Method for treating leachate based on coupling of microalgae with nitrification process

Abstract

The application discloses a leachate treatment method based on a microalgae coupling nitrification process, belongs to the field of wastewater biological treatment and intelligent control, and constructs initial reaction conditions suitable for microalgae proliferation and nitrification reaction; operation parameters such as ammonia nitrogen concentration, nitrite concentration, dissolved oxygen, temperature and microalgae concentration are collected to form real-time operation data; based on a weighted multi-parameter fitting relationship, the optimal combination of light intensity, carbon source addition rate, circulation ratio and reaction length is calculated; when the ammonia nitrogen removal rate is continuously lower than a threshold value, historical data are called to perform trend analysis, and the fitting weight is dynamically corrected; the correction result is applied to control light conditions and water inlet and outlet ratio; after the operation cycle ends, the fitting model is updated in an incremental learning mode, and continuous optimization of the control strategy is realized; the application improves nitrogen removal efficiency and system stability of the leachate and has strong self-adaptive adjustment capacity.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of wastewater biological treatment and intelligent control, and particularly relates to a leachate treatment method based on microalgae coupling nitrification process. BACKGROUND

[0002] Leachate is high-concentration organic wastewater generated in the process of landfill or stacking of municipal solid waste, and has the characteristics of high ammonia nitrogen concentration, large organic load, complex water quality composition, strong toxicity, etc., and is one of the typical difficult-to-treat wastewaters. If it is directly discharged without effective treatment, it is easy to cause environmental problems such as water eutrophication and groundwater pollution, so efficient removal of nitrogen pollutants in leachate has become an important research direction in current solid waste treatment.

[0003] At present, the ammonia nitrogen treatment of leachate mainly adopts traditional nitrification-denitrification biological denitrification process, but this kind of process usually relies on external carbon source, and has high requirements for operation condition control, and under the characteristics of low C / N ratio and high salinity of leachate, the denitrification efficiency is obviously decreased, and the system stability is poor and the operation cost is high.

[0004] Microalgae, as an autotrophic organism, can utilize carbon dioxide for photosynthesis under light conditions, and absorb nutrients such as ammonia nitrogen and phosphorus, and its oxygen production capacity can provide dissolved oxygen for the nitrification process, and has good denitrification potential. However, the growth of microalgae alone for treating high-ammonia-nitrogen leachate has the problems of growth inhibition and incomplete nitrogen removal.

[0005] In view of the above problems, some researches have tried to couple microalgae technology with autotrophic nitrification process, but the current coupling path is not perfect, and there is a lack of precise regulation of the synergistic relationship between microalgae growth and nitrification process, especially under the conditions of high ammonia nitrogen, low carbon and high toxicity of leachate, it is still difficult to stably achieve efficient denitrification. SUMMARY

[0006] The purpose of the present application is to provide a leachate treatment method based on microalgae coupling nitrification process to solve the problems in the background art.

[0007] In order to achieve the above purpose, the present application provides the following technical scheme: a leachate treatment method based on microalgae coupling nitrification process, comprising:

[0008] Introducing leachate into a reaction environment with controllable light conditions to establish initial conditions suitable for microalgae proliferation and ammonia oxidation reaction;

[0009] During the reaction process, the ammonia nitrogen concentration, nitrite concentration, dissolved oxygen concentration, pH value, temperature and microalgae concentration operating parameters are obtained to form real-time operating data;

[0010] Based on the obtained real-time operation data, the optimal combination of light intensity, carbon source addition rate, cycle ratio and reaction time is calculated through the established weighted multi-parameter fitting relationship;

[0011] When the monitoring results show that the ammonia nitrogen removal rate is lower than the preset threshold value for two consecutive periods, historical operation data is called for trend analysis, and the weight parameters in the weighted multi-parameter fitting relationship are dynamically corrected based on a time series prediction method to update the operation control strategy;

[0012] The light conditions and the water in and out ratio are controlled according to the corrected weight parameters;

[0013] After each operation period ends, the treated leachate is output, and all the operation data of the current period are used to update the weighted multi-parameter fitting relationship.

[0014] Preferably, the step of introducing the leachate into the reaction environment with controllable light conditions to establish initial conditions suitable for microalgae proliferation and ammonia oxidation reaction comprises:

[0015] The leachate is pretreated by coarse grid and introduced into the initial reaction unit, and a buffer solution is used to adjust the pH value to 6.8-7.5;

[0016] Based on the ammonia nitrogen concentration and total organic carbon value of the influent, the carbon-nitrogen ratio required is calculated and the amount of added carbon source is determined;

[0017] After the carbon source is added, a pre-illumination period is set, and the light in the reaction environment is gradually increased to the target illumination.

