second target time in ascending order of time is used as the filtering feature sequence of the data points of the second target time. The data point at the acquisition time is used as the data point of the second marker time. The set of data points at the second preset parameter acquisition times closest to the second marker time is used as the adjacent data set of the data points of the second marker time.

[0013] Preferably, the method for obtaining the blockage effect index of the data points at each acquisition time in the water flow velocity time series and water pressure time series based on the filtering feature sequence of the data points at each acquisition time and the adjacent data set is as follows:

[0014] Step S1: For each data point in the water flow velocity time series, calculate the coefficient of variation of all data points in the filtered feature sequence of the data points at the data points at the data points at the data points, and use the negative mapping result with the natural constant as the base and the coefficient of variation as the exponent as the numerator.

[0015] The absolute value of the difference between the value of the data point at the acquisition time and the value of each data point in the filtered feature sequence of the data point at the acquisition time is used as the first summing factor, and the sum of the first summing factor on the filtered feature sequence and the third preset parameter is used as the denominator.

[0016] The sum of the opposite of the ratio of the numerator to the denominator and the third preset parameter is used as the filtering anomaly index of the data point at the time of acquisition.

[0017] Step S2: Obtain the anti-blocking capability index of the data points at the time of acquisition based on the filtering feature sequence of the data points at the acquisition time and the adjacent data set;

[0018] Step S3: The negative mapping result with the natural constant as the base and the anti-clogging ability index of the data point at the time of collection as the index is used as the first product factor. The product of the first product factor and the filtration anomaly index is used as the clogging effect index of the data point at the time of collection in the water flow velocity time series.

[0019] Step S4: Replace the water flow velocity time series in step S1 with the water pressure time series, and repeat the calculations of steps S1, S2 and S3 to obtain the blockage effect index of the data points at the time of collection in the water pressure time series.

[0020] Preferably, the method for obtaining the anti-blocking capability index of the data points at the acquisition time based on the filtered feature sequence of the data points at the acquisition time and the adjacent data set is as follows:

[0021] The mean of all data points in the filtered feature sequence of the data points at the time of acquisition is used as the numerator;

[0022] Calculate the metric distance between the filtered feature sequence of the data point at the acquisition time and the filtered feature sequence of each data point in the adjacent data set of the data point at the acquisition time, and use the sum of the metric distance and the third preset parameter as the denominator;

[0023] The average of the sums of the ratios of the numerator to the denominator over the adjacent data sets is used as the anti-blocking capability index of the data point at the time of acquisition.

[0024] Preferably, the method for obtaining the joint blockage effect index at each sampling time based on the blockage effect index of each data point in the water flow velocity time series and water pressure time series is as follows:

[0025] The blockage effect index of each data point in the water flow velocity time series and water pressure time series is used as the input of the Kalman filter-based data fusion algorithm, and the output of the Kalman filter-based data fusion algorithm is used as the joint blockage effect index at each data collection time.

[0026] Preferably, the method for obtaining the water quality characteristic sequence and water quality change set of data points at each collection time in the water conductivity time series is as follows:

[0027] The water conductivity time series is used as the input of the Pettitt mutation detection algorithm, the output of the Pettitt mutation detection algorithm is used as all mutation data points in the water conductivity time series, and the sequence composed of all mutation data points in the water conductivity time series is used as the water quality mutation sequence.

[0028] For each data point in the water conductivity time series, calculate the difference between the data point at the data point at the data point and each element in the water quality mutation sequence, and use the sequence of all the differences in ascending order as the water quality characteristic sequence of the data point at the data point at the data point.

[0029] The data points at each collection time in the water conductivity time series are used as the data points at the third marked time. The data points at all collection times in the water conductivity time series are used as the input of the K-nearest neighbor algorithm. The output of the K-nearest neighbor algorithm is used as all the nearest data points of the data points at the third marked time. The set of all the nearest data points of the data points at the third marked time is used as the set of water quality changes of the data points at the third marked time.

