A real-time monitoring method for the water quality of discharge ponds in petrochemical industry

Abstract

The present invention relates to the technical field of data processing, and particularly relates to a real-time monitoring method for the water quality of an emission pond for petrochemical industry, including: collecting water quality conductivity data, pH value data, and the minimum positive operation period of the wastewater treatment system in the emission pond, calculating the local outliers of each conductivity data and pH value data, thereby obtaining the correlation difference values of all conductivity data, combining the differences in the correlation difference values of the conductivity data at the same position in other periods, obtaining the noise performance degree of each conductivity data, and further obtaining the prediction weight of each conductivity data; obtaining the predicted value of each conductivity data through the prediction weight of each conductivity data, and completing the real-time monitoring of the water quality of the emission pond for petrochemical industry. The present invention determines the prediction weight by analyzing the water quality conductivity data, avoids the interference of noise data, and accurately monitors the water quality of the drainage pond for petrochemical industry in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to a real-time monitoring method for the water quality of an emission pool for petrochemical industry. Background Art

[0002] Petrochemical emission pools may contain various chemical substances and pollutants, including organic substances, heavy metals, salts, etc. Real-time monitoring can help detect the excessive emission or abnormal conditions of pollutants in a timely manner, protect the surrounding environment from pollution, and ensure that the emissions meet the requirements and standards of laws and regulations. The water conductivity can reflect the electrolyte content in the water. By real-time monitoring the change of conductivity, the changing trend of water pollution degree can be evaluated, providing accurate data support for formulating pollution prevention and control measures. Currently, real-time monitoring of abnormal data of water conductivity in petrochemical emission pools is often carried out, and the water pollution situation in the petrochemical emission pool is reflected through the abnormal data of water conductivity, and then corresponding adaptive measures are taken.

[0003] The prior art often uses the AR (Autoregression) model to perform real-time prediction on the water conductivity data at the most recent moment when no event has occurred. The abnormal water conductivity data is judged by the difference between the predicted value and the actual value at the same moment, and then the real-time monitoring of water quality is realized through the abnormal water conductivity data. In the traditional AR prediction model, the prediction weights of each historical data are determined by artificial experience values. Due to the influence of noise in the historical data, the inaccuracy of the prediction weights will be caused, resulting in a low prediction accuracy of the predicted value and unable to accurately detect the water quality situation in real time. Summary of the Invention

[0004] The present invention provides a real-time monitoring method for the water quality of an emission pool for petrochemical industry to solve the existing problems.

[0005] The real-time monitoring method for the water quality of an emission pool for petrochemical industry of the present invention adopts the following technical solutions:

[0006] An embodiment of the present invention provides a real-time monitoring method for the water quality of an emission pool for petrochemical industry, and the method includes the following steps:

[0007] Collect water conductivity data and pH value data, and each collection moment corresponds to a conductivity data and a pH value data; obtain the minimum positive operation period of the wastewater treatment system of the emission pool;

[0008] Obtain the local outliers of each conductivity data based on the difference between each conductivity data and the surrounding conductivity data; obtain the local outliers of each pH value data; construct a local anomaly ratio curve according to the local anomaly ratio of each conductivity data and the pH value data at the corresponding acquisition time, and segment the local anomaly ratio curve to obtain several segmented data; obtain the correlation difference value of all conductivity data according to the local anomaly ratio difference and data correlation between the conductivity data of each segmented data and the pH value data at the corresponding acquisition time; obtain the noise performance degree of each conductivity data according to the correlation difference value of the minimum running positive period where each conductivity data is located and the difference between the correlation difference values of the conductivity data at the same position in other periods.

[0009] Take the time interval between each conductivity data and the prediction time as the weight, and combine the noise performance degree of each conductivity data to obtain the prediction weight of each conductivity data; obtain the predicted value of the conductivity data at each prediction time through the prediction weight of each conductivity data; perform real-time monitoring on the water quality of the emission pool in petrochemical industry according to the predicted value of the conductivity data.