[0018] Preferably, the step of calculating the optimal combination of light intensity, carbon source addition rate, cycle ratio and reaction time through the established weighted multi-parameter fitting relationship comprises:

[0019] The obtained ammonia nitrogen concentration, nitrite concentration, dissolved oxygen concentration, microalgae concentration and temperature are normalized;

[0020] Each normalized parameter is assigned a preset weight to construct a weighted multi-parameter fitting function, the fitting function takes the target control variable as the output, and the optimal combination scheme is selected using the least deviation principle.

[0021] Preferably, the step of constructing a weighted multi-parameter fitting function, the fitting function taking the target control variable as the output, and selecting the optimal combination scheme using the least deviation principle comprises:

[0022] Each weight parameter is substituted into a multi-element nonlinear fitting model, the fitting model is trained using a stepwise regression algorithm, and the predicted values of light intensity, carbon source addition rate, cycle ratio and reaction time in the control variable group are output;

[0023] For each set of predicted values, calculate the deviation between its last cycle running target function value and the current predicted target function value, the target function includes the pollutant removal rate and the microalgae growth rate, select the deviation value of the minimum set of control variables as the optimal output scheme.

[0024] Preferably, when the monitoring result shows that the ammonia nitrogen removal rate is lower than the preset threshold value for two consecutive periods, the historical operation data is called for trend analysis, including:

[0025] Record the actual measured value of the ammonia nitrogen removal rate in the last two operation periods, and determine whether they are lower than the set threshold value, if so, call the historical operation data similar to the environmental conditions of the reaction stage to form a time series data set;

[0026] Trend analysis is performed on the time series data set, and the short-term trend change value of the ammonia nitrogen removal rate is calculated by using the exponential smoothing prediction method, and the deviation between the predicted trend and the current measured value is used as the fitting weight adjustment trigger factor;

[0027] The weight of each operating parameter in the current weighted multi-parameter fitting relationship is dynamically corrected, so as to update the calculation path of the control variables in the next period.

[0028] Preferably, the dynamic correction of the weight of each operating parameter in the current weighted multi-parameter fitting relationship includes:

[0029] Calculate the Pearson correlation coefficient between each operating parameter and the ammonia nitrogen removal rate in the last ten periods, construct the correlation matrix between the parameters and the target performance indicators, and divide them into three categories of high correlation, medium correlation and low correlation according to the correlation matrix;

[0030] Combine the fluctuation amplitude of each operating parameter in the current period with its correlation category, and apply the weight adjustment function to modify the value, the weight of the parameter with large fluctuation amplitude in the high correlation group is preferentially increased; the weight of the parameter in the low correlation group is linearly decreased when the fluctuation amplitude is small in the current period;

[0031] The modified weight set is used to reconstruct the multi-parameter fitting relationship, which is used to update the control output of the light intensity, carbon source dosage rate, circulation ratio and reaction time in the next period.

[0032] Preferably, the control of the light condition and the water in and out according to the corrected weight parameter includes: determining the target light intensity and the water in and out according to the output result of the weighted fitting function after weight correction; adjusting the light source power supply time and the irradiation intermittent period to match the target light intensity, and setting the water in and out rate by adjusting the running time and start-stop frequency of the peristaltic pump.

[0033] Preferably, the using of the whole operation data of the current period to update the weighted multi-parameter fitting relationship comprises: updating the weighted multi-parameter fitting relationship in an incremental learning mode, and taking the data of the new period as an extended sample input.

[0034] Preferably, the normalized operation parameters collected in the current operation period and the corresponding target output values are combined to form training sample pairs, and a sliding sample window is constructed together with the historical samples retained in the recent several periods; the Euclidean distances of the new samples and the historical samples in the parameter distribution space are compared, and the samples with high difference are preferentially added to the weighted multi-parameter fitting relationship.

