Intelligent monitoring method for discharging oil melting wastewater

Abstract

The present application relates to the technical field of intelligent monitoring, in particular to a kind of oil melting wastewater discharge intelligent monitoring method.The method comprises: selecting the historical monitoring time interval as target production cycle time length integer multiple from current monitoring time as comparison historical monitoring time, and based on the water quality data variation difference between comparison historical monitoring time and current monitoring time, the target noise index value corresponding to the normalized water quality data under current monitoring time is obtained, the normalized water quality data under current monitoring time is processed based on target noise index value to obtain current target water quality data, and the oil melting wastewater discharge is monitored based on current target water quality data.The present application can avoid the interference of noise on the monitoring accuracy of oil melting wastewater discharge, and thus can improve the accuracy of monitoring oil melting wastewater discharge.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring technology, specifically to an intelligent monitoring method for wastewater discharge from oil smelting. Background Technology

[0002] The smelting of oils and fats generates a large amount of oily organic wastewater, characterized by high COD concentration, high oil content, and high suspended solids concentration. Currently, in order to prevent the high oil content, high suspended solids concentration, and high COD concentration of smelting wastewater from entering urban pipe networks or natural water bodies, and to avoid water ecological safety problems such as pipe blockage, water body hypoxia, eutrophication, and foul odor, it is usually necessary to monitor the discharge of smelting wastewater.

[0003] Existing technologies typically rely on comparing water quality data such as COD concentration and suspended solids concentration, directly monitored and collected at the outlet of oil smelting wastewater, with thresholds or normal ranges to indicate whether there are any abnormal emissions. However, the equipment used for data collection is easily affected by factors such as oil film adhesion, bubble interference, and electromagnetic interference, resulting in noise data in the collected data. Noise data is easily confused with real fluctuation data, which can easily lead to false monitoring. Therefore, how to distinguish noise data to improve the accuracy of monitoring oil smelting wastewater emissions has become an urgent problem to be solved. Summary of the Invention

[0004] To address the above problems, this invention provides an intelligent monitoring method for wastewater discharge from oil and fat smelting, the specific technical solution of which is as follows:

[0005] One embodiment of the present invention provides an intelligent monitoring method for the discharge of oil smelting wastewater, comprising the following steps:

[0006] Obtain normalized auxiliary impact data at historical monitoring times and normalized water quality data at the current monitoring time. The normalized auxiliary impact data includes normalized wastewater temperature and normalized wastewater discharge flow rate.

[0007] The system determines whether the change in water quality data at the current monitoring time is greater than a preset change threshold. If it is not greater, the normalized water quality data at the current monitoring time is recorded as the current target water quality data. If it is greater, the system clusters the historical monitoring times based on the normalized auxiliary influence data and local fitting slope at the historical monitoring times to obtain the sub-time period sequence corresponding to each target cluster. The system obtains the target production cycle time length based on the standard deviation and median of the time difference between adjacent sub-time periods in the sub-time period sequence. The system selects historical monitoring times that are an integer multiple of the target production cycle time length from the current monitoring time as comparison historical monitoring times. Based on the difference in water quality data change between the comparison historical monitoring times and the current monitoring time, the system obtains the target noise index value corresponding to the normalized water quality data at the current monitoring time. The system performs noise judgment processing on the normalized water quality data at the current monitoring time based on the target noise index value to obtain the current target water quality data.

[0008] The discharge of wastewater from oil smelting is monitored based on the current target water quality data.

[0009] Beneficial effects: This invention first acquires normalized auxiliary influence data at historical monitoring times and normalized water quality data at the current monitoring time; then, it determines whether the change in water quality data at the current monitoring time is greater than a preset change threshold. If it is not greater, the normalized water quality data at the current monitoring time is recorded as the current target water quality data. If it is greater, the results of clustering historical monitoring times based on the normalized auxiliary influence data and local fitting slope at historical monitoring times are used to obtain the sub-time period sequence corresponding to each target cluster. The target production cycle time length is obtained based on the standard deviation and median of the time difference between adjacent sub-time periods in the sub-time period sequence. Historical monitoring times that are an integer multiple of the target production cycle time length at the current monitoring time are selected as comparison historical monitoring times. Based on the difference in water quality data change between the comparison historical monitoring times and the current monitoring time, the target noise index value corresponding to the normalized water quality data at the current monitoring time is obtained. The normalized water quality data at the current monitoring time is then subjected to noise judgment processing based on the target noise index value to obtain the current target water quality data. Finally, the discharge of oil smelting wastewater is monitored based on the current target water quality data. Furthermore, this invention uses the target production cycle duration as the screening criterion, extracts and compares the difference in water quality data changes between historical monitoring times and current monitoring times as the basis for noise judgment, and finally conducts monitoring of oil smelting wastewater discharge based on the processed data. This can avoid the interference of noise on the monitoring accuracy of oil smelting wastewater discharge and significantly improve the accuracy of monitoring oil smelting wastewater discharge. Attached Figure Description

[0010] To more clearly illustrate the technical solutions and advantages 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.

[0011] Figure 1 This is a flowchart of an intelligent monitoring method for wastewater discharge from oil smelting according to the present invention. Detailed Implementation

[0012] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the protection scope of the embodiments of the present invention.

[0013] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art.

[0014] This embodiment provides an intelligent monitoring method for wastewater discharge from oil smelting, detailed as follows:

[0015] like Figure 1 As shown, the intelligent monitoring method for wastewater discharge from oil smelting includes the following steps:

[0016] Step S001: Obtain normalized auxiliary impact data at historical monitoring times and normalized water quality data at the current monitoring time.

[0017] Because equipment used for water quality monitoring at the discharge outlet of grease smelting wastewater is susceptible to factors such as oil film adhesion, bubble interference, and electromagnetic interference, the collected water quality data is prone to noise. This noise data may exceed the specified or preset normal range, which can easily lead to misidentification of the noise data as abnormal discharge, affecting the accuracy of monitoring grease smelting wastewater discharge. In order to improve the accuracy of monitoring grease smelting wastewater discharge, this embodiment will subsequently perform noise judgment processing based on the characteristic differences between the actual water quality fluctuations caused by the grease smelting process and the noise data. Then, the data obtained from the noise judgment processing will be used for monitoring grease smelting wastewater discharge, thereby improving the accuracy of grease smelting wastewater discharge monitoring.