[0010] Further, obtaining the local outliers of each conductivity data according to the difference between each conductivity data and the surrounding conductivity data includes the following specific formula:

[0011]

[0012] In the formula, represents the local outlier of the i-th data in the conductivity data; represents the conductivity value of the i-th data in the conductivity data; represents the mean value of the conductivity values of all conductivity data; represents the number of neighboring data of the i-th data in the conductivity data; represents the conductivity value of the j-th neighboring data of the i-th data in the conductivity data; represents the standard deviation between the i-th data and its neighboring data in the conductivity data; represents the linear normalization function, represents the absolute value function.

[0013] Further, the method for obtaining neighboring data includes the following specific steps:

[0014] Construct a neighborhood centered on the i-th data in the conductivity data, and the conductivity data points within the neighborhood are recorded as the neighboring data of the i-th data.

[0015] Further, the steps of constructing a local anomaly ratio curve based on the local anomaly ratio of each conductivity data and the pH value data at the corresponding acquisition time, and segmenting the local anomaly ratio curve to obtain a number of segmented data are as follows:

[0016] Denote the ratio of the local anomaly value of each conductivity data to the local anomaly value of the pH value data at the corresponding acquisition time as the local anomaly ratio. Use all local anomaly ratios to construct a curve space, with the conductivity data as the horizontal axis and the value of the local anomaly ratio as the vertical axis in the curve space. Project all local anomaly ratios into the curve space to obtain the local anomaly ratio curve;

[0017] Segment the local anomaly ratio curve according to the acquisition time on average to obtain a number of segmented data.

[0018] Further, the steps of obtaining the correlation difference value of all conductivity data according to the local anomaly ratio difference and data correlation between the conductivity data and the pH value data at the corresponding acquisition time of each segmented data include the following specific formula:

[0019]

[0020] In the formula, represents the correlation difference value of the i-th data in the conductivity data; represents the i-th local anomaly ratio in the local anomaly ratio curve; represents the mean value of all local anomaly ratios in the local anomaly ratio curve; represents the number of segments of the local anomaly ratio curve; represents the segmented data where the i-th local anomaly ratio in the local anomaly ratio curve is located; represents the x-th segmented data in the local anomaly ratio curve except the segmented data where the i-th local anomaly ratio is located; is the exponential function with the natural constant as the base; represents the function for calculating the Pearson correlation coefficient, represents the absolute value function.

[0021] Further, the steps of obtaining the noise performance degree of each conductivity data according to the correlation difference value of the minimum operating positive period where each conductivity data is located and the difference between the correlation difference values of the conductivity data at the same position in other periods include the following specific formula:

[0022]

[0023] In the formula, represents the noise performance degree of the i-th data in the conductivity data; is the first parameter, is the second parameter; Represents the normalization function.

[0024] Furthermore, the specific steps for obtaining the first parameter and the second parameter are as follows:

[0025] First parameter The calculation method is:

[0026]

[0027] Second parameter The calculation method is:

[0028]

[0029] Where, Represents the correlation difference value of the i-th data in the conductivity data; Indicates the number of all minimum operating cycles; It represents the correlation difference value of the conductivity data in the yth period except the minimum operation period of the i-th data, which is at the same collection time as the minimum operation period of the i-th data; Indicates the maximum value of the correlation difference value of the i-th data in the conductivity data in the minimum operating cycle; Indicates the maximum value of the correlation difference value in the y-th period of the conductivity data except the minimum operating period of the i-th data. Represents the absolute value function.

[0030] Furthermore, the time interval between each conductivity data and the prediction time is used as a weight, and the prediction weight of each conductivity data is obtained in combination with the noise performance level of each conductivity data. The specific formula included is as follows:

[0031]

[0032] Where, Represents the prediction weight of the i-th data in the conductivity data; represents the initial weight; Indicates the noise performance of the i-th conductivity data; Represents the time interval between the i-th data in the conductivity data and the prediction time; Represents the normalization function.

[0033] Furthermore, the predicted value of the conductivity data at each prediction moment is obtained by using the prediction weight of each conductivity data, and the specific steps include the following:

[0034] Input each conductivity data and the predicted weight of the conductivity data into the AR model to obtain the predicted value of the k-th data at the prediction moment;

[0035]

[0036] In the formula, represents the predicted value of the k-th data at the prediction moment; represents the predicted weight of the i-th data in the conductivity data; represents the conductivity value of the i-th data in the conductivity data; represents the number of conductivity data used for prediction.