[0030] Preferably, the method for obtaining the filtered water quality excellence index at each sampling time based on the water quality characteristic sequence and water quality change set of data points at each sampling time in the water conductivity time series is as follows:

[0031] The filtration performance stability index for each sampling moment is obtained based on the water quality characteristic sequence and water quality change set of each data point in the water conductivity time series.

[0032] For each data point in the water conductivity time series, a negative mapping result is calculated with the natural constant as the base and the value of each data point in the water quality change set as the exponent. The product of the cumulative sum of the negative mapping result on the water quality change set of the data points and the filtration performance stability index is used as the filtration water quality excellence index at the time of collection.

[0033] Preferably, the method for obtaining the filtration performance stability index at each sampling time based on the water quality characteristic sequence and water quality change set of data points at each sampling time in the water conductivity time series is as follows:

[0034] For each data point in the water conductivity time series, the numerator will be the negative mapping result with the natural constant as the base and the variance of all data points in the water quality change set as the exponent.

[0035] Calculate the metric distance between the water quality feature sequence of the data points and the water quality feature sequence of each data point in the water quality change set of the data points, and use the sum of the metric distances on the water quality change set of the data points and the sum of the third preset parameter as the denominator;

[0036] The ratio of the numerator to the denominator is used as the stability index of the filtration performance at the time of data collection.

[0037] Preferably, the method for obtaining the filtration performance anomaly index at each sampling time based on the filtration water quality excellence index and the combined clogging effect index at each sampling time, and for obtaining the filtration performance test dataset based on the filtration performance anomaly indices at all sampling times, is as follows:

[0038] For each data collection time, the combined clogging effect index at that time is used as the numerator, the excellent filtration water quality index at that time is used as the denominator, the ratio of the numerator to the denominator is used as the abnormal filtration performance index at that time, and the set of abnormal filtration performance indices at all data collection times is used as the filtration performance test dataset.

[0039] Preferably, the method for obtaining the filter performance detection model based on the filter performance detection dataset using an SVM classifier, and obtaining the filter performance detection result based on the filter performance detection model, is as follows:

[0040] The segmentation threshold of the filtering performance test dataset is obtained by using the maximum inter-class variance algorithm. Data in the filtering performance test dataset that is below the segmentation threshold is marked as 0, and data in the filtering performance test dataset that is above the segmentation threshold is marked as 1. The labeling results of the data in the filtering performance test dataset are used as the label data of the filtering performance test dataset.

[0041] The SVM classifier model is trained using the labeled data of the filter performance detection dataset to obtain the trained model, which is then used as the detection model for filter performance.

[0042] The abnormal filtration performance index at each moment during the actual operation of the filter is used as the input to the filter performance detection model, and the output of the filter performance detection model is used as the filter performance detection result at each moment during the actual operation.

[0043] The beneficial effects of this invention are as follows: This invention obtains a filtration anomaly index and an anti-clogging ability index based on water flow velocity time series and water pressure time series; it obtains a clogging effect index based on the filtration anomaly index and the anti-clogging ability index; it obtains a combined clogging effect index based on the clogging effect index; simultaneously, it obtains a filtration performance stability index based on water conductivity time series; it obtains a filtered water quality excellence index based on the filtration performance stability index; and it obtains a filtration performance anomaly index based on the combined clogging effect index and the filtered water quality excellence index. A filtration performance testing dataset is obtained based on the filtration performance anomaly index, and an SVM classifier is used to obtain the filter performance testing results for evaluating the filtration performance of farmland irrigation pipeline filters. Its beneficial effect lies in combining the clogging phenomenon observed in farmland irrigation pipeline filters with the quality of the filtered water source, and then using an SVM classifier to obtain the filter performance testing results, making the filter performance evaluation more complete and improving the accuracy of filtration performance testing for farmland irrigation pipeline filters. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a schematic flowchart of a method for testing the filtration performance of a filter for farmland irrigation pipelines according to an embodiment of the present invention.