[0018] This embodiment first collects monitoring data reflecting water quality during the discharge of oil smelting wastewater. Specifically, during the discharge of oil smelting wastewater, water quality monitoring equipment deployed at the discharge outlet is used to monitor and collect water quality data. The collected data is recorded as the water quality data to be treated at the corresponding monitoring time. The water quality monitoring equipment deployed at the discharge outlet includes, but is not limited to, online COD analyzers, infrared spectrophotometers, pH monitors, and online SS monitors. Therefore, the water quality data types in this embodiment include, but are not limited to, COD, oil content, pH, and suspended solids concentration. That is, the water quality data to be treated includes the COD to be treated, the oil content to be treated, the pH to be treated, and the suspended solids concentration to be treated. The online SS monitor is a water quality analysis device used to monitor the suspended solids concentration in the discharge outlet water. The online COD analyzer is an automated water quality analysis device used to monitor the chemical oxygen demand (COD) in the discharge outlet water. The infrared spectrophotometer is a precision analysis instrument used to monitor the petroleum and animal / vegetable oil content in the discharge outlet water. The pH monitor is used to monitor the pH value of the discharge outlet water. This embodiment requires that the water quality monitoring equipment deployed at the wastewater discharge outlet of oil smelting be synchronously collected. Since the COD online monitoring instrument is limited by reaction time, its monitoring frequency is relatively low. Therefore, if the monitoring frequency of the COD online monitoring instrument is lower than the monitoring frequency required by this embodiment, the COD data to be processed at each monitoring time can be obtained through existing interpolation methods, such as cubic spline interpolation. For example, if the water quality monitoring equipment required by this embodiment collects data once per minute (i.e., the time interval between adjacent monitoring times is 1 minute), and the COD online analyzer can only collect data once every 5 minutes at the fastest, then for a certain monitoring time, if the COD online analyzer does not collect data at that monitoring time, the COD data obtained through cubic spline interpolation at that monitoring time will be used as the COD data at that monitoring time. The data obtained through cubic spline interpolation is the data before normalization.

[0019] Furthermore, water quality monitoring during the oil smelting process is affected by wastewater temperature and discharge flow rate. These parameters are key reference data for real water quality fluctuations and noise. For example, increased temperature increases oil solubility. Therefore, in this embodiment, wastewater temperature and discharge flow rate must also be considered when identifying noise. These parameters are collectively referred to as auxiliary impact data. Thus, in this embodiment, auxiliary impact data monitoring equipment is used to monitor and collect auxiliary impact data during the oil smelting wastewater discharge process, obtaining the auxiliary impact data to be treated at different monitoring times. This allows us to obtain the wastewater temperature and discharge flow rate to be treated at different monitoring times. The auxiliary impact data monitoring equipment includes a temperature sensor and a flow meter. The temperature sensor monitors the wastewater temperature, and the flow meter monitors the wastewater flow rate. In this embodiment, both can be placed at the discharge outlet. In this embodiment, auxiliary impact data and water quality data are collected synchronously. The auxiliary impact data and water quality data to be treated in this embodiment are directly collected or interpolated by the monitoring equipment without normalization.

[0020] In this embodiment, after obtaining the water quality data and auxiliary influence data at each monitoring time, the water quality data and auxiliary influence data at each monitoring time are normalized. The normalized data are recorded as the normalized water quality data and normalized auxiliary influence data at the corresponding monitoring time. This yields the normalized COD, normalized oil content, normalized pH, normalized suspended solids concentration, normalized wastewater temperature, and normalized wastewater discharge flow rate at each monitoring time. Here, an existing normalization method is selected for normalization, such as minimum-maximum normalization. Normalization is used to avoid the influence of dimensions on subsequent analysis.

[0021] Therefore, this embodiment can obtain normalized water quality data and normalized auxiliary influence data at each monitoring time through the above process. This embodiment then needs to perform noise judgment processing on the normalized water quality data at the current monitoring time based on the historical normalized water quality data after noise judgment processing. In this embodiment, the data after noise judgment processing of the normalized water quality data at the current monitoring time is the current target water quality data. Therefore, this embodiment needs to obtain the historical monitoring time period used for noise judgment processing of the normalized water quality data at the current monitoring time, which is the historical monitoring time period corresponding to the current monitoring time. In specific applications, the implementer needs to set the length of the historical monitoring time period according to the actual situation, but it is required to cover at least one complete production cycle. A complete production cycle is the entire process from the input of raw materials into the refining tank to the production of crude oil during the oil smelting process. For example, in this embodiment, one month before the current monitoring time can be selected as the historical monitoring time period corresponding to the current monitoring time. To ensure the accuracy of subsequent analysis, this embodiment uses historical water quality data processed by normalization and noise assessment when performing noise discrimination or noise judgment processing on the normalized water quality data at the current monitoring time. In other words, the historical water quality data used is the normalized water quality data processed by noise assessment, and the normalized water quality data processed by noise assessment is the corresponding historical target water quality data at that historical monitoring time. Furthermore, the method for obtaining historical target water quality data at historical monitoring times in this embodiment is logically consistent with the method for obtaining current target water quality data at the current monitoring time, as described later in this embodiment. Therefore, the process of obtaining historical target water quality data will not be described further. However, if the total monitoring time before a certain historical monitoring time is less than the preset monitoring time, noise assessment is not performed; instead, the normalized water quality data at that historical monitoring time is directly used as the historical target water quality data at that historical monitoring time.

[0022] Therefore, this embodiment can obtain normalized water quality data and normalized auxiliary impact data at the current monitoring time through the above process, as well as historical target water quality data and normalized auxiliary impact data at each historical monitoring time in the historical monitoring period of the current monitoring time. The normalized auxiliary impact data includes normalized wastewater temperature and normalized wastewater discharge flow rate. Moreover, this embodiment first identifies the noise and then performs targeted noise reduction. Compared with blindly performing uniform noise reduction on each data point, this method can avoid erasing real fluctuations or real abnormal information and improve the accuracy of discharge monitoring. Furthermore, since the method for determining whether each type of normalized water quality data at the current monitoring time is noise or requires denoising processing is consistent in this embodiment, that is, the method for obtaining the target water quality data corresponding to each type of normalized water quality data at the current monitoring time by performing noise judgment processing is consistent in this embodiment, for ease of description and understanding, the following description will take the process of noise identification, denoising or noise judgment processing of normalized water quality data corresponding to any type at the current monitoring time as an example. For example, the process of noise identification, denoising or noise judgment processing of normalized water quality data with oil content as the data type at the current monitoring time can be used as an example. That is, the type of water quality data that appears in the subsequent noise identification, denoising or noise judgment processing process in this embodiment is oil content.

