Sewage treatment water quality prediction method and system

Abstract

The invention discloses a sewage treatment water quality prediction method and system, and belongs to the technical field of sewage treatment.The sewage treatment water quality prediction method includes the steps that sewage inlet flow and water quality indexes are collected in real time, the pollution condition of inlet water is reflected, sewage residence time is calculated according to the sewage inlet flow at each moment, and the sewage residence time is compared with reference time; the method comprises the following steps: constructing a residence time deviation sequence, constructing a gas explosion quantity deviation sequence according to the difference between the aeration quantity provided by the history and the required gas explosion quantity, evaluating the water quality treatment adaptability according to the deviation between the residence time and the gas explosion quantity, and reflecting the water quality treatment difficulty in the time period by the water quality index change. The water quality treatment fitness score is obtained by combining the sewage water quality treatment process and the water quality treatment difficulty, and is used for evaluating inflow water in the time period and predicting the water quality treatment fitness score in advance, so that workers are warned to adjust sewage treatment parameters in time.

Description

Technical Field

[0001] The present invention relates to the technical field of sewage treatment, and in particular to a sewage treatment water quality prediction method and system. Background Art

[0002] Wastewater treatment is an important part of modern urban environmental management. With the acceleration of urbanization and the advancement of industrial development, the amount of wastewater discharged has increased significantly, and water quality issues have become increasingly serious. In order to protect water resources and improve water quality, a series of wastewater treatment measures have been adopted worldwide. These measures not only include physical and chemical treatment methods, but also emphasize the importance of biological treatment in removing organic matter and nutrients in wastewater.

[0003] Secondary biological treatment is one of the key links in sewage treatment. Through the metabolic activities of microorganisms, the chemical oxygen demand (COD) and biological oxygen demand (BOD) in sewage can be effectively reduced to meet the discharge standards. Although the water quality indicators in the sewage treatment process can be directly detected, there will be a certain lag and uncertainty in adjusting the parameters of sewage treatment according to the real-time water quality indicators, which will also prolong the sewage treatment time and make it impossible to predict the adaptability of the current water quality treatment process. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, a method and system for predicting water quality in sewage treatment provided by the present invention solves the problem that the prior art cannot predict the adaptability of the current water quality treatment process.

[0005] In order to achieve the above-mentioned purpose of the invention, the technical solution adopted by the present invention is: a method for predicting water quality of sewage treatment, comprising the following steps: S1. Collect the sewage inflow flow and inflow water quality indicators at each time, wherein the inflow water quality indicators include: inflow chemical oxygen demand, inflow biological oxygen demand and inflow ammonia nitrogen concentration; S2. Obtain multiple sewage residence times in the biochemical pool according to the sewage inflow flow at each moment; S3, constructing a residence time deviation sequence according to the difference between multiple sewage residence times and reference times; S4. Constructing a gas volume deviation sequence according to the difference between the gas volume provided at the historical moment and the required gas volume; S5. Construct a water quality index change sequence according to the influent water quality index at each moment; S6. According to the residence time deviation sequence, the aeration volume deviation sequence and the water quality index change sequence, the water quality treatment fitness score is predicted based on the water quality prediction model.

[0006] Furthermore, the reference time in S3 includes: minimum stay time T min , maximum residence time Tmax and standard residence time T s , S3 includes the following sub-steps: S31, in (T min +T t ) / 2≤T t ≤(T max +T t ) / 2, the corresponding sewage retention time T is assigned t The time deviation coefficient is 0; S32, in T min ≤T t <(T min +T s ) / 2, calculate the corresponding sewage retention time T t Time deviation coefficient: , where θ is the time deviation coefficient; S33, in (T max +T s ) / 2<T t ≤T max Calculate the corresponding sewage retention time T t Time deviation coefficient: ; S34, in T t <T min Calculate the corresponding sewage retention time T t Time deviation coefficient: ; S35, in T t >T max Calculate the corresponding sewage retention time T t Time deviation coefficient: ; S36, each time deviation coefficient is calculated according to the sewage residence time T t The occurrence order is arranged to obtain the residence time deviation sequence.