[0046] Figure 2 This is a flowchart illustrating the implementation of a method for testing the filtration performance of a filter used in farmland irrigation pipelines, as provided in one embodiment of the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Please see Figure 1 The diagram illustrates a flowchart of a method for testing the filtration performance of a filter for farmland irrigation pipelines according to an embodiment of the present invention. The method includes the following steps:

[0049] Step S001: Obtain the time series of relevant parameters that affect the filtration performance of the filter.

[0050] The purpose of this invention is to test the filtration performance of a filter in a farmland irrigation pipeline. This involves using multiple sensors to collect relevant parameters affecting the filter's performance, including a water conductivity sensor, a water flow sensor, and a water pressure sensor. The relevant parameters affecting the filter's performance include the water conductivity of the water source at the outlet of the farmland irrigation pipeline, as well as the water flow velocity and water pressure within the outlet pipeline.

[0051] The conductivity of the water source at the outlet pipe of the farmland irrigation pipeline is collected using a water conductivity sensor. At the same time, the water flow velocity and water pressure of the water source in the outlet pipe are collected using a water flow sensor and a water pressure sensor. The collection time interval is 10 seconds, and the collection time is 3 hours. The implementer can select the values ​​according to the actual situation.

[0052] For each relevant parameter affecting the filtration performance of the filter, a sequence of relevant parameter data collected at all times, arranged in ascending order of time, is used as the time series of the relevant parameter. To avoid the influence of different data units on the filtration performance analysis, Z-score normalization is performed on all data within the time series of each relevant parameter, resulting in the normalized time series of each relevant parameter. The normalized time series of each relevant parameter includes the time series of water conductivity, water flow velocity, and water pressure. Z-score normalization is a well-known technique and will not be elaborated further.

[0053] Thus, the time series of each relevant parameter after normalization is obtained.

[0054] Step S002: Obtain the filtration feature sequence and adjacent data set based on the water flow velocity time series and water pressure time series; obtain the clogging effect index based on the filtration feature sequence and adjacent data set; and obtain the joint clogging effect index based on the clogging effect index.

[0055] Generally, during the use of farmland irrigation pipeline filters, when the filter becomes clogged, significant changes occur in water flow rate and pressure, with some becoming abnormally low. This indicates poor filter performance. To identify filter clogging in farmland irrigation, it is necessary to analyze the time series of water flow rate and water pressure. The implementation flowchart of this invention is as follows: Figure 2 As shown.

[0056] Specifically, to more accurately analyze the clogging phenomenon caused by the filter, each data point in the water flow velocity time series is taken as the data point of the first target time. The sequence of data points from the m closest data points to the first target time in ascending order of time is taken as the filtering feature sequence of the data points of the first target time. The empirical value of m is 30. Each data point in the water flow velocity time series is taken as the data point of the first marker time. The set of data points from the n closest data points to the first marker time is taken as the neighbor set of the data points of the first marker time. The empirical value of n is 2.

[0057] Similarly, each data point at each acquisition time in the water pressure time series is used as the data point of the second target time. The sequence of data points at the m acquisition times closest to the second target time, arranged in ascending order of time, is used as the filtering feature sequence of the data points at the second target time. Each data point at each acquisition time in the water pressure time series is used as the data point of the second marked time. The set of data points at the n acquisition times closest to the second marked time is used as the neighbor set of the data points at the second marked time.

[0058] Thus, the filtered feature sequence and adjacent data set for each data point in the water flow velocity time series and water pressure time series are obtained respectively.

[0059] It should be noted that the greater the change in water flow velocity, and the lower the water flow velocity, the more likely the filter in the farmland irrigation pipe will become clogged. Similarly, the greater the change in water pressure, and the lower the water pressure, the more likely the filter in the farmland irrigation pipe will become clogged. Since the principle by which changes in water flow velocity and water pressure reflect filter clogging is the same, the analysis of clogging phenomena is based on the time series of water flow velocity, and the time series of water pressure is processed in the same way.