[0023] Step S002: Determine whether the change in water quality data at the current monitoring time is greater than a preset change threshold. If it is not greater, record the normalized water quality data at the current monitoring time as the current target water quality data. If it is greater, cluster the historical monitoring times based on the normalized auxiliary influence data and local fitting slope at the historical monitoring times to obtain the sub-time period sequence corresponding to each target cluster. Obtain the target production cycle time length based on the standard deviation and median of the time difference between adjacent sub-time periods in the sub-time period sequence. Select historical monitoring times that are integer multiples of the target production cycle time length from the current monitoring time as comparison historical monitoring times. Obtain the target noise index value corresponding to the normalized water quality data at the current monitoring time based on the difference in water quality data change between the comparison historical monitoring times and the current monitoring time. Perform noise judgment processing on the normalized water quality data at the current monitoring time based on the target noise index value to obtain the current target water quality data.

[0024] Normally, non-noise data is relatively stable, generally fluctuating within a small range, while noise data is usually not correlated with actual oil smelting wastewater discharge and typically fluctuates more significantly than non-noise data. Therefore, this embodiment will first obtain the change in water quality data at the current monitoring time based on the magnitude of the change in water quality data between the current monitoring time and the previous monitoring time. The change in water quality data can reflect the probability that the normalized water quality data at the current monitoring time is noise or non-noise. Therefore, this embodiment will then obtain the change in water quality data at the current monitoring time based on the normalized water quality data at the current monitoring time and the historical target water quality data at the previous monitoring time. The specific process for obtaining the change in water quality data at the current monitoring time is as follows:

[0025] The relative change of the normalized water quality data at the current monitoring time compared to the historical target water quality data at the previous monitoring time is taken as the change in water quality data at the current monitoring time. This is an expression for the change in water quality data at the current monitoring time. This is the normalized water quality data at the current monitoring time. This refers to the historical target water quality data from the previous monitoring time. This is a preset constant to avoid a denominator of 0. In specific applications, implementers can set the value of the preset constant according to the actual situation, such as 0.01. The smaller the change in water quality data at the current monitoring time, the smaller the change in the normalized water quality data at the current monitoring time relative to the target water quality data at the previous monitoring time. In this case, the normalized water quality data at the current monitoring time is more likely to be non-noise. The larger the change in water quality data at the current monitoring time, the larger the change in the normalized water quality data at the current monitoring time relative to the target water quality data at the previous monitoring time. In this case, the normalized water quality data at the current monitoring time is more likely to be noise.

[0026] Therefore, after obtaining the change in water quality data at the current monitoring time, this embodiment performs preliminary noise identification, that is, it determines whether the change in water quality data at the current monitoring time is greater than a preset change threshold. If it is not greater, it indicates that the normalized water quality data at the current monitoring time is non-noise data, and no denoising processing or subsequent analysis is required. The normalized water quality data at the current monitoring time can be directly recorded as the current target water quality data. Subsequent monitoring of oil smelting wastewater discharge at the current monitoring time is based on the current target water quality data. However, if it is determined that the change in water quality data at the current monitoring time is greater than the preset change threshold, it indicates that the normalized water quality data at the current monitoring time may be noise data or that the normalized water quality data at the current monitoring time is suspected noise data. However, it cannot be directly determined to be noise data because the oil smelting production stage... Transition processes or sudden changes in wastewater flow can cause significant fluctuations in monitoring data, or result in non-noise or real fluctuation data showing a large change in the target water quality data at the previous monitoring time. In other words, the existence of transition processes or sudden changes in wastewater flow during the oil smelting production stage can lead to non-noise or real fluctuation data even when the change in water quality data at the current monitoring time exceeds a preset change threshold. Therefore, to ensure the accuracy of noise identification, this embodiment, when determining that the change in water quality data at the current monitoring time exceeds the preset change threshold, first obtains the target production cycle time length, and then obtains the target noise index value corresponding to the normalized water quality data at the current monitoring time based on the obtained target production cycle time length. Subsequently, the noise of the normalized water quality data at the current monitoring time will be judged again based on the target noise index value. In practical applications, implementers can set preset change thresholds based on historical statistical distribution characteristics and other actual conditions. For example, if a box plot is constructed for the change in water quality data at all monitoring times, the upper limit of the normal range can be used as the preset change threshold. Thus, Q3 + 1.5 × IQR can be selected as the preset change threshold, where Q3 is the upper quartile and IQR is the interquartile range.

[0027] Since the production cycle length of oil smelting is not completely fixed, the cycle length may vary between different batches of oil smelting due to adjustments in some processes or other factors. To ensure the accuracy of noise identification in this embodiment, it is necessary to select data from historical monitoring periods that are at a similar or close production stage to the current monitoring time to assess the noise situation of the water quality data at the current monitoring time. Therefore, this embodiment needs to first obtain data that can be filtered out from historical monitoring periods to determine the target production cycle length as much as possible. Specifically, the target production cycle length should be selected from historical monitoring periods that are at a similar or close production stage to the current monitoring time. The specific process for obtaining the target production cycle length is as follows:

[0028] Since the auxiliary impact data is directly related to the production stage, this embodiment will first cluster the historical monitoring times based on the normalized auxiliary impact data and local fitting slopes of the historical monitoring time periods to obtain the sub-time period sequence corresponding to each target cluster. Then, based on the standard deviation and median of the time difference between adjacent sub-time periods in the sub-time period sequence corresponding to each target cluster, the target production cycle length will be obtained. That is, the sub-time period sequence corresponding to the target cluster is the key to determining the target production cycle length.