[0007] Furthermore, the S4 comprises the following sub-steps: S41, calculating the required explosion volume for the inlet water quality index at each moment, and obtaining a required explosion volume sequence; S42, obtaining a comparison explosion volume sequence according to the explosion volume provided at each historical moment; S43, subtracting the comparison explosion volume sequence from the required explosion volume sequence to obtain an explosion volume difference sequence; S44. Calculate the explosion volume deviation coefficient according to each explosion volume difference in the explosion volume difference sequence, and construct the explosion volume deviation sequence.

[0008] Furthermore, the formula for calculating the explosion volume deviation coefficient in S44 is: , where μ is the explosion volume deviation coefficient, △Q a is the difference in explosion volume, and | | is the absolute value.

[0009] Furthermore, the S5 comprises the following sub-steps: S51, subtracting the inlet water quality indicators at adjacent moments to obtain a water quality indicator difference; S52. Calculate the water quality index variation coefficient based on the water quality index difference and construct a water quality index variation sequence.

[0010] Furthermore, the water quality index difference in S51 is: , where △C t is the water quality index difference at time t, C COD,in,t+1 is the chemical oxygen demand of the influent at time t+1, C BOD,in,t+1 is the biological oxygen demand of the influent at time t+1, C NH,in,t+1 is the influent ammonia nitrogen concentration at time t+1, C COD,in,t is the chemical oxygen demand of the influent at time t, C BOD,in,t is the biological oxygen demand of the influent at the tth moment, C NH,in,t is the influent ammonia nitrogen concentration at the tth moment, and t is the moment number.

[0011] Furthermore, the formula for calculating the coefficient of variation of water quality index in S52 is: , where ε is the coefficient of variation of water quality index and | | is the absolute value operation.

[0012] Furthermore, the water quality prediction model in S6 includes: a time deviation processing channel, a gas volume deviation processing channel, a water quality index change processing channel and an output layer; The input end of the time deviation processing channel is used to input the residence time deviation sequence to obtain the time impact value; the input end of the aeration volume deviation processing channel is used to input the aeration volume deviation sequence to obtain the aeration volume impact value; the input end of the water quality index change processing channel is used to input the water quality index change sequence to obtain the water quality index impact value; the output layer is used to weight the time impact value, the aeration volume impact value and the water quality index impact value to obtain the water quality treatment fitness score.

[0013] Furthermore, the time deviation processing channel, the explosion volume deviation processing channel and the water quality index change processing channel all include: a maximum pooling layer, a significant value enhancement layer and a fully connected layer connected in sequence, wherein the maximum pooling layer is a one-dimensional pooling layer, the pooling window size is 4, and the stride is 4; The expression of the saliency enhancement layer is: , where y i is the i-th enhanced saliency value output by the saliency enhancement layer, x iis the ith significant value output by the maximum pooling layer, and e is a natural constant.

[0014] A sewage treatment water quality prediction system, comprising: a collection unit, a sewage residence time acquisition unit, a first sequence construction unit, a second sequence construction unit, a third sequence construction unit and a prediction unit; The collection unit is used to collect sewage inflow flow and inflow water quality indicators at each moment; The sewage residence time acquisition unit is used to obtain multiple sewage residence times in the biochemical pool according to the sewage inflow flow rate at each moment; The first sequence construction unit is used to construct a residence time deviation sequence according to the difference between a plurality of sewage residence times and a reference time; The second sequence construction unit is used to construct a gas volume deviation sequence according to the difference between the gas volume provided at the historical moment and the required gas volume; The third sequence construction unit is used to construct a water quality index change sequence according to the influent water quality index at each moment; The prediction unit is used to predict the water quality treatment fitness score based on the water quality prediction model according to the residence time deviation sequence, the explosion volume deviation sequence and the water quality index change sequence.

[0015] The beneficial effects of the present invention are as follows: the present invention effectively reflects the pollution of the influent by collecting the sewage influent flow and water quality indicators (including chemical oxygen demand, biological oxygen demand and ammonia nitrogen concentration) in real time. According to the sewage influent flow at each moment, the residence time of the sewage in the biochemical pool is calculated, and it is compared with the reference time to construct the residence time deviation sequence. This process can reveal the adaptability of sewage treatment time.