[0037] Furthermore, the real-time monitoring of the water quality of the emission pool in petrochemical industry according to the predicted value of the conductivity data includes the following specific steps:

[0038] Obtain the residual data between the predicted value of the data and the conductivity value at each prediction moment, perform Z-score statistics on all residual data, and then obtain the Z-score of each residual data by the Z-score statistical algorithm. Mark the residual data with the absolute value of the Z-score greater than the preset abnormal threshold as abnormal water quality conductivity data. When the number of abnormal water quality conductivity data exceeds the preset abnormal number, perform water quality cleaning measures.

[0039] The beneficial effects of the technical solution of the present invention are: collecting water quality conductivity data, pH value data and the minimum positive operation period of the wastewater treatment system in the emission pool, calculating the local abnormal values of each conductivity data and pH value data, which initially reflects the characteristics of noise data; thereby obtaining the correlation difference values of all conductivity data, and combining the differences of the correlation difference values of the conductivity data at the same position in other cycles, obtaining the noise performance degree of each conductivity data, which is beneficial to appropriately adjust the predicted weight of each conductivity data, and then obtaining the predicted value of each conductivity data through the predicted weight of each conductivity data, avoiding the interference of noise data and improving the accuracy of real-time monitoring of the water quality of the emission pool in petrochemical industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0041] Figure 1 It is a flowchart of the steps of a real-time water quality monitoring method for an emission pool in petrochemical industry according to the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0042] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following specifically describes, in conjunction with the accompanying drawings and preferred embodiments, a real-time water quality monitoring method for an oil and chemical industrial effluent pond according to the present invention, including its specific implementation manner, structure, features, and effects. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0044] The following specifically describes, in conjunction with the accompanying drawings, the specific solution of a real-time water quality monitoring method for an oil and chemical industrial effluent pond provided by the present invention.

[0045] Please refer to Figure 1 , which shows a step flow chart of a real-time water quality monitoring method for an oil and chemical industrial effluent pond provided by an embodiment of the present invention. The method includes the following steps:

[0046] Step S001: Collect water quality conductivity data and pH value data, and obtain the minimum normal operation period of the wastewater treatment system in the effluent pond.

[0047] The main purpose of this embodiment is to predict the water quality conductivity data in the oil and chemical industrial effluent pond, and realize abnormal monitoring according to the difference between the prediction result and the actual water quality conductivity. First, install a water quality conductivity sensor in the oil and chemical industrial effluent pond to be monitored, and use the sensor to collect the water quality conductivity data at each sampling moment. At the same time, use a pH value detection instrument to synchronously collect the water quality pH value to obtain the water quality pH value data pH. Record the time interval for the wastewater treatment system to clean the wastewater. One wastewater cleaning is regarded as one treatment cycle, and the average value of the number of conductivity data in all treatment cycles is used as the minimum normal operation period of the wastewater treatment system in the oil and chemical industrial effluent pond. In this embodiment, the time interval between the collection moments of the conductivity data and the pH value data is described by taking a measurement every 10 minutes as an example, and the range of the collection moments of the conductivity data and the pH value data is described by taking the most recent 40 days as an example. Other embodiments can set other values for the collection moment interval and the collection moment range, and this embodiment does not make specific limitations. It should be noted that since the neutral value of the pH value is 7, the pH value data in this embodiment is the absolute value of the difference between the value collected by the pH value detection instrument and the neutral value 7.

[0048] Step S002: Obtain the local outlier of each conductivity data according to the difference between each conductivity data and the surrounding conductivity data.

[0049] It should be noted that the purpose of this embodiment is to adaptively determine the prediction weights of historical data in the AR prediction model, and then obtain the predicted value of water quality conductivity from the prediction weights. Since noise data will affect the predicted value in the AR prediction model, it is necessary to judge the noise probability of each water quality conductivity data and reduce the prediction weight of the noise data. Since noise data has the characteristics of sharp rises and falls among all data, in this embodiment, the noise probability of water quality conductivity data is initially judged based on the local characteristics of the data. The AR model described in this embodiment is a well-known prior art and will not be elaborated herein.