[0029] The specific process of clustering historical monitoring moments based on normalized auxiliary influence data and local fitting slopes within a historical monitoring period to obtain the sub-time period sequence corresponding to each target cluster is as follows:

[0030] Because different production stages may differ not only in temperature and flow rate, but more importantly, in the direction of temperature and flow rate changes—for example, the initial temperature in the preheating stage is low, but gradually increases as heating progresses, while the temperature in the cooling stage gradually decreases and remains in the medium-low temperature range—the clustering process also needs to consider the trend values ​​of temperature and flow rate, namely, the fitting slopes of local wastewater temperature and local wastewater discharge flow rate. Therefore, in this embodiment, the fitting slopes of local wastewater temperature and local wastewater discharge flow rate at each historical monitoring time are first calculated. Furthermore, for any historical monitoring time within the historical monitoring period, the a historical monitoring times closest in time to that historical monitoring time, along with the data generated from that historical monitoring time, are considered. The time period is denoted as the local time period corresponding to the historical monitoring time. The slope obtained by linearly fitting the time series data segment formed by the normalized wastewater temperature at all historical monitoring times within the local time period corresponding to the historical monitoring time is denoted as the local wastewater temperature fitting slope at the historical monitoring time. The slope obtained by linearly fitting the time series data segment formed by the normalized wastewater discharge flow rate at all historical monitoring times within the local time period corresponding to the historical monitoring time is denoted as the local wastewater discharge flow rate fitting slope at the historical monitoring time. In specific applications, the implementer can set the value of 'a' according to the actual situation. For example, in this embodiment, 'a' can be set to 5. The local wastewater temperature fitting slope and the local wastewater discharge flow rate fitting slope... Specific instructions for obtaining the rate: Construct a temperature coordinate system. Map each historical monitoring moment and the normalized wastewater temperature at each historical monitoring moment within the local time period corresponding to the historical monitoring moment onto the temperature coordinate system. This yields the data points corresponding to each historical monitoring moment within the local time period. Use the least squares method to perform linear fitting on the data points corresponding to each historical monitoring moment within the local time period. The slope of the fitted line is the local wastewater temperature fitting slope. The local wastewater discharge flow rate fitting slope is obtained similarly, except that the coordinate system mapped when obtaining the local wastewater discharge flow rate fitting slope is the flow rate coordinate system, and the horizontal axis of the temperature coordinate system is used instead. The time coordinate system uses the vertical axis to represent wastewater temperature, while the flow rate coordinate system uses the horizontal axis to represent time and the vertical axis to represent wastewater discharge flow rate. Then, the fitting slopes for local wastewater temperature and local wastewater discharge flow rate are normalized, and the normalized results are denoted as the normalized temperature fitting slope and the normalized flow rate fitting slope. Existing normalization methods are used here, such as max-min normalization. Next, the feature vectors for each historical monitoring time are obtained, and the vector formed by the normalized wastewater temperature, normalized temperature fitting slope, normalized wastewater discharge flow rate, and normalized flow rate fitting slope for each historical time is denoted as the feature vector for the corresponding historical monitoring time.Next, based on the feature vectors of historical monitoring moments within the historical monitoring period, the metric distance between historical monitoring moments within the historical monitoring period is calculated. The metric distance between any two historical monitoring moments is the Euclidean distance between their feature vectors. Then, based on the metric distance between historical monitoring moments within the historical monitoring period, k-means clustering is performed on all historical monitoring moments within the historical monitoring period to obtain each initial cluster. The number of cluster centers is determined using the elbow method. The purpose of clustering is to group moments with similar production stages into the same cluster as much as possible. The process of clustering historical monitoring moments using the k-means clustering algorithm based on the known metric distance is a well-known technique. However, due to the possibility of misclassification... Furthermore, this leads to the occurrence of isolated monitoring moments. Therefore, in this embodiment, the monitoring moments adjacent to the unmonitored moments (isolated monitoring moments) within the initial clusters are migrated, and all the migrated initial clusters are recorded as target clusters. The isolated monitoring moments within the initial clusters are migrated to the initial cluster to which the time nearest neighbor monitoring moment of the corresponding monitoring moment belongs. Then, the time periods consisting of continuous historical monitoring moments in each target cluster are recorded as the sub-time periods corresponding to the corresponding target cluster. That is, the sub-time periods corresponding to any target cluster are composed of continuous historical monitoring moments in that target cluster. The result of arranging all the sub-time periods corresponding to each target cluster in chronological order is recorded as the sub-time period sequence corresponding to the corresponding target cluster.

[0031] The specific process for obtaining the target cluster is as follows: First, the initial clusters are sorted in descending order of the number of parameters within the cluster. Then, each cluster is checked in the sorted order. If a monitoring time has no adjacent points within the current cluster (i.e., an isolated monitoring time), it is moved to the cluster to which the previous monitoring time belongs. This process is repeated until all clusters have been traversed. Finally, the target clusters with optimized time continuity are obtained. That is, all the initial clusters after traversal and migration are target clusters. The above traversal and migration need to skip the processed clusters that have no isolated monitoring times and only perform migration on the clusters that have isolated monitoring times. If a certain time to be migrated has no previous monitoring time, it is moved to the cluster to which the next monitoring time belongs.

[0032] The specific process for obtaining the target production cycle length based on the standard deviation and median of the time difference values ​​between adjacent sub-time periods in the sub-time period sequence corresponding to each target cluster is as follows: The set of time differences between all adjacent sub-time periods in the sub-time period sequence corresponding to each target cluster is denoted as the time difference set corresponding to the corresponding target cluster; and the b-th time difference in the time difference set corresponding to any target cluster is the time interval between the start time of the (b+1)-th sub-time period and the start time of the b-th sub-time period in the sub-time period sequence corresponding to that target cluster; since sub-time periods belonging to the same cluster are more likely to be different... Within the same production cycle and at the same production stage, the time difference between adjacent sub-time periods in the same cluster can characterize the production cycle length. Then, the median of the time difference set corresponding to each target cluster is selected as the representative cycle time length of the corresponding target cluster. Since there may be phenomena where the sub-time period length is too short or too long due to abnormalities or other factors, such as abnormal shutdowns or cleaning, the obtained sub-time period length may be too short or too long. To avoid these factors affecting the determination of the subsequent target production cycle time length, this embodiment selects the median of the time difference set for determining the target production cycle time length.