[0016] At the same time, the present invention also takes into account the gap between the actual aeration volume provided during the sewage treatment process and the required aeration volume, and constructs an aeration volume deviation sequence. The establishment of this sequence is helpful to evaluate the impact of aeration conditions on the sewage treatment effect.

[0017] In addition, changes in water quality indicators reflect the difficulty of water quality treatment. Rapid changes in water quality indicators may increase the difficulty of treatment and affect the treatment effect. By combining the residence time deviation, aeration volume deviation and water quality indicator changes, the adaptability of sewage treatment is comprehensively evaluated, and finally the water quality treatment fitness score is obtained. The water quality treatment fitness score is used to evaluate the influent during this period of time, and predict the fitness of the water quality treatment process in advance, so as to warn the staff to make timely adjustments to the parameters of sewage treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of a method for predicting water quality of sewage treatment; Figure 2Schematic diagram of the water quality prediction model. DETAILED DESCRIPTION

[0019] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0020] Embodiment 1, as Figure 1 As shown, a method for predicting water quality of sewage treatment includes the following steps: S1. Collect the sewage inflow flow and inflow water quality indicators at each time, wherein the inflow water quality indicators include: inflow chemical oxygen demand, inflow biological oxygen demand and inflow ammonia nitrogen concentration; S2. Obtain multiple sewage residence times in the biochemical pool according to the sewage inflow flow at each moment; S3, constructing a residence time deviation sequence according to the difference between multiple sewage residence times and reference times; S4. Constructing a gas volume deviation sequence according to the difference between the gas volume provided at the historical moment and the required gas volume; S5. Construct a water quality index change sequence according to the influent water quality index at each moment; S6. According to the residence time deviation sequence, the aeration volume deviation sequence and the water quality index change sequence, the water quality treatment fitness score is predicted based on the water quality prediction model.

[0021] In this embodiment, the calculation formula for the sewage residence time in the biochemical pool in S2 is: , where T t is the sewage residence time corresponding to the sewage inflow flow at time t, Q t is the sewage inflow flow rate at the tth moment, and V is the volume of the biochemical pool.

[0022] In this embodiment, the reference time in S3 includes: minimum stay time T min , maximum residence time T max and standard residence time T s , S3 includes the following sub-steps: S31, in (T min +T t ) / 2≤T t ≤(T max +T t ) / 2, the corresponding sewage retention time T is assigned t The time deviation coefficient is 0; S32, in Tmin ≤T t <(T min +T s ) / 2, calculate the corresponding sewage retention time T t Time deviation coefficient: , where θ is the time deviation coefficient; S33, in (T max +T s ) / 2<T t ≤T max Calculate the corresponding sewage retention time T t Time deviation coefficient: ; S34, in T t <T min Calculate the corresponding sewage retention time T t Time deviation coefficient: ; S35, in T t >T max Calculate the corresponding sewage retention time T t Time deviation coefficient: ; S36, each time deviation coefficient is calculated according to the sewage residence time T t The occurrence order is arranged to obtain the residence time deviation sequence.

[0023] In the secondary treatment process of the sewage treatment plant: minimum retention time T min = 4 hours, maximum residence time T max = 8 hours, standard residence time T s = 6 hours.

[0024] 1. When the actual stay time T t When in the ideal range (5-7 hours): t = 6 hours, at which time θ = 0, indicating that it is in the optimal residence time range; 2. When the stay time is shorter (4-5 hours): t = 4.5 hours, θ = 0.4, indicating a slight time deficit; 3. When the stay time is slightly longer (7-8 hours): T t = 7.5 hours, θ = 0.4, indicating a slight overtime; 4. When the residence time is seriously insufficient (<4 hours): T t = 3 hours, θ = 0.67, indicating a more serious time shortage; 5. When the stay time is too long (>8 hours): T t= 9 hours, θ = 0.83, indicating a serious over-long period of time.