[0060] Calculate the blockage effect index for each data point in the water flow velocity time series:

[0061]

[0062]

[0063] D i =A i *exp(-C i )

[0064] In the formula, A i The filter anomaly index represents the data point at the i-th acquisition time in the water flow velocity time series, where exp() represents the exponential function with base to the natural constant, μ. iLet a represent the coefficient of variation of all data points in the filtered feature sequence of the data point at the i-th acquisition time in the water flow velocity time series, m represent the number of data points in the filtered feature sequence of the data point at the i-th acquisition time in the water flow velocity time series, and a represent the coefficient of variation of all data points in the filtered feature sequence of the data point at the i-th acquisition time in the water flow velocity time series. i This represents the value of the data point at the i-th sampling time in the water flow velocity time series. C represents the value of the j-th data point in the filtered feature sequence of the data point at the i-th acquisition time in the water flow velocity time series; i β represents the anti-clogging index of the data point at the i-th acquisition time in the water flow velocity time series, where n represents the number of data points in the adjacent data set of the data point at the i-th acquisition time in the water flow velocity time series. i The mean of all data points in the filtered feature sequence of the data point at the i-th acquisition time in the water flow velocity time series is represented by dtw(), which represents the dtw distance function. i This represents the filtered feature sequence of the data point at the i-th acquisition time in the water flow velocity time series. This represents the filtered feature sequence of the b-th data point in the neighboring dataset of the data point at the i-th acquisition time in the water flow velocity time series. The dtw distance represents the distance between the filtered feature sequence of the data point at the i-th acquisition time in the water flow velocity time series and the filtered feature sequence of the b-th data point in its neighboring dataset; D i This represents the blockage effect index at the i-th data collection time in the water flow velocity time series. It should be noted that the calculation of the coefficient of variation is a well-known technique and will not be elaborated upon further.

[0065] The coefficient of variation μ of all data points in the filtered feature sequence of the data point at the i-th acquisition time in the water flow velocity time series i The larger the value, and the greater the difference between the value of the data point at the i-th acquisition time in the water flow velocity time series and the value of the j-th data point in its filtered feature sequence, the greater the first summation factor. The larger the value, the greater the change in water flow velocity, reflecting abnormal changes in water flow to some extent. This increases the likelihood that a filter malfunction is causing the significant change in water flow velocity, thus a higher filtration anomaly index. Additionally, the dtw distance between the filter feature sequence of the data point at the i-th acquisition time in the water flow velocity time series and the filter feature sequence of the b-th data point in its neighboring dataset is also considered. The larger the value, and the greater the mean β of all data points in the filtering feature sequence of the data point at the i-th acquisition time in the water flow velocity time series. iThe smaller the value, the lower the similarity between the filtering feature sequences of data points in adjacent datasets, the greater the change in water flow velocity, and the lower the water flow velocity, the more likely the farmland irrigation pipe filter is to become clogged, thus the lower the anti-clogging ability index. Therefore, the filtering anomaly index A of the data point at the i-th acquisition time in the water flow velocity time series is... i The larger the value, the greater the anti-clogging index C of the data point at the i-th acquisition time in the water flow velocity time series. i The smaller the first product factor exp(-C) i The larger the value, the more obvious the abnormal phenomenon of the filter, that is, the more likely the filter is to become clogged, and the larger the clogging effect index is.

[0066] Similarly, by applying the above calculation method to the water pressure time series, we can obtain the filtration anomaly index, anti-clogging capacity index, and clogging effect index for each data point in the water pressure time series. It should be noted that a larger water pressure change reflects an abnormal change in water pressure within the pipes, and is more likely due to a filter malfunction causing the large pressure change; therefore, the filtration anomaly index for each data point in the water pressure time series will be larger. Conversely, a larger water pressure change, especially with lower water pressure, indicates a greater likelihood of clogging in the agricultural irrigation pipe filter; therefore, the anti-clogging capacity index for each data point in the water pressure time series will be smaller. Furthermore, a larger filtration anomaly index and a smaller anti-clogging capacity index for each data point in the water pressure time series indicate a more pronounced filter malfunction, meaning the filter is more likely to clog; therefore, the clogging effect index for each data point in the water pressure time series will be larger.