[0033] Furthermore, since sub-time periods within the target cluster may not belong to the same production stage or may have significant differences in production stages, excessive reference to the representative cycle time length of the corresponding target cluster when determining the target production cycle time length may lead to historical monitoring times selected based on the final obtained target production cycle time length not being in the same production stage as the current monitoring time. Therefore, it is necessary to reduce the contribution of the representative cycle time length of such target clusters to the subsequent determination of the target production cycle time length. Also, because the lower the consistency of the time difference set corresponding to the target cluster, the better... This indicates that the likelihood of sub-time periods that are not at the same production stage or sub-time periods with significant differences in production stages within the corresponding target cluster is higher. Furthermore, the standard deviation of the time difference set can characterize consistency. Therefore, in this embodiment, the production cycle reflection value of each target cluster will be obtained based on the standard deviation of the time difference set corresponding to each target cluster. In this embodiment, the standard deviation of the time difference set corresponding to each target cluster is inversely normalized and then proportionally normalized, and the result is used as the production cycle reflection value of the corresponding target cluster. The expression for the production cycle reflection value of the w-th target cluster is:

[0034]

[0035]

[0036] in, The production cycle of the w-th target cluster reflects the characteristic value, where M is the number of target clusters. The result of inverse normalization of the standard deviation of the time difference set corresponding to the w-th target cluster, using the Norm() normalization function. Let $\begin{cases} \ ... This is a preset constant to avoid the denominator being 0. After taking the reciprocal, normalize to obtain the pair Perform reverse normalization. The larger the value, the greater the likelihood that there are time periods in the corresponding target cluster that are not at the same production stage, or that there are sub-time periods in the corresponding target cluster with significant differences in production stages. This further indicates that the representative cycle length of the w-th target cluster contributes less to the subsequent determination of the target production cycle length. When it is larger, The smaller, therefore The smaller the value, the lower the contribution of the representative cycle time length of the w-th target cluster to the subsequent determination of the target production cycle time length. The larger the value, the higher the contribution of the representative cycle time length of the w-th target cluster to the subsequent determination of the target production cycle time length. In other words, the production cycle reflection value can reflect the contribution of the representative cycle time length of the target cluster to the subsequent determination of the target production cycle time length.

[0037] Next, the target production cycle time is obtained based on the production cycle reflection characteristic value of each target cluster and the representative cycle time length of the corresponding target cluster. In this embodiment, the sum of the product of the production cycle reflection characteristic value of each target cluster and the representative cycle time length of the corresponding target cluster is denoted as the target production cycle time length, expressed as: , Let w be the representative period length of the w-th target cluster.

[0038] Therefore, this embodiment can obtain the target production cycle time length through the above process. After obtaining the target production cycle time length, this embodiment filters the time points in the historical monitoring period based on the target production cycle time length. The filtered time points have similar characteristics in data changes to the current monitoring time point, or are more likely to belong to similar production stages of different production cycles. The specific filtering process is as follows: in the historical monitoring period, obtain all historical monitoring time points whose time interval with the current monitoring time point is an integer multiple of the target production cycle time length, and record them as the comparative historical monitoring time points of the current monitoring time point.

[0039] After obtaining the historical monitoring times for comparison, the noise level of the normalized water quality data at the current monitoring time is reassessed by comparing the historical monitoring times, or the normalized water quality data is subjected to noise judgment processing again. Specifically, in all the sub-time periods in the sub-time period sequence corresponding to each target cluster, the sub-time period to which each historical monitoring time belongs is first obtained. Then, based on the sub-time period to which each historical monitoring time belongs and the difference in water quality data change between the historical monitoring time and the current monitoring time, the target noise index value corresponding to the normalized water quality data at the current monitoring time is obtained.

[0040] In this embodiment, the calculation logic for the change in water quality data at historical monitoring times is the same as that for the change in water quality data at the current monitoring time. The only difference is that the data used to calculate the change in water quality data at historical monitoring times is historical target water quality data, but the formula remains the same. For example, the change in water quality data at any historical monitoring time is the relative change in the historical target water quality data at that historical monitoring time compared to the historical target water quality data at the previous historical monitoring time, i.e., the change in water quality data at the current monitoring time is expressed as follows: Replace with historical target water quality data at that historical monitoring time. The expression for the change in water quality data at the previous historical monitoring time is obtained by replacing the target water quality data with the historical water quality data of the previous historical monitoring time.

[0041] In this embodiment, the specific process of obtaining the target noise index value corresponding to the normalized water quality data at the current monitoring time, based on the sub-time period to which each historical monitoring time belongs and the difference in water quality data changes between the historical monitoring time and the current monitoring time, is as follows:

[0042] First, based on the differences in water quality data changes between each historical monitoring time and the current monitoring time, a difference characterization value is obtained between each historical monitoring time and the current monitoring time. The difference characterization value between any historical monitoring time and the current monitoring time is the normalized result of the absolute value of the difference between the water quality data change at that historical monitoring time and the water quality data change at the current monitoring time. Here, the normalization function Norm() is used to achieve normalization. The larger the difference characterization value between the historical monitoring time and the current monitoring time, the greater the possibility that the normalized water quality data at the current monitoring time is noise, and vice versa. Subsequently, the difference characterization value is mainly used to measure the situation of the water quality data at the current monitoring time.