[0025] In the process of sewage treatment, too short a residence time may result in the failure to fully degrade organic matter and nutrients in the sewage, while too long a residence time may result in reduced sewage treatment efficiency and the risk of secondary pollution. The present invention gives an optimal evaluation (θ=0) for the residence time within the ideal range, a smaller deviation coefficient for slight deviations, and a larger deviation coefficient for severe deviations.

[0026] In this embodiment, S4 includes the following sub-steps: S41, calculating the required explosion volume for the inlet water quality index at each moment, and obtaining a required explosion volume sequence; S42, obtaining a comparison explosion volume sequence according to the explosion volume provided at each historical moment; S43, subtracting the comparison explosion volume sequence from the required explosion volume sequence to obtain an explosion volume difference sequence; S44. Calculate the explosion volume deviation coefficient according to each explosion volume difference in the explosion volume difference sequence, and construct the explosion volume deviation sequence.

[0027] In this embodiment, the gas volume data provided in S42 is derived from: the gas volume provided to the sewage during the sewage treatment process in the most recent historical period, that is, the historical gas volume near the current time.

[0028] In this embodiment, the formula for calculating the required explosion volume in S41 is: , where Q a is the required explosion volume, C COD,in,t is the chemical oxygen demand of the influent at time t, C BOD,in,t is the biological oxygen demand of the influent at the tth moment, C NH,in,t is the influent ammonia nitrogen concentration at time t, t is the time number, K1 is the influent chemical oxygen demand C COD,in,t The proportional constant, K2 is the biological oxygen demand C of the influent BOD,in,t The proportional constant, K3 is the influent ammonia nitrogen concentration C NH,in,t The proportionality constant.

[0029] In this embodiment, other linear regression models may be used to calculate the required explosion volume, and K1, K2 and K3 may be determined through experiments or experience.

[0030] In this embodiment, the formula for calculating the explosion volume deviation coefficient in S44 is: , where μ is the explosion volume deviation coefficient, △Q a is the difference in explosion volume, and | | is the absolute value.

[0031] When the difference in the amount of explosion is greater than or equal to 0, the amount of explosion provided by the present invention is sufficient. When the difference in the amount of explosion is less than 0, the smaller the difference in the amount of explosion is, the larger the deviation coefficient of the amount of explosion is, and the larger the amount of explosion actually required is.

[0032] In this embodiment, S5 includes the following sub-steps: S51, subtracting the inlet water quality indicators at adjacent moments to obtain a water quality indicator difference; S52. Calculate the water quality index variation coefficient based on the water quality index difference and construct a water quality index variation sequence.

[0033] In this embodiment, the water quality index difference in S51 is: , where △C t is the water quality index difference at time t, C COD,in,t+1 is the chemical oxygen demand of the influent at time t+1, C BOD,in,t+1 is the biological oxygen demand of the influent at time t+1, C NH,in,t+1 is the influent ammonia nitrogen concentration at time t+1, C COD,in,t is the chemical oxygen demand of the influent at time t, C BOD,in,t is the biological oxygen demand of the influent at the tth moment, C NH,in,t is the influent ammonia nitrogen concentration at the tth moment, and t is the moment number.

[0034] In this embodiment, the formula for calculating the water quality index variation coefficient in S52 is: , where ε is the coefficient of variation of water quality index and | | is the absolute value operation.

[0035] The present invention subtracts the inlet water quality indicators at adjacent moments to obtain the water quality indicator difference, which reflects the change amplitude of the water quality indicator, thereby calculating the water quality indicator change coefficient, and the water quality indicator change coefficient is used to reflect whether the inlet water has a sudden change in a short period of time.

[0036] When water quality indicators suddenly increase, the growth environment of microorganisms will change dramatically, which may lead to the over-growth of certain microbial populations, while other populations may decrease, thus affecting the overall efficiency of sewage treatment. Treatment facilities cannot adapt to sudden changes in water quality in a timely manner, resulting in insufficient treatment capacity, substandard effluent quality, and increased risk of environmental pollution. When water quality indicators suddenly decrease, it means a decrease in pollutants in the influent, which may lead to a decrease in the activity of microorganisms, affecting their ability to treat remaining pollutants, thereby reducing overall treatment efficiency. For sewage treatment systems, both sudden and substantial increases and decreases in water quality indicators will increase the difficulty of sewage treatment.

