Water quality on-line detection method

Abstract

The invention provides a water quality online detection method which comprises the following steps: (1) arranging a plurality of monitoring sections in a water flow direction, arranging a detection assembly on each monitoring section, and establishing a water quality detection system to obtain water quality detection values of the monitoring sections; (2) establishing corresponding water quality prediction models through the water quality detection values of the monitoring sections, wherein the water quality prediction models of different monitoring sections form a water quality prediction system; (3) carrying out fusion operation on the water quality detection system and the water quality prediction system; (4) analyzing a water quality prediction value of the water quality prediction system, and performing early warning and model updating; and (5) analyzing the water quality detection value of the water quality detection system, and carrying out early warning and equipment maintenance. According to the water quality on-line detection method, water quality detection and prediction are fused and cooperatively applied, the detection and prediction complement each other and support each other, monitoring data are generated as a whole, and the detection frequency and the operation cost of an on-line detection system are greatly reduced.

Description

Technical Field

[0001] The invention belongs to the field of water quality detection, and in particular relates to an online water quality detection method. Background Art

[0002] As environmental protection supervision becomes increasingly stringent, online detection instruments have been widely used in the monitoring of inlet / outlet water and process water quality of sewage treatment plants to ensure that operators can understand water quality changes in a timely manner, so as to adjust the operating parameters of sewage treatment facilities and instruments in a timely manner to ensure that emissions meet standards.

[0003] Traditional online detection instruments based on chemical methods are all periodic sampling and testing (sampling once every 1-2 hours) to perceive water quality changes in real time. The high detection frequency, coupled with the large number of detection instruments, significantly increases its operation and maintenance costs. For example, online COD, TP, TN and ammonia nitrogen based on the national standard method require at least 3 reagents for each indicator detection, and a single test produces 4-6mL of waste liquid. The cost of purchasing these reagents and treating the waste liquid (disposing of it as hazardous waste), as well as the power consumption of equipment operation, parts replacement, maintenance labor and other costs, have reduced the benefits of sewage treatment plants to a certain extent.

[0004] At the same time, the more online detection equipment is invested and the higher the detection frequency, the higher the frequency of failure of the online detection system as a whole, which in turn affects the effective operation of the water plant's water system. In serious cases, misleading control signals may be generated, causing the effluent water quality to exceed the control target, which reduces the stability and reliability of the water system. Summary of the invention

[0005] In view of this, the present invention aims to overcome the defects in the prior art and proposes an online water quality detection method.

[0006] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0007] The present invention provides a method for online water quality detection, comprising the following steps:

[0008] (1) Setting up a number of monitoring sections in the direction of water flow, setting up detection components on each monitoring section, establishing a water quality detection system, and obtaining water quality detection values ​​of the monitoring sections;

[0009] (2) Establishing a corresponding water quality prediction model based on the water quality detection value of the monitoring section. The water quality prediction models of different monitoring sections form a water quality prediction system to obtain the water quality prediction value of the monitoring section;

[0010] (3) The water quality detection system and the water quality prediction system are integrated and operated;

[0011] (4) Analyze the water quality prediction values ​​of the water quality prediction system, conduct early warning and model update;

[0012] (5) Analyze the water quality test values ​​of the water quality detection system and carry out early warning and equipment maintenance.

[0013] Furthermore, the coordinates of the monitoring section in step (1) are calculated by formula (I):

[0014]

[0015] Where X is the position coordinate of the water flow direction, the water inlet end of the treatment unit is the zero point, and the water flow direction is the positive direction; K is the water quality change value, L is the water flow length of the treatment unit, HRT is the hydraulic retention time of the treatment unit, and f(K) is the mapping function of time and K.

[0016] Furthermore, the water quality change value K is calculated by formula (II):

[0017] K=-(dS / dT) Formula (Ⅱ)

[0018] Wherein, S is a water quality index, T is time, and the moment when the treatment starts is taken as time zero; the water quality index is at least one of COD, TN, TP or NH3-N.

[0019] Furthermore, we set K max is the maximum value of K of the processing unit; K = K max The calculated coordinate is X max , when K≤0, the calculated coordinate is X min , K = 0.5 × K max The calculated coordinate is X mid ; when X≤X max The control monitoring section is set in the interval of X≥X min The interval setting effect monitoring section, in X mid ≤X≤X min Set the dynamic monitoring section in the interval.

[0020] Furthermore, the detection component in step (1) is at least one of a sampling device, a sample pretreatment device or a water quality detection instrument; the sampling period of the sampling device is 1-2 hours, and the monitoring period of the water quality monitoring instrument is 1-2 hours.

[0021] Furthermore, the water quality prediction model in step (2) is an algorithm model with incremental learning characteristics, which is at least one of a long short-term memory model, an incremental decision tree or an online support vector machine; the control value H of the water quality prediction model in step (2) is 0-0.5. After the water quality detection system has been running for 3-6 months, the water quality detection data generated by each monitoring section is used for training and evaluation of the water quality prediction model.