[0035] In the above technical solutions, the technical effects and advantages provided by the present application are as follows:

[0036] 1. In the leachate treatment method based on the weighted multi-parameter fitting relationship, the multi-source operation parameter normalization processing, the dynamic weight adjustment mechanism and the incremental learning strategy are introduced, so that the accurate calculation and optimal control of the key control variables such as light intensity, carbon source addition rate, circulation ratio and reaction time are effectively realized. The method not only improves the synergistic efficiency of microalgae and nitrification process, but also enhances the adaptability of the system to leachate water quality fluctuations, and significantly improves the ammonia nitrogen removal rate and the stability of microalgae biomass.

[0037] 2. Compared with the existing process using fixed control parameters or static model, the present application has the advantages of significant data-driven intelligence: through time series trend prediction, weight dynamic correction and sliding sample window management, a sustainable self-adaptive optimization control path is constructed, and the stability of the system operation is improved. At the same time, the present application has good algorithm expandability and can be widely applied to the resourceization and ecological treatment process of high ammonia nitrogen difficult wastewater. BRIEF DESCRIPTION OF DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art according to these drawings.

[0039] Figure 1 The method flowchart of the present application. DETAILED DESCRIPTION

[0040] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0041] Embodiments, please refer to Figure 1 The leachate treatment method based on the microalgae coupled nitrification process described in the embodiments includes:

[0042] The leachate is introduced into a reaction environment with controllable illumination conditions to establish initial conditions suitable for microalgae proliferation and ammonia oxidation reaction.

[0043] During the reaction process, the ammonia nitrogen concentration, nitrite concentration, dissolved oxygen concentration, pH value, temperature and microalgae concentration operating parameters are obtained to form real-time operating data.

[0044] Based on the obtained real-time operating data, the optimal combination of illumination intensity, carbon source addition rate, circulation ratio and reaction time is calculated through the established weighted multi-parameter fitting relationship.

[0045] When the monitoring results show that the ammonia nitrogen removal rate is lower than the preset threshold value in two consecutive periods, the historical operating data is called for trend analysis, and the weight parameters in the weighted multi-parameter fitting relationship are dynamically corrected based on the time series prediction method to update the operating control strategy.

[0046] The illumination conditions and the water in-out ratio are controlled according to the corrected weight parameters to realize the dynamic balance between the oxygen production rate of microalgae and the oxygen demand of nitrification reaction.

[0047] At the end of each operating period, the treated leachate is output, and all the operating data of the current period are used to update the weighted multi-parameter fitting relationship.

[0048] In the present application, the "leachate is introduced into a reaction environment with controllable illumination conditions, the pH value is adjusted to a constant interval, and the external carbon source is supplemented according to a preset carbon-nitrogen ratio to establish initial conditions suitable for microalgae proliferation and ammonia oxidation reaction" specifically includes the following technical measures:

[0049] Firstly, the leachate collected from the landfill or other sources is preliminarily screened by a mechanical grid to remove suspended particles, floating objects and coarse impurities, so as to avoid blockage or illumination interference to the subsequent reaction unit. This step not only guarantees the stability of the subsequent reaction environment, but also provides a relatively clean growth medium for microalgae.

[0050] After the grid pretreatment is completed, the leachate is introduced into the reaction unit with controllable light device. Considering that the ammonia nitrogen concentration in the leachate is usually high (often in hundreds to thousands of milligrams per liter), which has a certain inhibitory effect on the growth of microalgae, therefore, the pH value needs to be adjusted before the reaction starts to slow down the ammonia nitrogen volatilization, and at the same time provide a suitable enzyme activity environment for nitrifying bacteria. In the present embodiment, a buffer solution formed by compounding sodium bicarbonate and phosphate (such as potassium dihydrogen phosphate) is used for pH regulation, in which sodium bicarbonate is mainly used to provide alkalinity, and phosphate provides phosphorus source and enhances pH buffering capacity. Preferably, the pH is controlled to be stable between 6.8 and 7.5. The adjustment mode adopts dropwise addition and is assisted by pH online monitoring probe to ensure that the set range is reached within the specified time.

[0051] Secondly, in order to ensure the smooth progress of the nitrification reaction, sufficient carbon source must be provided for the initial growth demand of microalgae in the ammonia oxidation stage, while avoiding the early start of denitrification reaction. Therefore, in the present application, the external carbon source is dynamically calculated by carbon nitrogen ratio (C / N), which is the ratio of total organic carbon content of carbon source to ammonia nitrogen concentration. Specifically, according to the initial concentration of ammonia nitrogen in the influent (in milligrams per liter) and the real-time monitoring value of TOC (total organic carbon), the ratio is analyzed, and when the C / N is lower than the set threshold (preferably 6), the carbon source is automatically supplemented. In the present application, the external carbon source is preferably easily biodegradable carbon sources such as sodium acetate, glycerol or ethanol, which are diluted by an online dilution system by a ratio of 1:10 before being added to avoid local concentration being too high to cause microalgae cell membrane rupture or early start of denitrification chain reaction. The diluted carbon source is added by a precision peristaltic pump to ensure uniform addition and controllable rate.