[0037] In this embodiment, the water quality prediction model can use a BP neural network, and the residence time deviation sequence, the aeration volume deviation sequence and the water quality index change sequence are input into the input end of the BP neural network to obtain the water quality treatment fitness score. Figure 2 As shown, the water quality prediction model in S6 includes: a time deviation processing channel, a gas volume deviation processing channel, a water quality index change processing channel and an output layer; The input end of the time deviation processing channel is used to input the residence time deviation sequence to obtain the time impact value; the input end of the aeration volume deviation processing channel is used to input the aeration volume deviation sequence to obtain the aeration volume impact value; the input end of the water quality index change processing channel is used to input the water quality index change sequence to obtain the water quality index impact value; the output layer is used to weight the time impact value, the aeration volume impact value and the water quality index impact value to obtain the water quality treatment fitness score.

[0038] The time deviation processing channel calculates the impact of time on the treatment effect by receiving the residence time deviation sequence; the aeration volume deviation processing channel evaluates the effect of oxygen supply on microbial activity by analyzing the deviation of the aeration volume; and the water quality index change processing channel focuses on the dynamic impact of water quality changes. Finally, these impact values ​​are weighted and synthesized in the output layer to form a water quality treatment fitness score.

[0039] In the output layer, corresponding weights are assigned to the time impact value, the gas volume impact value, and the water quality index impact value to obtain the water quality treatment fitness score.

[0040] The time deviation processing channel, the gas volume deviation processing channel and the water quality index change processing channel all include: a maximum pooling layer, a significant value enhancement layer and a fully connected layer connected in sequence, wherein the maximum pooling layer is a one-dimensional pooling layer, the pooling window size is 4, and the stride is 4; The expression of the saliency enhancement layer is: , where y i is the i-th enhanced saliency value output by the saliency enhancement layer, x i is the ith significant value output by the maximum pooling layer, and e is a natural constant.

[0041] Since the larger the time deviation coefficient in the residence time deviation sequence of the present invention, the more serious the time deviation, the larger the aeration volume deviation coefficient in the aeration volume deviation sequence, the more serious the aeration volume deviation, and the larger the water quality index change coefficient in the water quality index change sequence, the water quality index has a mutation. Therefore, the present invention uses a maximum pooling layer with a size of 4 and a stride of 4 to extract a maximum value (i.e., a significant value) from every 4 moments in the sequence, and uses the maximum value to reflect the most significant changes, filters out noise and unimportant fluctuations, and then uses a significant value enhancement layer to enhance the significant value, so that the impact value can better reflect the deviation or change.

[0042] In this embodiment, the fully connected layer is used to process the enhanced saliency value output by the saliency enhancement layer, and the activation functions used by the fully connected layer include: Sigmoid function and tanh activation function.

[0043] In this embodiment, the water quality prediction model can be trained using the gradient descent method.

[0044] Embodiment 2, a sewage treatment water quality prediction system, comprising: a collection unit, a sewage residence time acquisition unit, a first sequence construction unit, a second sequence construction unit, a third sequence construction unit and a prediction unit; The collection unit is used to collect sewage inflow flow and inflow water quality indicators at each moment; The sewage residence time acquisition unit is used to obtain multiple sewage residence times in the biochemical pool according to the sewage inflow flow rate at each moment; The first sequence construction unit is used to construct a residence time deviation sequence according to the difference between a plurality of sewage residence times and a reference time; The second sequence construction unit is used to construct a gas volume deviation sequence according to the difference between the gas volume provided at the historical moment and the required gas volume; The third sequence construction unit is used to construct a water quality index change sequence according to the influent water quality index at each moment; The prediction unit is used to predict the water quality treatment fitness score based on the water quality prediction model according to the residence time deviation sequence, the explosion volume deviation sequence and the water quality index change sequence.

[0045] The specific implementation process of Example 2 is the same as that of Example 1.