[0022] Furthermore, the fusion operation step in step (3) is specifically as follows: a water quality prediction value is generated at each hour; when the relationship between the water quality prediction value and the water quality detection value satisfies formula (III), a water quality detection is performed after one monitoring cycle; when the relationship between the water quality prediction value and the water quality detection value does not satisfy formula (III), a water quality detection is performed in the next monitoring cycle; formula (III) is as follows:

[0023]

[0024] Among them, Dp is the water quality prediction value and Dc is the water quality detection value.

[0025] Furthermore, the early warning step in step (4) is specifically as follows: a comprehensive analysis is performed on the water quality prediction value, and when it satisfies formula (IV), an early warning is issued, and incremental learning of the prediction model is started to update the prediction model; formula (IV) is as follows:

[0026] and

[0027] Among them, m is the amount of valid water quality detection data within 30 days, ranging from 50 to 300, i is the data ordinal number, R is the prediction deviation, Ra is the mean deviation of the m prediction values, and F is the standard deviation of the deviation of the m prediction values.

[0028] Furthermore, the early warning step described in step (5) is specifically: calculating the water quality detection value at any time, and when it satisfies formula (V), issuing an early warning and prompting the operator to perform equipment maintenance; formula (V) is as follows:

[0029] and

[0030] Among them, t is any moment, and t-1 is the moment before t.

[0031] Furthermore, when the water quality detection system is abnormal or fails, the abnormal value or fault value generated by it is marked, and the predicted value is used instead of the abnormal value or fault detection value to improve the stability of the production operation of the sewage treatment plant.

[0032] Compared with the prior art, the present invention has the following advantages:

[0033] The online water quality detection method of the present invention integrates the detection and prediction of water quality and applies them in a coordinated manner. The two complement and support each other and generate monitoring data as a whole, which greatly reduces the detection frequency and operating cost of the online detection system.

[0034] The detection system and prediction model in the online water quality detection method described in the present invention monitor each other, realize "mutual warning" between the detection system and the prediction model, improve the operation and maintenance efficiency of the detection system through detection and warning, and update and improve the prediction model through prediction and warning and incremental learning, so that its prediction accuracy meets the operation and maintenance requirements.

[0035] The detection value and the predicted value in the online water quality detection method described in the present invention can "corroborate and replace each other", which greatly improves the reliability and stability of the monitoring data, provides a more solid data foundation for the stable and standard operation of the sewage treatment plant, and provides a solid guarantee for the effective operation of the water system; by establishing a monitoring section through this method, the monitoring data can also better play a guiding role in the operation and control of the sewage treatment plant. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The figure is a flow chart of the online water quality detection method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] Unless otherwise defined, the technical terms used in the following examples have the same meanings as those generally understood by those skilled in the art to which the present invention belongs. The test reagents used in the following examples, unless otherwise specified, are all conventional biochemical reagents; the experimental methods, unless otherwise specified, are all conventional methods.

[0038] The present invention will be described in detail below with reference to the embodiments.

[0039] Example 1

[0040] The treatment process of a municipal sewage treatment plant is: screen + grit chamber + anaerobic chamber + anoxic chamber + aerobic chamber + secondary sedimentation chamber + disinfection chamber. A carbon source is added to the anoxic chamber, and a phosphorus removal agent (PAC) is added to the effluent of the aerobic chamber. The effluent meets the Class A discharge standard.

[0041] The monitoring section and its detection indexes and prediction indexes designed according to the water quality online detection method of the present invention are shown in Table 1.

[0042] Table 1 Monitoring sections and detection and prediction indicators

[0043]

[0044] According to the monitoring sections and indicators in Table 1, a water quality detection system and a corresponding water quality prediction model are established.

[0045] The water quality detection system samples and tests once an hour, and generates 24 test values ​​in a day. After the system has been running for 3 months, a single water quality indicator generates 2160 test values. The water quality data of each monitoring section is cleaned and sorted to form a data set for algorithm model training. The data set is divided into a training set (80% of the data volume) and a test set (20% of the data volume). The prediction deviation rate control value H of each water quality indicator is shown in Table 2.

[0046] Table 2 Deviation control values ​​of predicted values ​​of indicators in each monitoring section

[0047]

[0048] When the prediction deviation statistics meet the requirements of formula (IV), the incremental learning of the water quality prediction model is started to optimize the prediction model. When the detection value and the prediction value meet the requirements of formula (V), a detection warning is issued to remind the operation and maintenance personnel to pay attention to the operation status of the online detection equipment and restore it in time if a fault or abnormality is found.