[0067] Furthermore, a data fusion algorithm based on Kalman filtering is used to obtain the joint blockage effect index at each acquisition time. The blockage effect index of each data point in the water flow velocity time series and water pressure time series is used as the input of the data fusion algorithm based on Kalman filtering, and the output of the data fusion algorithm based on Kalman filtering is used as the joint blockage effect index at each acquisition time. The data fusion algorithm based on Kalman filtering is a well-known technology and will not be described in detail.

[0068] Thus, the joint congestion effect index at each acquisition moment is obtained.

[0069] Step S003: Obtain water quality characteristic sequence and water quality change set based on water conductivity time series; obtain excellent filtration water quality index based on water quality characteristic sequence and water quality change set; obtain abnormal filtration performance index based on combined clogging effect index and excellent filtration water quality index.

[0070] Furthermore, from the perspective of water conductivity, the effectiveness of the filter in filtering water sources is analyzed. Generally, the better the filtration performance of a farmland irrigation pipeline filter, the purer the filtered water and the lower its conductivity; conversely, the worse the filtration performance, the more impurities are present in the filtered water and the higher its conductivity. Because of the varying quality of filters, filtration often leads to changes in conductivity. A greater change in conductivity indicates less stable filtration performance, and a higher conductivity in the filtered water indicates poorer filtration performance.

[0071] Specifically, to clearly analyze the quality of filtration performance, the water conductivity time series is used as the input to the Pettitt mutation detection algorithm, and the output of the Pettitt mutation detection algorithm is used as all mutation data points in the water conductivity time series. The sequence composed of all mutation data points in the water conductivity time series is taken as the water quality mutation sequence. The Pettitt mutation detection algorithm is a well-known technology and will not be elaborated further. For each data point in the water conductivity time series at each acquisition time, the difference between the data point and each element in the water quality mutation sequence is calculated. The sequence composed of all the differences in ascending order is taken as the water quality feature sequence of the data point at the acquisition time.

[0072] Furthermore, in order to clearly reflect the changes in water quality, the data points at each collection time in the water conductivity time series are used as the data points at the third marked time. The data points at all collection times in the water conductivity time series are used as the input of the K-nearest neighbor algorithm. The empirical value of the neighborhood parameter K is 10. The output of the K-nearest neighbor algorithm is used as all the nearest data points of the data points at the third marked time. The set of all the nearest data points of the data points at the third marked time is used as the set of water quality changes of the data points at the third marked time.

[0073] Calculate the water quality excellence index at each sampling time:

[0074]

[0075]

[0076] In the formula, G i S represents the stability index of the filtering performance at the i-th acquisition time, exp() represents the exponential function with the natural constant as the base, and S i Let K represent the variance of all data points in the set of water quality changes at the i-th sampling time point in the water conductivity time series, K represent the number of data points in the set of water quality changes at the i-th sampling time point in the water conductivity time series, dtw() represent the dtw distance function, and f if represents the water quality characteristic sequence of the data point at the i-th acquisition time in the water conductivity time series. i g dtw(f) represents the water quality characteristic sequence of the g-th data point in the set of water quality changes at the i-th data point in the water quality conductivity time series. i ,f i g H represents the dtw distance between the water quality characteristic sequence of the data point at the i-th acquisition time in the water conductivity time series and the water quality characteristic sequence of the g-th data point in its water quality change set; i This represents the water quality excellence index at the i-th sampling time. This represents the value of the g-th data point in the set of water quality changes at the i-th data point in the water conductivity time series.