[0052] In order to further improve the adaptability of microalgae to the reaction environment, especially the stress resistance under high ammonia nitrogen initial conditions, the present application sets a pre-activation stage. Specifically, after the leachate enters the reaction unit and the pH and carbon source are regulated, a pre-illumination time period is set, the light intensity is gradually increased to the target value (preferably in the range of 30 to 80 micromole photons per square meter per second), and a pre-adaptation period of not less than 30 minutes is maintained. In this stage, microalgae have not proliferated in large numbers but have started to start photosynthesis, and the oxygen released by them can be naturally enriched in the reaction unit to provide initial oxygen source support for the subsequent nitrification process. At the same time, the slowly increasing light intensity helps to activate the photosynthetic pigment system and improve the physiological response ability of microalgae in high ammonia nitrogen environment, reducing the inhibition phenomenon.

[0053] In the present application, after the leachate enters the reaction stage, a plurality of groups of online sensors are deployed at different key positions to collect real-time operation parameters. The above-mentioned sensors include:

[0054] Ammonia and nitrite sensor: using selective ion electrode method sensor, through the concentration response of ammonia and nitrite ions in the reaction solution, online continuous detection is realized. The electrode has high selectivity and anti-interference design for the interference of complex background ions in the leachate;

[0055] Dissolved oxygen, pH value and temperature sensor: using fluorescence method dissolved oxygen probe, composite pH glass electrode and thermistor thermometer, all probes are packaged into high corrosion resistant structure to adapt to high salinity and high ammonia nitrogen environment of leachate;

[0056] Microalgae concentration collection device: a light density detection window is arranged, and the colorimetric detection method is used to obtain the light density value (OD value) in the reaction solution, which is used as an indirect indication index of microalgae concentration.

[0057] In order to ensure the time consistency and response sensitivity of the data, all sensors collect data synchronously with ten minutes as the reference collection period. Different from the traditional fixed-time collection method, the application further introduces a sampling frequency dynamic adjustment mechanism, that is, when the system detects that the fluctuation rate of any key parameter (such as ammonia nitrogen concentration or dissolved oxygen) exceeds the set threshold value, the collection frequency is automatically increased to five minutes once; if the parameter tends to be stable, the normal frequency is restored. This strategy takes into account the system resource consumption and response accuracy, and has higher real-time adaptability.

[0058] In terms of microalgae concentration collection, the conventional light density method is easily disturbed by bubbles, resulting in abnormal readings, because the leachate contains a large amount of suspended solids and biological foam. Therefore, the application introduces a sliding average and outlier rejection algorithm in the light density reading program. That is, after continuously obtaining the OD values of three sampling periods, the weighted average is calculated, and the abnormal data points deviating from the mean value by more than two standard deviations are removed. This significantly improves the stability and reliability of the microalgae concentration data in the high foam background.

[0059] All collected data are arranged in a unified structured data format. Each group of data consists of three core elements, namely time stamp (collection time), measured value (such as ammonia nitrogen concentration of 30 mg / L), and measurement confidence (determined by probe calibration accuracy and algorithm bias, divided into high, medium and low levels). The three-element data structure not only facilitates subsequent modeling and control input, but also supports data quality backtracking, avoiding overall control logic deviation caused by error propagation.

[0060] After the data collection is completed, enter the real-time data verification and calibration phase. The present application designs a multi-parameter linkage verification mechanism in this link: when two or more operating parameters are detected to deviate from their respective preset intervals at the same time (for example, the ammonia nitrogen concentration increases while the dissolved oxygen decreases), the linkage verification mechanism is triggered. At this time, the current abnormal data is compared with the data at the same time point in the last valid operation period, and if the deviation amplitude exceeds 15%, the data is marked as “verification required state” and is temporarily not used for control parameter input, and an alarm is issued. This mechanism effectively reduces the control errors caused by single-point sensor failure or temporary interference, and improves the stable operation ability of the system.