[0043] However, if the overall trend of the local data segment at the historical monitoring time differs significantly from the overall trend of the local data segment at the current monitoring time, then the calculated difference values ​​are likely not due to random noise, but rather to normal changes caused by systematic factors (such as changes in temperature, flow rate, or production process). Therefore, to ensure the accuracy of subsequent noise identification, the participation and contribution of the difference values ​​in noise assessment should be reduced. Furthermore, the lower the participation and contribution of the difference values ​​obtained from a particular historical monitoring time in noise assessment, the lower the reliability of that historical monitoring time. Therefore, this embodiment further requires evaluating the reliability values ​​of each historical monitoring time based on the difference between the overall trend of the local data segment at the historical monitoring time and the overall trend of the local data segment at the current monitoring time. The specific acquisition of the reliability values... The process is as follows: First, obtain the time period from the start time of the sub-time period to the previous monitoring time of each historical monitoring time, and record it as the historical time period to be analyzed for the corresponding historical monitoring time. That is, the historical time period to be analyzed for any historical monitoring time is the time period from the start time of the sub-time period to the previous monitoring time. Then, before the current monitoring time, obtain a continuous historical time period with the same length as the historical time period to be analyzed for each historical monitoring time, and record it as the current historical time period to be analyzed for the corresponding historical monitoring time. That is, if the length of the historical time period to be analyzed for any historical monitoring time is T0, then the continuous historical time period with a length of T0 before the current monitoring time is the current historical time period to be analyzed for the historical monitoring time. The current historical time periods to be analyzed for each historical monitoring time are all adjacent to the current monitoring time. Then, based on the differences in water quality data changes between the corresponding monitoring times in the historical time period to be analyzed and the current time period to be analyzed for each comparative historical monitoring time, the reliability characterization value of each comparative historical monitoring time is obtained. That is, for any comparative historical monitoring time, the normalized result of the mean of the absolute value of the difference in water quality data changes between the corresponding monitoring times in the historical time period to be analyzed and the current time period to be analyzed for that comparative historical monitoring time is the reliability characterization value of that comparative historical monitoring time. The expression for the reliability characterization value of that comparative historical monitoring time is as follows: R represents the total number of monitoring times in the historical time period to be analyzed or the current time period to be analyzed, compared to the historical monitoring time. This represents the change in water quality data at the r-th monitoring time within the historical time period to be analyzed, compared to the historical monitoring time. This represents the change in water quality data at the r-th monitoring time within the current time period to be analyzed, compared to historical monitoring times. The larger the value, the greater the difference in the trend between the historical time period to be analyzed and the current time period to be analyzed. In this case, the reliability of the historical time period to be analyzed is lower. The lower the reliability of the historical time period to be analyzed, the lower the participation and contribution of the difference characterization value obtained based on the historical time period to the noise assessment. In addition, if a certain historical time period to be analyzed is the start time of its sub-time period, then this embodiment defaults to taking the minimum reliability characterization value of the historical time period to be analyzed. In this embodiment, the minimum reliability characterization value is set to 0.01.

[0044] Next, based on the reliability characterization values ​​of each historical monitoring time and the difference characterization values ​​between each historical monitoring time and the current monitoring time, the initial noise index value corresponding to the normalized water quality data at the current monitoring time is obtained; and the mean of the product of the difference characterization value between each historical monitoring time and the current monitoring time and the corresponding reliability characterization value of the historical monitoring time is the initial noise index value corresponding to the normalized water quality data at the current monitoring time. The expression for the initial noise index value corresponding to the normalized water quality data at the current monitoring time is:

[0045]

[0046] G1 represents the initial noise index value corresponding to the normalized water quality data at the current monitoring time, and J represents the total number of historical monitoring times compared. Let j be the difference value between the historical monitoring time and the current monitoring time. Let be the reliability characterization value for the j-th comparison with historical monitoring time; The larger G1 is, the greater the probability that the normalized water quality data at the current monitoring time is noise, or the greater the probability that the water quality data to be treated corresponding to the normalized water quality data at the current monitoring time is noise. The smaller G1 is, the less probability that the normalized water quality data at the current monitoring time is noise, or the less probability that the water quality data to be treated corresponding to the normalized water quality data at the current monitoring time is noise. The water quality data to be treated corresponding to the normalized water quality data at the current monitoring time refers to the water quality data to be treated at the current monitoring time before normalization. The data type of the water quality data to be treated corresponding to the normalized water quality data is consistent with the data type of the corresponding normalized water quality data.

[0047] Furthermore, during oil smelting production, the wastewater discharge flow rate may suddenly increase due to equipment flushing, potentially causing sudden changes in the monitored water quality data. Therefore, to further ensure the accuracy of subsequent noise identification, this embodiment, based on the obtained initial noise index value, combines the degree of change in wastewater discharge flow rate at the current monitoring time to obtain the final noise index value, which is the target noise index value. Specifically, this embodiment will next obtain the target noise index value corresponding to the normalized water quality data at the current monitoring time based on the initial noise index value corresponding to the normalized water quality data at the current monitoring time and the difference in wastewater discharge flow rate between the current and previous monitoring times. The specific acquisition process is as follows: subtract the normalized result of the normalized wastewater discharge flow rate at the previous monitoring time from the normalized wastewater discharge flow rate at the current monitoring time. The initial noise index is denoted as the flow change characterization value at the current monitoring time. Normalization is achieved using the normalization function Norm(). Since sudden changes in data are often caused by a sudden increase in flow, there is no need to add an absolute value. The product of the initial noise index value and the flow change characterization value is calculated and denoted as the target noise index value corresponding to the normalized water quality data at the current monitoring time. The larger the initial noise index value and the larger the flow change characterization value, that is, the larger the target noise index value, the greater the probability that the normalized water quality data at the current monitoring time is noise or the greater the probability that the water quality data to be treated corresponding to the normalized water quality data at the current monitoring time is noise. The smaller the target noise index value, the smaller the probability that the normalized water quality data at the current monitoring time is noise or the smaller the probability that the water quality data to be treated corresponding to the normalized water quality data at the current monitoring time is noise.

[0048] In this embodiment, after obtaining the target noise index value corresponding to the normalized water quality data at the current monitoring time, noise judgment processing is performed on the normalized water quality data at the current monitoring time based on the target noise index value corresponding to the normalized water quality data at the current monitoring time to obtain the current target water quality data. The specific process of performing noise judgment processing on the normalized water quality data at the current monitoring time based on the target noise index value corresponding to the normalized water quality data at the current monitoring time to obtain the current target water quality data is as follows:

[0049] The system determines whether the target noise index value corresponding to the normalized water quality data at the current monitoring time is greater than a preset noise judgment threshold. If it is greater, the normalized water quality data at the current monitoring time or the water quality data to be treated corresponding to the normalized water quality data at the current monitoring time is determined to be noise data, and the normalized water quality data at the current monitoring time needs to be denoised to obtain the current target water quality data. If it is not greater, the normalized water quality data at the current monitoring time or the water quality data to be treated corresponding to the normalized water quality data at the current monitoring time is determined to be non-noise data, and the normalized water quality data at the current monitoring time can be directly recorded as the current target water quality data at the current monitoring time without denoising. In specific applications, implementers need to set a preset noise judgment threshold based on historical statistical characteristics, the range of target noise index values, and other actual conditions. If based on historical statistical characteristics, the target noise index value corresponding to the historical target water quality data at each historical monitoring time in the historical monitoring period can be calculated according to the above method, and the largest target noise index value can be selected as the preset noise judgment threshold. If based on the range of target noise index values, 0.6 can be selected as the preset noise judgment threshold.