[0046] In the present embodiments 1 and 2, the higher the water quality treatment fitness score, the worse the existing sewage treatment situation. According to the water quality treatment fitness score, the following specific guidance and suggestions can be provided to the staff: score range and corresponding measures, high fitness (1~2 points): the system is in good operating condition and maintains the current operating parameters; medium and high fitness (2~3 points): check whether the residence time is within the optimal range, fine-tune the aeration volume, and ensure that the dissolved oxygen is maintained at an ideal level; medium fitness (3~4 points): adjust the influent flow rate to make the residence time closer to the reference value, and appropriately increase or decrease the aeration volume according to actual needs; low fitness (4~5 points): immediately check whether there is a sudden change in the influent water quality indicators, adjust the biochemical pool operating parameters, including the return ratio and sludge concentration, increase the aeration volume, improve the treatment capacity, consider launching an emergency treatment plan, increase the monitoring frequency, and test key indicators every 2 hours; extremely low fitness (5 points): launch an emergency plan and temporarily reduce the influent load.

[0047] The present invention effectively reflects the pollution of the influent water by collecting the sewage influent flow and water quality indicators (including chemical oxygen demand, biological oxygen demand and ammonia nitrogen concentration) in real time. According to the sewage influent flow at each moment, the residence time of the sewage in the biochemical pool is calculated, and it is compared with the reference time to construct the residence time deviation sequence. This process can reveal the adaptability of sewage treatment time.

[0048] At the same time, the present invention also takes into account the gap between the actual aeration volume provided during the sewage treatment process and the required aeration volume, and constructs an aeration volume deviation sequence. The establishment of this sequence is helpful to evaluate the impact of aeration conditions on the sewage treatment effect.

[0049] In addition, changes in water quality indicators reflect the difficulty of water quality treatment. Rapid changes in water quality indicators may increase the difficulty of treatment and affect the treatment effect. By combining the residence time deviation, aeration volume deviation and water quality indicator changes, the adaptability of sewage treatment is comprehensively evaluated, and finally the water quality treatment fitness score is obtained. The water quality treatment fitness score is used to evaluate the influent during this period of time, and predict the fitness of the water quality treatment process in advance, so as to warn the staff to make timely adjustments to the parameters of sewage treatment.

[0050] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting water quality in sewage treatment, characterized in that: The following steps are involved: S1. Collect the sewage inflow flow and inflow water quality indicators at each time, wherein the inflow water quality indicators include: inflow chemical oxygen demand, inflow biological oxygen demand and inflow ammonia nitrogen concentration; S2. Obtain multiple sewage residence times in the biochemical pool according to the sewage inflow flow at each moment; S3, constructing a residence time deviation sequence according to the difference between multiple sewage residence times and reference times; S4. Constructing a gas volume deviation sequence according to the difference between the gas volume provided at the historical moment and the required gas volume; S5. Construct a water quality index change sequence based on the influent water quality index at each moment; S6. According to the residence time deviation sequence, the aeration volume deviation sequence and the water quality index change sequence, the water quality treatment fitness score is predicted based on the water quality prediction model.

2. The method for predicting sewage treatment water quality according to claim 1, characterized in that: The reference time in S3 includes: minimum stay time T min , maximum residence time T max and standard residence time T s , S3 includes the following sub-steps: S31, in (T min +T t ) / 2≤T t ≤(T max +T t ) / 2, the corresponding sewage retention time T is assigned t The time deviation coefficient is 0; S32, in T min ≤T t <(T min +T s ) / 2, calculate the corresponding sewage retention time T t Time deviation coefficient: , where θ is the time deviation coefficient; S33, in (T max +T s ) / 2<T t ≤T max Calculate the corresponding sewage retention time T t Time deviation coefficient: ; S34, in T t <T min Calculate the corresponding sewage retention time T t Time deviation coefficient: ; S35, in T t >T max Calculate the corresponding sewage retention time T t Time deviation coefficient: ; S36, each time deviation coefficient is calculated according to the sewage residence time T t The occurrence order is arranged to obtain the residence time deviation sequence.