[0077] The variance S of all data points in the set of water quality changes at the i-th data collection time in the water conductivity time series is... i The smaller the value, the greater the dtw distance dtw(f) between the water quality characteristic sequence of the data point at the i-th collection time in the water conductivity time series and the water quality characteristic sequence of the g-th data point in its water quality change set. i ,f i g The smaller the value of G, the lower the dispersion of data points in the water quality change set and the higher the similarity between water quality characteristic sequences. This reflects, to some extent, a smaller change in water conductivity, meaning the more stable the filter's performance, the larger the filtration performance stability index. Additionally, the filtration performance stability index G at the i-th sampling time... i The larger the value of the filter, and the smaller the value of the g-th data point in the set of water quality changes at the i-th collection time in the water conductivity time series, the smaller the conductivity, which means the purer the filtered water source and the better the filtration performance of the filter. Therefore, the greater the water quality excellence index.

[0078] Furthermore, by combining the combined clogging effect index and the excellent filtration water quality index, the abnormal filtration performance index at each sampling time was calculated:

[0079]

[0080] In the formula, V i E represents the filtering performance anomaly index at the i-th acquisition time. i H represents the joint congestion effect index at the i-th acquisition time. i This represents the water quality quality index at the i-th sampling time.

[0081] The joint blockage effect index E at the i-th acquisition time point iThe larger the value, the better the filtered water quality index H at the i-th sampling time. i The smaller the value, the more obvious the clogging of the filter, and the lower the quality of the filtered water. In other words, the greater the possibility of abnormal filtration performance, the higher the filtration performance abnormality index.

[0082] Thus, the filtering performance anomaly index for each data acquisition moment is obtained.

[0083] Step S004: Obtain the filter performance detection dataset based on the filter performance anomaly index, obtain the filter performance detection model based on the filter performance detection dataset, and obtain the filter performance detection result using the filter performance detection model.

[0084] The set of filtering performance anomaly indices collected at all times is defined as the filtering performance detection dataset. This dataset contains two main categories: one representing poor filtering performance and the other representing good filtering performance. To differentiate the data in the filtering performance detection dataset, it is used as the input to the Otsu's algorithm (Maximum Inter-Class Variance). The output of the Otsu's algorithm is used as the segmentation threshold for the filtering performance detection dataset. The Otsu's algorithm is a well-known technique and will not be elaborated upon further.

[0085] Furthermore, the data in the filtering performance test dataset are labeled to obtain the labeled data of the filtering performance test dataset. Data in the filtering performance test dataset that is below the segmentation threshold is marked as 0, representing data with poor filtering performance; data in the filtering performance test dataset that is above the segmentation threshold is marked as 1, representing data with good filtering performance.

[0086] The SVM classifier is a supervised machine learning model that can effectively handle binary classification problems in practical applications. The SVM classifier model is trained using labeled data from a filter performance detection dataset. The trained model is then used as the filter performance detection model. Model training is a well-known technique, and the specific process will not be elaborated further. Next, following the above procedure, the filter performance anomaly index at each moment during the actual operation of the farmland irrigation pipeline filter is obtained as the input to the trained detection model. The output of the detection model is then used as the filter performance detection result at each moment during actual operation.

[0087] This concludes the method for testing the filtration performance of filters used in farmland irrigation pipelines.