[0050] In this embodiment, the specific process of denoising the normalized water quality data at the current monitoring time to obtain the current target water quality data is as follows:

[0051] The time series sequence STL decomposition is formed by the historical target water quality data and the normalized water quality data at each monitoring time in the current monitoring period. The trend component and periodic component of the normalized water quality data at the current monitoring time are obtained. The sum of the trend component and periodic component of the normalized water quality data at the current monitoring time is recorded as the current target water quality data. STL time series decomposition is a well-known technique.

[0052] Therefore, this embodiment can obtain the current target water quality data of each type of water quality data being monitored at the current monitoring time through the above process, and the number of current target water quality data obtained by this embodiment at the current monitoring time according to the above process is consistent with the number of water quality data types collected and monitored.

[0053] Step S003: Monitor the discharge of oil smelting wastewater based on the current target water quality data.

[0054] In this embodiment, after obtaining the current target water quality data at the current monitoring time, the discharge of oil smelting wastewater is monitored based on the current target water quality data. The specific process is as follows:

[0055] Since the target water quality data is normalized data, before monitoring the discharge of oil smelting wastewater, it is necessary to normalize and restore the current target water quality data at the current monitoring time. The data obtained by normalization and restoration is recorded as the current target restored water quality data at the current monitoring time. Normalization and restoration refers to the process of restoring the normalized data to its original numerical scale through inverse transformation. It is the reverse process of normalization operation in data preprocessing. If the above-mentioned data collection in this embodiment uses max-min normalization for normalization, then the current target water quality data is normalized and restored using max-min normalization.

[0056] Then, it is determined whether any of the current target restored water quality data at the current monitoring time exceeds the preset normal range corresponding to the water quality data type of the current target restored water quality data. If so, it indicates that the water quality data monitored at the current monitoring time does not meet the discharge requirements, and the discharge of oil smelting wastewater at the current monitoring time is determined to be abnormal. If none of them exceed the limits, it indicates that the water quality data monitored at the current monitoring time meets the discharge requirements, and the discharge of oil smelting wastewater at the current monitoring time is determined to be normal. In specific applications, the preset normal range needs to be determined based on the actual situation such as the specified wastewater discharge standards. For example, in the field of oil smelting wastewater discharge, if COD (Chemical Oxygen Demand) is required... If COD ≤ 500 mg / L, suspended solids (SS) ≤ 400 mg / L, oil content ≤ 100 mg / L, and pH is 6 to 9, then the preset normal range for COD is ≤ 500 mg / L, the preset normal range for suspended solids (SS) is ≤ 400 mg / L, the preset normal range for oil content is ≤ 100 mg / L, and the preset normal range for pH is 6 to 9. mg / L refers to milligrams per liter. Therefore, among all the current target reduced water quality data at the current monitoring time, only when none of them exceed the corresponding preset normal range can it be considered that the wastewater discharge is normal. Otherwise, it indicates that the wastewater discharge is abnormal and will affect the ecological safety of the water body.

[0057] Thus, this embodiment completes the intelligent monitoring of wastewater discharge from oil smelting. Furthermore, this embodiment uses the difference in water quality data between historical monitoring times and the current monitoring time, which are similar to the current monitoring time and are selected based on the analyzed production cycle length, as the basis for noise judgment and processing. Based on the data obtained after noise judgment and processing, the monitoring of wastewater discharge from oil smelting can avoid the interference of noise on the monitoring accuracy of wastewater discharge from oil smelting, thereby improving the accuracy of monitoring wastewater discharge from oil smelting.

[0058] In summary, this embodiment first acquires normalized auxiliary influence data at historical monitoring times and normalized water quality data at the current monitoring time. Then, it determines whether the change in water quality data at the current monitoring time exceeds a preset change threshold. If not, the normalized water quality data at the current monitoring time is recorded as the current target water quality data. If it exceeds the threshold, the results of clustering historical monitoring times based on the normalized auxiliary influence data and local fitting slope at historical monitoring times are used to obtain sub-time period sequences corresponding to each target cluster. The target production cycle length is obtained based on the standard deviation and median of the time difference between adjacent sub-time periods in the sub-time period sequence. Historical monitoring times with an interval equal to an integer multiple of the target production cycle length at the current monitoring time are selected as comparison historical monitoring times. Based on the difference in water quality data changes between the comparison historical monitoring times and the current monitoring time, the target noise index value corresponding to the normalized water quality data at the current monitoring time is obtained. Noise judgment processing is performed on the normalized water quality data at the current monitoring time based on the target noise index value to obtain the current target water quality data. Finally, the discharge of oil smelting wastewater is monitored based on the current target water quality data. Furthermore, this embodiment uses the target production cycle duration as the screening criterion, extracts and compares the difference in water quality data changes between historical monitoring times and current monitoring times as the basis for noise judgment, and finally conducts monitoring of oil smelting wastewater discharge based on the processed data. This can avoid the interference of noise on the monitoring accuracy of oil smelting wastewater discharge and significantly improve the accuracy of monitoring oil smelting wastewater discharge.