[0050] Specifically, if the difference between a certain conductivity data and the data mean of the overall conductivity data is larger, it indicates that the noise probability of this conductivity data is higher. At the same time, if the numerical difference between a certain conductivity data and its neighboring data is larger, and the standard deviation of the local data segment formed by this conductivity data and its neighboring data is larger, it indicates that the possibility of this conductivity data being noise data is higher. The calculation method of the local outlier of the i-th data in the conductivity data is as follows:

[0051]

[0052] In the formula, represents the local outlier of the i-th data in the conductivity data; represents the conductivity value of the i-th data in the conductivity data; represents the mean of the conductivity values of all conductivity data; represents the number of neighboring data of the i-th data in the conductivity data. The acquisition method of the neighboring data is as follows: A neighborhood is constructed with the i-th data in the conductivity data as the center, and the conductivity data within the neighborhood is recorded as the neighboring data of the i-th data. In this embodiment, taking 8 conductivity data with the i-th conductivity data as the center as an example for description, other values can be set in other embodiments, which are not limited in this embodiment; represents the conductivity value of the j-th neighboring data of the i-th data in the conductivity data; represents the standard deviation between the i-th data in the conductivity data and its neighboring data; represents the linear normalization function, represents the absolute value function.

[0053] represents the numerical difference between the i-th data in the conductivity data and the overall conductivity data. The larger this value is, the greater the difference between the i-th conductivity data and the mean level of the overall conductivity data, indicating that the noise probability of this conductivity data is higher; It represents the difference in conductivity values between the i-th conductivity data and its neighboring data. The larger this value is, the more obvious the local anomaly of the i-th conductivity data, and the greater the possibility that it is noise data. The larger it is, the more obvious the fluctuation characteristics of the local data segment corresponding to the i-th conductivity data, which indicates that the noise probability of the i-th conductivity data is greater.

[0054] Similarly, according to the differences between each pH value data and the surrounding pH value data, the local anomaly values of each pH value data are obtained. It should be noted that the calculation method of the local anomaly value of pH value data is the same as that of conductivity data.

[0055] So far, the local anomaly values of each conductivity data and the local anomaly values of each pH value data have been obtained.

[0056] Step S003: Construct a local anomaly ratio curve and evenly segment the local anomaly ratio curve; according to the differences in local anomaly ratios and data correlations between the conductivity data and pH value data of each segmented data, the correlation difference values of all conductivity data are obtained.

[0057] It should be noted that when analyzing the characteristics of water quality conductivity data, the pH value of the water quality will affect the water quality conductivity data, thereby affecting the prediction weight of the predicted value in the AR prediction model. The correlation characteristics of the water quality pH value and the water quality conductivity can be combined for analysis. The pH value is an index to measure the acidity and alkalinity of the water body. Hydrogen ions in acidic solutions and hydroxide ions in alkaline solutions can both increase the conductivity. When the water body is acidic or alkaline, its conductivity is usually high. When the water quality pH value tends to be more acidic or alkaline, the water quality conductivity data is larger; when the water quality pH value tends to be more neutral, the water quality conductivity data is smaller. Therefore, the correlation difference value of the water quality conductivity data can be obtained according to the differences in local anomaly values in the water quality conductivity data and the water quality pH value data.

[0058] Specifically, the ratio of the local anomaly value of each conductivity data to the local anomaly value of the pH value data at the corresponding acquisition time is recorded as the local anomaly ratio. All local anomaly ratios are obtained and a curve space is constructed. In the curve space, the conductivity data is used as the horizontal axis and the value of the local anomaly ratio is used as the vertical axis. All local anomaly ratios are projected into the curve space to obtain the local anomaly ratio curve.

[0059] Further, the local anomaly ratio curve is evenly divided into M segments according to the acquisition time to obtain M segments of segmented data. In this embodiment, M , taking this as an example for description, it can be set to other values in other embodiments, which is not limited in this embodiment.