[0088] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. The above descriptions are merely preferred embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for testing the filtration performance of filters used in farmland irrigation pipelines, characterized in that, The method includes the following steps: Obtain the time series of relevant parameters, including the time series of water flow velocity, water pressure, and water conductivity. Based on the water flow velocity time series and water pressure time series, obtain the filtering feature sequence and adjacent data set of each data point at each acquisition time in the water flow velocity time series and water pressure time series, respectively; based on the filtering feature sequence and adjacent data set of each data point at each acquisition time in the water flow velocity time series and water pressure time series, obtain the clogging effect index of each data point at each acquisition time in the water flow velocity time series and water pressure time series, respectively; based on the clogging effect index of each data point at each acquisition time in the water flow velocity time series and water pressure time series, obtain the joint clogging effect index for each acquisition time. Based on the water conductivity time series, obtain the water quality characteristic sequence and water quality change set of data points at each collection time in the water conductivity time series; obtain the filtration water quality excellence index at each collection time based on the water quality characteristic sequence and water quality change set of data points at each collection time in the water conductivity time series; obtain the filtration performance anomaly index at each collection time based on the filtration water quality excellence index and the combined clogging effect index at each collection time; obtain the filtration performance detection dataset based on the filtration performance anomaly index at all collection times. An SVM classifier is used to obtain a detection model of filter performance based on a filter performance detection dataset, and the filter performance detection results are obtained based on the filter performance detection model. The method for obtaining the blockage effect index of each data point in the water flow velocity time series and water pressure time series based on the filtered feature sequence and adjacent data sets of each data point at each acquisition time is as follows: Step S1: For each data point in the water flow velocity time series, calculate the coefficient of variation of all data points in the filtered feature sequence of the data points at the data points at the data points at the data points, and use the negative mapping result with the natural constant as the base and the coefficient of variation as the exponent as the numerator. The absolute value of the difference between the value of the data point at the acquisition time and the value of each data point in the filtered feature sequence of the data point at the acquisition time is used as the first summing factor, and the sum of the first summing factor on the filtered feature sequence and the constant 1 is used as the denominator. The sum of the opposite of the ratio of the numerator to the denominator and the constant 1 is used as the filtering anomaly index of the data point at the time of acquisition. Step S2: Obtain the anti-blocking capability index of the data points at the time of acquisition based on the filtered feature sequence of the data points at the acquisition time and the adjacent data set; Step S3: The negative mapping result with the natural constant as the base and the anti-clogging ability index of the data point at the time of collection as the index is used as the first product factor. The product of the first product factor and the filtration anomaly index is used as the clogging effect index of the data point at the time of collection in the water flow velocity time series. Step S4: Replace the water flow velocity time series in step S1 with the water pressure time series, and repeat the calculations of steps S1, S2 and S3 to obtain the blockage effect index of the data points at the collection time in the water pressure time series. The method for obtaining the anti-blocking capability index of data points at the time of acquisition based on the filtered feature sequence of data points at the time of acquisition and adjacent data sets is as follows: The mean of all data points in the filtered feature sequence of the data points at the time of acquisition is used as the numerator; Calculate the metric distance between the filtered feature sequence of the data point at the acquisition time and the filtered feature sequence of each data point in the adjacent data set of the data point at the acquisition time, and use the sum of the metric distance and the constant 1 as the denominator; The average of the sums of the ratios of the numerator to the denominator over the adjacent data sets is used as the anti-blocking capability index of the data point at the time of acquisition. The method for obtaining the joint blockage effect index at each sampling time based on the blockage effect index of each data point in the water flow velocity time series and water pressure time series is as follows: The blockage effect index of each data point in the water flow velocity time series and water pressure time series is used as the input of the Kalman filter-based data fusion algorithm, and the output of the Kalman filter-based data fusion algorithm is used as the joint blockage effect index at each data collection time. The method for obtaining the filtered water quality excellence index at each sampling time based on the water quality characteristic sequence and water quality change set of each data point in the water conductivity time series is as follows: The filtration performance stability index for each sampling moment is obtained based on the water quality characteristic sequence and water quality change set of each data point in the water conductivity time series. For each data point in the water conductivity time series, a negative mapping result is calculated with the natural constant as the base and the value of each data point in the water quality change set as the exponent. The product of the sum of the negative mapping result on the water quality change set of the data points and the filtration performance stability index is used as the filtration water quality excellence index at the time of collection. The method for obtaining the filtration performance stability index at each sampling time based on the water quality characteristic sequence and water quality change set of each data point in the water conductivity time series is as follows: For each data point in the water conductivity time series, the numerator will be the negative mapping result with the natural constant as the base and the variance of all data points in the water quality change set as the exponent. Calculate the metric distance between the water quality feature sequence of the data points and the water quality feature sequence of each data point in the water quality change set of the data points, and use the sum of the metric distances on the water quality change set of the data points and the constant 1 as the denominator; The ratio of the numerator to the denominator is used as the stability index of the filtration performance at the time of data collection. The method for obtaining the filtration performance abnormality index at each sampling time based on the filtration water quality excellence index and the combined clogging effect index at each sampling time is as follows: For each sampling time, the combined clogging effect index at the sampling time is used as the numerator, the excellent filtration water quality index at the sampling time is used as the denominator, and the ratio of the numerator to the denominator is used as the abnormal filtration performance index at the sampling time.