[0049] Monitoring data generation: When the detection value is generated, a decision analysis is performed on the time of the next sampling test: if the current prediction value and the detection value meet the requirements of formula (III), then sampling and testing will be carried out after one cycle (that is, after 2 hours); otherwise, sampling and testing will be carried out immediately in the next monitoring cycle (that is, after 1 hour). As shown in Table 3, the indicators of each monitoring section generate one prediction data every hour. When there is a detection value at the hour, the detection value is used as the monitoring data, and when there is no detection value at the hour, the prediction value is used as the monitoring data. The monitoring data at the hour is sent to the control system (ICS) to provide data support for intelligent operation control.

[0050] Table 3 Monitoring data fusion table (taking the COD index of aerobic effect monitoring section on a certain day as an example)

[0051]

[0052]

[0053] According to the conventional detection frequency of this embodiment, the method of the present invention can save the loss and consumption of detection equipment and reagents. The theoretical upper limit of the saving is 75%, and the actual saving is about 42%, which can significantly reduce the operating cost.

[0054] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A water quality online detection method, characterized in that: The steps include: (1) Setting up a number of monitoring sections in the direction of water flow, setting up detection components on each monitoring section, establishing a water quality detection system, and obtaining water quality detection values ​​of the monitoring sections; (2) Establishing a corresponding water quality prediction model based on the water quality detection value of the monitoring section. The water quality prediction models of different monitoring sections form a water quality prediction system to obtain the water quality prediction value of the monitoring section; (3) The water quality detection system and the water quality prediction system are integrated and operated; (4) Analyze the water quality prediction values ​​of the water quality prediction system, conduct early warning and model update; (5) Analyze the water quality test values ​​of the water quality detection system and carry out early warning and equipment maintenance.

2. The method for online water quality detection according to claim 1, characterized in that: The coordinates of the monitoring section in step (1) are calculated by formula (I): Where X is the position coordinate of the water flow direction, the water inlet end of the treatment unit is the zero point, and the water flow direction is the positive direction; K is the water quality change value, L is the water flow length of the treatment unit, HRT is the hydraulic retention time of the treatment unit, and f(K) is the mapping function of time and K.

3. The water quality online detection method according to claim 2, characterized in that: The water quality change value K is calculated by formula (II): K=-(dS / dT) Formula (Ⅱ) Wherein, S is a water quality index, T is time, and the moment when the treatment starts is taken as time zero; the water quality index is at least one of COD, TN, TP or NH3-N.

4. The method for online water quality detection according to claim 3, characterized in that: Set K max is the maximum value of K of the processing unit; K = K max The calculated coordinate is X max , when K≤0, the calculated coordinate is X min , K = 0.5K max The calculated coordinate is X mid ; when X≤X max The control monitoring section is set in the interval of X≥X min The interval setting effect monitoring section, in X mid ≤X≤X min Set the dynamic monitoring section in the interval.

5. The water quality online detection method according to claim 1, characterized in that: The detection component in step (1) is at least one of a sampling device, a sample pretreatment device or a water quality detection instrument; the sampling period of the sampling device is 1-2 hours, and the monitoring period of the water quality monitoring instrument is 1-2 hours.

6. The method for online water quality detection according to claim 1, characterized in that: The water quality prediction model in step (2) is an algorithm model with incremental learning characteristics, which is at least one of a long short-term memory network, an incremental decision tree or an online support vector machine; the deviation control value H of the water quality prediction model in step (2) is 0-0.

5.

7. The water quality online detection method according to claim 1, characterized in that: The fusion operation step in step (3) is specifically as follows: a water quality prediction value is generated at each hour; when the relationship between the water quality prediction value and the water quality detection value satisfies formula (III), a water quality detection is performed after one monitoring cycle; when the relationship between the water quality prediction value and the water quality detection value does not satisfy formula (III), a water quality detection is performed in the next monitoring cycle; formula (III) is as follows: Among them, Dp is the water quality prediction value and Dc is the water quality detection value.

8. The water quality online detection method according to claim 1, characterized in that: The early warning step in step (4) is specifically as follows: a comprehensive analysis is performed on the water quality prediction value, and when it satisfies formula (IV), an early warning is issued, and incremental learning of the prediction model is started to update the prediction model; formula (IV) is as follows: Among them, m is the amount of valid water quality detection data within 30 days, ranging from 50 to 300, i is the data ordinal, R is the prediction deviation, Ra is the average deviation of the m prediction values, and F is the standard deviation of the deviation of the m prediction values.

9. The water quality online detection method according to claim 1, characterized in that: The early warning step described in step (5) is specifically: calculating the water quality detection value at any time, and when it satisfies formula (V), issuing an early warning and prompting the operator to perform equipment maintenance; formula (V) is as follows: Among them, t is any moment, and t-1 is the moment before t.

10. The method for online water quality detection according to claim 1, characterized in that: When the water quality detection system is abnormal or fails, the abnormal value or fault value generated by it is marked, and the predicted value is used instead of the abnormal value or fault detection value.

Images