3. The method for predicting sewage treatment water quality according to claim 1, characterized in that: The S4 comprises the following sub-steps: S41, calculating the required explosion volume for the inlet water quality index at each moment, and obtaining a required explosion volume sequence; S42, obtaining a comparison explosion volume sequence according to the explosion volume provided at each historical moment; S43, subtracting the comparison explosion volume sequence from the required explosion volume sequence to obtain an explosion volume difference sequence; S44. Calculate the explosion volume deviation coefficient according to each explosion volume difference in the explosion volume difference sequence, and construct the explosion volume deviation sequence.

4. The method for predicting sewage treatment water quality according to claim 3, characterized in that: The formula for calculating the explosion volume deviation coefficient in S44 is: , where μ is the explosion volume deviation coefficient, △Q a is the difference in explosion volume, and | | is the absolute value.

5. The method for predicting sewage treatment water quality according to claim 1, characterized in that: The S5 comprises the following sub-steps: S51, subtracting the inlet water quality indicators at adjacent moments to obtain a water quality indicator difference; S52. Calculate the water quality index variation coefficient based on the water quality index difference and construct a water quality index variation sequence.

6. The method for predicting sewage treatment water quality according to claim 5, characterized in that: The water quality index difference in S51 is: , where △C t is the water quality index difference at time t, C COD,in,t+1 is the chemical oxygen demand of the influent at time t+1, C BOD,in,t+1 is the biological oxygen demand of the influent at time t+1, C NH,in,t+1 is the influent ammonia nitrogen concentration at time t+1, C COD,in,t is the chemical oxygen demand of the influent at time t, C BOD,in,t is the biological oxygen demand of the influent at the tth moment, C NH,in,t is the influent ammonia nitrogen concentration at the tth moment, and t is the moment number.

7. The method for predicting sewage treatment water quality according to claim 5, characterized in that: The formula for calculating the coefficient of variation of water quality index in S52 is: , where ε is the coefficient of variation of water quality index and | | is the absolute value operation.

8. The method for predicting sewage treatment water quality according to claim 1, characterized in that: The water quality prediction model in S6 includes: a time deviation processing channel, a gas volume deviation processing channel, a water quality index change processing channel and an output layer; The input end of the time deviation processing channel is used to input the residence time deviation sequence to obtain the time impact value; the input end of the aeration volume deviation processing channel is used to input the aeration volume deviation sequence to obtain the aeration volume impact value; the input end of the water quality index change processing channel is used to input the water quality index change sequence to obtain the water quality index impact value; the output layer is used to weight the time impact value, the aeration volume impact value and the water quality index impact value to obtain the water quality treatment fitness score.

9. The method for predicting sewage treatment water quality according to claim 8, characterized in that: The time deviation processing channel, the gas volume deviation processing channel and the water quality index change processing channel all include: a maximum pooling layer, a significant value enhancement layer and a fully connected layer connected in sequence, wherein the maximum pooling layer is a one-dimensional pooling layer, the pooling window size is 4, and the stride is 4; The expression of the saliency enhancement layer is: , where y i is the i-th enhanced saliency value output by the saliency enhancement layer, x i is the ith significant value output by the maximum pooling layer, and e is a natural constant.

10. A sewage treatment water quality prediction system, implemented based on the sewage treatment water quality prediction method according to any one of claims 1 to 9, characterized in that: include: A collection unit, a sewage residence time acquisition unit, a first sequence construction unit, a second sequence construction unit, a third sequence construction unit and a prediction unit; The collection unit is used to collect sewage inflow flow and inflow water quality indicators at each moment; The sewage residence time acquisition unit is used to obtain multiple sewage residence times in the biochemical pool according to the sewage inflow flow rate at each moment; The first sequence construction unit is used to construct a residence time deviation sequence according to the difference between a plurality of sewage residence times and a reference time; The second sequence construction unit is used to construct a gas volume deviation sequence according to the difference between the gas volume provided at the historical moment and the required gas volume; The third sequence construction unit is used to construct a water quality index change sequence according to the influent water quality index at each moment; The prediction unit is used to predict the water quality treatment fitness score based on the water quality prediction model according to the residence time deviation sequence, the explosion volume deviation sequence and the water quality index change sequence.

Images