2. The method for testing the filtration performance of a filter for farmland irrigation pipelines according to claim 1, characterized in that, The method for obtaining the filtered feature sequence and adjacent data set of data points at each acquisition time in the water flow velocity time series and water pressure time series, respectively, is as follows: For each data point in the water flow velocity time series, the data point at the time of collection is used as the data point of the first target time. The sequence of data points at the first preset parameter times closest to the first target time in ascending order of time is used as the filtering feature sequence of the data points of the first target time. The data points at the acquisition time are used as the data points at the first marked time, and the set of data points at the second preset parameter acquisition times that are closest to the first marked time is used as the adjacent data set of the data points at the first marked time. For each data point in the water pressure time series, the data point at the acquisition time is used as the data point of the second target time. The sequence of data points at the first preset parameter acquisition times closest to the second target time in ascending order of time is used as the filtering feature sequence of the data points of the second target time. The data points at the acquisition time are used as the data points at the second marked time, and the set of data points at the second preset parameter acquisition times that are closest to the second marked time is used as the adjacent data set of the data points at the second marked time.

3. The method for testing the filtration performance of a filter for farmland irrigation pipelines according to claim 1, characterized in that, The method for obtaining the water quality characteristic sequence and water quality change set of each data point in the water conductivity time series based on the water conductivity time series is as follows: The water conductivity time series is used as the input of the Pettitt mutation detection algorithm, the output of the Pettitt mutation detection algorithm is used as all mutation data points in the water conductivity time series, and the sequence composed of all mutation data points in the water conductivity time series is used as the water quality mutation sequence. For each data point in the water conductivity time series, calculate the difference between the data point at the data point at the data point and each element in the water quality mutation sequence, and use the sequence of all the differences in ascending order as the water quality characteristic sequence of the data point at the data point at the data point. The data points at each collection time in the water conductivity time series are used as the data points at the third marked time. The data points at all collection times in the water conductivity time series are used as the input of the K-nearest neighbor algorithm. The output of the K-nearest neighbor algorithm is used as all the nearest data points of the data points at the third marked time. The set of all the nearest data points of the data points at the third marked time is used as the set of water quality changes of the data points at the third marked time.

4. The method for testing the filtration performance of a filter for farmland irrigation pipelines according to claim 1, characterized in that, The method for obtaining the filtering performance detection dataset based on the filtering performance anomaly index at all collection times is as follows: The set of filtering performance anomaly indices at all collection times is used as the filtering performance detection dataset.

5. The method for testing the filtration performance of a filter for farmland irrigation pipelines according to claim 1, characterized in that, The method for obtaining the filter performance detection model based on the filter performance detection dataset using an SVM classifier, and obtaining the filter performance detection result based on the filter performance detection model, is as follows: The segmentation threshold of the filtering performance test dataset is obtained by using the maximum inter-class variance algorithm. Data in the filtering performance test dataset that is below the segmentation threshold is marked as 0, and data in the filtering performance test dataset that is above the segmentation threshold is marked as 1. The labeling results of the data in the filtering performance test dataset are used as the label data of the filtering performance test dataset. The SVM classifier model is trained using the labeled data of the filter performance detection dataset to obtain the trained model, which is then used as the detection model for filter performance. The abnormal filtration performance index at each moment during the actual operation of the filter is used as the input to the filter performance detection model, and the output of the filter performance detection model is used as the filter performance detection result at each moment during the actual operation.

Images