[0060] Then the calculation method of the correlation difference value of the i-th data in the conductivity data is as follows:

[0061]

[0062] In the formula, represents the correlation difference value of the i-th data in the conductivity data; represents the i-th local anomaly ratio in the local anomaly ratio curve; represents the mean value of all local anomaly ratios in the local anomaly ratio curve; represents the number of segments of the local anomaly ratio curve; represents the segment data where the i-th local anomaly ratio in the local anomaly ratio curve is located; represents the x-th segment data in the local anomaly ratio curve except for the segment data where the i-th local anomaly ratio is located; is the exponential function with the natural constant as the base; represents the function for calculating the Pearson correlation coefficient, represents the absolute value function.

[0063] represents the difference between the i-th ratio in the local anomaly ratio curve and the mean level of all ratios. The larger this value is, the less the ratio satisfies the correlation between the conductivity data and the water quality pH value data, and the smaller the correlation difference value is; represents the mean value of the Pearson correlation coefficients between the segment where the i-th local anomaly ratio in the local anomaly ratio curve is located and other segments. The smaller this value is, the worse the correlation between the segment where the i-th local anomaly ratio in the local anomaly ratio curve is located and other segments, the greater the noise probability of the i-th conductivity data, and the smaller the correlation difference value is.

[0064] Similarly, according to the differences and correlations between the conductivity data and the pH value data on the local anomaly ratio curve, the correlation difference values of all conductivity data are obtained.

[0065] Step S004: Obtain the noise performance degree of all data in the conductivity data according to the correlation difference value of each conductivity data in the smallest running positive period and the difference between the correlation difference values of the conductivity data at the same position in other periods.

[0066] It should be noted that the operation periodicity of the wastewater treatment system in the discharge pool may affect the concentrations of various chemical substances in the wastewater, thereby affecting the conductivity value of the water quality. Since the petrochemical production process often operates according to a certain periodicity, the composition and properties of the wastewater change periodically, and then the conductivity data of the water quality in the discharge pool shows periodic changes.

[0067] Specifically, for the minimum positive operation period of the wastewater treatment system in the discharge pond, the greater the data difference at the corresponding position within each period, the greater the noise performance degree of the data. The calculation method for the noise performance degree of the i-th data in the conductivity data is as follows:

[0068]

[0069] Among them, the first parameter The calculation method is:

[0070]

[0071] The second parameter The calculation method is:

[0072] In the formula, represents the noise performance degree of the i-th data in the conductivity data; represents the correlation difference value of the i-th data in the conductivity data; represents the number of all minimum operation periods; represents the correlation difference value of the conductivity data at the same acquisition moment in the y-th period other than the minimum operation period where the i-th data is located and the minimum operation period where the i-th data is located; represents the maximum value of the correlation difference value in the minimum operation period where the i-th data in the conductivity data is located; represents the maximum value of the correlation difference value in the y-th period other than the minimum operation period where the i-th data is located; represents the normalization function, represents the absolute value function.

[0073] The first parameter represents the numerical difference between the i-th data in the conductivity data and the values at other corresponding period positions. The larger this value, the greater the abnormal degree of the correlation difference value of the i-th data and the greater the noise probability of the i-th conductivity data; represents the difference between the correlation difference value of the i-th data in the conductivity data and the maximum value of the large correlation difference value within the period where the i-th data is located, denoted as the difference characteristic value of the i-th data; represents the correlation difference value of the conductivity data at the same position as the i-th data in the -th other period except the period where the i-th data is located, and the difference characteristic value of the maximum value of the correlation difference value in the -th other period. Then the second parameter It represents the difference between the difference eigenvalue of the \(i\)-th data in the conductivity data and the difference eigenvalues of the data at the same position in other periods. The larger this difference value is, the greater the difference in the correlation of the \(i\)-th data in different periods, and the greater the degree of noise manifestation of the \(i\)-th data.

[0074] Similarly, according to the correlation difference values of each data in the conductivity data, the degree of noise manifestation of all data in the conductivity data is obtained.

[0075] Step S005: According to the degree of noise manifestation of each data in the conductivity data and the time interval from the prediction moment, obtain the prediction weight of each data in the conductivity data; obtain the predicted value of each data at the prediction moment through the prediction weight, and perform real-time monitoring on the water quality of the emission pool in petrochemical industry.