[0059] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for intelligent monitoring of wastewater discharge from oil smelting, characterized in that, The method includes the following steps: Obtain normalized auxiliary impact data at historical monitoring times and normalized water quality data at the current monitoring time. The normalized auxiliary impact data includes normalized wastewater temperature and normalized wastewater discharge flow rate. The system determines whether the change in water quality data at the current monitoring time is greater than a preset change threshold. If it is not greater, the normalized water quality data at the current monitoring time is recorded as the current target water quality data. If it is greater, the system clusters the historical monitoring times based on the normalized auxiliary influence data and local fitting slope at the historical monitoring times to obtain the sub-time period sequence corresponding to each target cluster. The system obtains the target production cycle time length based on the standard deviation and median of the time difference between adjacent sub-time periods in the sub-time period sequence. The system selects historical monitoring times that are an integer multiple of the target production cycle time length from the current monitoring time as comparison historical monitoring times. Based on the difference in water quality data change between the comparison historical monitoring times and the current monitoring time, the system obtains the target noise index value corresponding to the normalized water quality data at the current monitoring time. The system performs noise judgment processing on the normalized water quality data at the current monitoring time based on the target noise index value to obtain the current target water quality data. The discharge of wastewater from oil smelting is monitored based on the current target water quality data.

2. The intelligent monitoring method for wastewater discharge from oil smelting as described in claim 1, characterized in that, The change in water quality data at the current monitoring time is the relative change between the normalized water quality data at the current monitoring time and the historical target water quality data at the previous monitoring time. The change in water quality data at the historical monitoring time is the relative change between the historical target water quality data at the historical monitoring time and the historical target water quality data at the previous historical monitoring time.

3. The intelligent monitoring method for wastewater discharge from oil smelting as described in claim 1, characterized in that, The methods for obtaining the sub-time period sequences corresponding to each target cluster include: Based on the feature vectors of each historical monitoring moment, all historical monitoring moments are clustered to obtain initial clusters. Isolated monitoring moments within the initial clusters are migrated to the initial clusters of their nearest neighbor monitoring moments. All migrated initial clusters are recorded as target clusters. The time periods formed by consecutive historical monitoring moments in each target cluster are recorded as sub-time periods corresponding to the target cluster. The result of arranging all sub-time periods corresponding to each target cluster in chronological order is recorded as the sub-time period sequence corresponding to the target cluster. The feature vector at any historical monitoring moment includes the normalized wastewater temperature, the normalized result of the local wastewater temperature fitting slope, the normalized wastewater discharge flow rate, and the normalized result of the local wastewater discharge flow rate fitting slope at the historical monitoring moment.

4. The intelligent monitoring method for wastewater discharge from oil smelting as described in claim 1, characterized in that, Methods for obtaining the target production cycle time include: The set of time differences between all adjacent sub-time periods in the sub-time period sequence corresponding to each target cluster is denoted as the time difference set corresponding to the target cluster. The median of the time difference set corresponding to each target cluster is taken as the representative period length of the target cluster. The result of reverse normalization and then proportional normalization of the standard deviation of the time difference set corresponding to each target cluster is denoted as the production cycle reflection characteristic value of the corresponding target cluster. The sum of the product of the production cycle reflection characteristic value of each target cluster and the representative cycle time length of the corresponding target cluster is recorded as the target production cycle time length.

5. The intelligent monitoring method for wastewater discharge from oil smelting as described in claim 1, characterized in that, Methods for obtaining target noise index values ​​include: The time period from the start time of each sub-time period to the previous monitoring time of the corresponding historical monitoring time is recorded as the historical time period to be analyzed for the corresponding historical monitoring time. The continuous time period before the current monitoring time that is equal in length to the historical time period to be analyzed for each historical monitoring time is recorded as the current time period to be analyzed for the corresponding historical monitoring time. Based on the differences in water quality data changes between the corresponding monitoring times at each comparative historical monitoring time point and the current time point to be analyzed, a reliability characterization value for each comparative historical monitoring time point is obtained. Based on the reliability characterization values ​​of each comparative historical monitoring time and the difference in water quality data changes between each comparative historical monitoring time and the current monitoring time, the initial noise index value corresponding to the normalized water quality data at the current monitoring time is obtained. Based on the initial noise index value and the difference in normalized wastewater discharge flow between the current monitoring time and the previous monitoring time, the target noise index value corresponding to the normalized water quality data at the current monitoring time is obtained.

6. The intelligent monitoring method for wastewater discharge from oil smelting as described in claim 5, characterized in that, The reliability characterization value of any comparative historical monitoring time is the normalized result of the mean of the absolute value of the difference between the water quality data change amount between the historical time period to be analyzed and the corresponding monitoring time in the current time period to be analyzed.

7. The intelligent monitoring method for wastewater discharge from oil smelting as described in claim 5, characterized in that, The method for obtaining the initial noise index value corresponding to the normalized water quality data at the current monitoring time includes: The normalized result of the absolute value of the difference between the water quality data changes between each historical monitoring time and the current monitoring time is recorded as the difference characterization value between the corresponding historical monitoring time and the current monitoring time. The mean value of the product of the difference characterization value between each historical monitoring time and the current monitoring time and the reliability characterization value of the corresponding historical monitoring time is recorded as the initial noise index value corresponding to the normalized water quality data at the current monitoring time.

8. The intelligent monitoring method for wastewater discharge from oil smelting as described in claim 5, characterized in that, A method for obtaining the target noise index value corresponding to the normalized water quality data at the current monitoring time based on the initial noise index value and the difference in normalized wastewater discharge flow between the current monitoring time and the previous monitoring time includes: The normalized wastewater discharge flow rate at the current monitoring time is subtracted from the normalized wastewater discharge flow rate at the previous monitoring time, and the result is recorded as the flow rate change characterization value at the current monitoring time. The product of the initial noise index value and the flow rate change characterization value is recorded as the target noise index value corresponding to the normalized water quality data at the current monitoring time.

9. The intelligent monitoring method for wastewater discharge from oil smelting as described in claim 1, characterized in that, A method for obtaining current target water quality data by performing noise assessment processing on normalized water quality data at the current monitoring time based on the target noise index value includes: Determine whether the target noise index value is greater than a preset noise judgment threshold. If it is greater, perform noise reduction processing on the normalized water quality data at the current monitoring time to obtain the current target water quality data. If it is not greater, record the normalized water quality data at the current monitoring time as the current target water quality data.

10. The intelligent monitoring method for wastewater discharge from oil smelting as described in claim 9, characterized in that, A method for denoising the normalized water quality data at the current monitoring time to obtain the current target water quality data includes: The sum of the trend component and periodic component of the normalized water quality data at the current monitoring time obtained from STL time series decomposition is denoted as the current target water quality data.

Images