[0076] It should be noted that if the degree of noise manifestation of a certain conductivity data is greater and the interval from the prediction moment is farther, then when using the AR model for prediction, since the conductivity may change greatly over time, the reference of the conductivity data with a farther interval from the prediction moment is lower during prediction, that is, the prediction weight of this conductivity data with a farther interval from the prediction moment for the predicted value is smaller. Therefore, the prediction weight of this conductivity data will be reduced in the AR prediction model.

[0077] Specifically, calculate the prediction weight of the \(i\)-th data in the conductivity data according to the degree of noise manifestation of the \(i\)-th data in the conductivity data and the time interval from the prediction moment, and input each conductivity data and the prediction weight of the conductivity data into the AR model to obtain the predicted value of the \(k\)-th data at the prediction moment.

[0078] Then the calculation method of the prediction weight of the \(i\)-th data in the conductivity data is:

[0079]

[0080] The calculation method of the predicted value of the \(k\)-th data at the prediction moment is:

[0081]

[0082] In the formula, represents the predicted value of the \(k\)-th data at the prediction moment; represents the prediction weight of the \(i\)-th data in the conductivity data; represents the conductivity value of the \(i\)-th data in the conductivity data; represents the number of conductivity data used for prediction; represents the initial weight; represents the degree of noise manifestation of the \(i\)-th conductivity data; represents the time interval between the \(i\)-th data in the conductivity data and the prediction moment; represents a normalization function. The number of conductivity data used for prediction selected in this embodiment is 80, and the initial weight is , and this is used as an example for description. In other embodiments, it can be set to other values, and this embodiment does not limit it.

[0083] When the conductivity value at the prediction moment is collected, the residual data between the data prediction value and the conductivity value at each prediction moment is obtained. All the residual data is subjected to Z - score statistics, and then the Z - score of each residual data is obtained by the Z - score statistical algorithm. The residual data with the absolute value of the Z - score greater than the preset anomaly threshold is recorded as abnormal water quality conductivity data. When the number of abnormal water quality conductivity data exceeds the preset abnormal number, it indicates that there is an anomaly in the water quality of the drainage pool in the current petrochemical industry, and water quality cleaning measures are taken. In this embodiment, the preset anomaly threshold is 3, and the preset abnormal number is 8, and this is used as an example for description. In other embodiments, it can be set to other values, and this embodiment does not limit it. The Z - score algorithm is a well - known prior art, and this embodiment will not elaborate on it.

[0084] It should be noted that the model in this embodiment only represents a negative correlation relationship and restricts the result of the model output to be within the interval, where is the input of this model. In specific implementation, it can be replaced with other models with the same purpose. This embodiment only uses the model as an example for description and does not make specific limitations.

[0085] So far, the present invention is completed.

[0086] The above - mentioned are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A real-time monitoring method for the water quality of an emission pond in petrochemical industry, characterized in that, The method includes the following steps: Collect water quality conductivity data and pH value data, where each collection moment corresponds to a conductivity data and a pH value data; obtain the minimum positive operation period of the wastewater treatment system in the discharge pool; According to the difference between each conductivity data and the surrounding conductivity data, obtain the local outliers of each conductivity data. The specific formula is as follows: In the formula, represents the local outlier of the i-th data in the conductivity data; represents the conductivity value of the i-th data in the conductivity data; represents the mean value of the conductivity values of all conductivity data; represents the number of neighboring data of the i-th data in the conductivity data; represents the conductivity value of the j-th neighboring data of the i-th data in the conductivity data; represents the standard deviation between the i-th data and its neighboring data in the conductivity data; represents the linear normalization function, represents the absolute value function; Obtain the local outliers of each pH value data; construct a local outlier ratio curve based on the ratio of the local outliers of each conductivity data and the pH value data at the corresponding collection moment, and segment the local outlier ratio curve to obtain several segmented data; According to the difference in local outlier ratios and data correlations between the conductivity data of each segmented data and the pH value data at the corresponding collection moment, obtain the correlation difference values of all conductivity data. The specific formula is as follows: Wherein, represents the correlation difference value of the i-th data in the conductivity data; represents the i-th local anomaly ratio in the local anomaly ratio curve; represents the mean value of all local anomaly ratios in the local anomaly ratio curve; represents the number of segments of the local anomaly ratio curve; represents the segment data where the i-th local anomaly ratio in the local anomaly ratio curve is located; represents the x-th segment data in the local anomaly ratio curve except for the segment data where the i-th local anomaly ratio is located; is an exponential function with the natural constant as the base; represents a function for calculating the Pearson correlation coefficient, represents an absolute value function; According to the correlation difference value of each conductivity data in the minimum positive operation period and the difference in the correlation difference values of the conductivity data at the same position in other periods, obtain the noise performance degree of each conductivity data. The specific formula is as follows: In the formula, represents the noise performance degree of the i-th data in the conductivity data; is the first parameter, is the second parameter, represents the normalization function, represents the correlation difference value of the i-th data in the conductivity data; represents the number of all minimum operation cycles; represents the correlation difference value of the conductivity data at the same acquisition moment as that of the minimum operation cycle where the i-th data is located, on the y-th cycle except the minimum operation cycle where the i-th data is located in the conductivity data; represents the maximum value of the correlation difference values in the minimum operation cycle where the i-th data in the conductivity data is located; represents the maximum value of the correlation difference values on the y-th cycle except the minimum operation cycle where the i-th data is located in the conductivity data, represents the absolute value function; Use the time interval between each conductivity data and the prediction moment as a weight, and combine the noise performance degree of each conductivity data to obtain the prediction weight of each conductivity data; obtain the predicted value of the conductivity data at each prediction moment through the prediction weight of each conductivity data. The specific steps are as follows: Input each conductivity data and the prediction weight of the conductivity data into the AR model to obtain the predicted value of the k-th data at the prediction moment; In the formula, represents the k-th data prediction value at the prediction moment; represents the prediction weight of the i-th data in the conductivity data; represents the conductivity value of the i-th data in the conductivity data; represents the number of conductivity data used for prediction; Perform real-time monitoring of the water quality in the discharge pool of petrochemical industry according to the predicted value of the conductivity data.

2. The real-time water quality monitoring method for an emission pond in petrochemical industry according to claim 1, characterized in that, The method for obtaining the adjacent data includes the following specific steps: Construct a neighborhood centered on the i-th data in the conductivity data, and the conductivity data points within the neighborhood are recorded as the adjacent data of the i-th data.

3. The real-time monitoring method for the water quality of an emission pond for petrochemical industry according to claim 1, characterized in that The steps of constructing a local outlier ratio curve based on the ratio of the local outliers of each conductivity data and the pH value data at the corresponding collection moment, and segmenting the local outlier ratio curve to obtain several segmented data include the following specific steps: Record the ratio of the local outlier of each conductivity data to the local outlier of the pH value data at the corresponding collection moment as the local outlier ratio. Use all local outlier ratios to construct a curve space, with the conductivity data as the horizontal axis and the value of the local outlier ratio as the vertical axis in the curve space, and project all local outlier ratios into the curve space to obtain the local outlier ratio curve; Segment the local outlier ratio curve evenly according to the collection moment to obtain several segmented data.

4. The real-time water quality monitoring method for an emission pond in petrochemical industry according to claim 1, wherein, The steps of using the time interval between each conductivity data and the prediction moment as a weight, and combining the noise performance degree of each conductivity data to obtain the prediction weight of each conductivity data include the following specific formula: In the formula, represents the predicted weight of the i-th data in the conductivity data; represents the initial weight; represents the degree of noise manifestation of the i-th conductivity data; represents the time interval between the i-th data in the conductivity data and the prediction time; represents the normalization function.

5. The real-time monitoring method for the water quality of an emission pond for petrochemical industry according to claim 1, characterized in that, The steps of performing real-time monitoring of the water quality in the discharge pool of petrochemical industry according to the predicted value of the conductivity data include the following specific steps: Obtain the residual data between the data prediction value and the conductivity value at each prediction moment, perform Z-score statistics on all the residual data, and then obtain the Z-score of each residual data by the Z-score statistical algorithm. Denote the residual data with an absolute value of the Z-score greater than the preset anomaly threshold as abnormal water quality conductivity data. When the number of abnormal water quality conductivity data exceeds the preset abnormal quantity, implement water quality cleaning measures.

Images