[0061] In addition, during the data collection process, to ensure the long-term stability of the sensor under high ammonia nitrogen concentration and complex background, the present application adopts a timing self-cleaning mechanism and an intermittent calibration mechanism to ensure the accuracy of each sensor under long-period operation. The probe cleaning program is automatically triggered every 48 hours, and calibration is performed once every seven days by manual or automatic means using standard solutions or known concentration simulation liquids.

[0062] In the present application, the whole process of “calculating the optimal combination of control variables through the established weighted multi-parameter fitting relationship” includes the following technical details:

[0063] After obtaining the operating parameters such as ammonia nitrogen concentration, nitrite concentration, dissolved oxygen concentration, microalgae concentration and temperature, in order to eliminate the influence between different physical dimensions and numerical ranges, it is necessary to first normalize each parameter. Specifically, the present application preferably uses linear interval scaling method for normalization, that is, the current value of each parameter is subtracted from the historical minimum value of the parameter, and then divided by the historical range (maximum value minus minimum value) of the parameter, thereby normalizing it to a dimensionless standard value between 0 and 1.

[0064] For example, the ammonia nitrogen concentration at a certain time point is 300 mg / L, the historical minimum value is 100, and the maximum value is 500, then the normalization result is: .

[0065] All operating parameters involved in the calculation are standardized according to the above method, thereby ensuring that in the subsequent weighting and fitting process, each parameter has an equivalent mathematical expression basis, avoiding the disproportionate influence of a parameter due to its large value on the output result.

[0066] After completing the normalization process, the system assigns a weight value to each parameter. Instead of manually setting the weight based on experience, the present application dynamically calculates the weight factor based on the historical correlation between the parameters and the key performance indicators. In the specific implementation process, the system calls the statistical correlation coefficients of each operating parameter and the pollutant removal rate, microalgae concentration growth rate and reaction liquid pH stability in the past continuous periods, and uses them as the initial weight value.

[0067] After the weight assignment, the parameter combination of the fitting function input end is realized by constructing a weighted linear combination function or a nonlinear combination function.

[0068] If the normalized parameters are A1 (ammonia nitrogen), A2 (nitrite), A3 (DO), A4 (microalgae concentration), and A5 (temperature), and the corresponding weights are W1 to W5, the input expression of the fitting function is W1 x A1 + W2 x A2 + W3 x A3 + W4 x A4 + W5 x A5.

[0069] To further improve the adaptability and generalization ability of the function, the present application adopts stepwise regression algorithm to construct a multivariate nonlinear fitting model. Stepwise regression is a variable selection process based on significance test, which can effectively eliminate redundant variables and improve model interpretability. The regression process includes variable introduction, variable elimination and interaction term setting, and the fitting target is to predict the optimal value range of the control variable.

[0070] For example, the system can construct four nonlinear regression sub-models corresponding to light intensity, carbon source addition rate, circulation ratio and reaction time, respectively. Each model takes the weighted combination of the operating parameters as input and the historical optimal value of the control variable as output, and finally obtains four predicted values.

[0071] The light intensity, carbon source addition rate, circulation ratio and reaction time calculated by the fitting model are the candidate value set. Since the model has certain error, to ensure the reliability of the actual running effect, the present application further designs a minimum deviation optimization mechanism.

[0072] Specifically, the system retrieves the target function output value of the last running period, which includes but is not limited to:

[0073] Ammonia nitrogen removal rate;

[0074] Microalgae biomass growth rate;

[0075] Reaction liquid pH change amplitude.

[0076] Then, the fitting predicted values are brought into the current running control logic to predict the possible target function output values, and the predicted values are compared with the actual values of the last period to calculate the deviation. The deviation preferably adopts the "relative difference absolute value" method, that is, the predicted target value is subtracted from the actual target value, then divided by the target value of the last period, and the absolute value is taken.

[0077] After simulating and evaluating the deviation of each of the four combinations of control variables, the combination with the smallest deviation is selected as the final output scheme for the operation control in the next cycle. This approach not only ensures that the actual effect of the predicted value meets the system's expectations, but also enhances the control strategy's resistance to abnormal values and sudden changes in operating conditions.

[0078] The finally selected control parameters act on the reaction light source device, the carbon source addition controller, the liquid reflux regulating valve, and the periodic liquid discharge timer, respectively, to ensure that the reaction process operates under new optimal control conditions.

[0079] During actual operation, the system records the ammonia nitrogen removal rate of each cycle and compares it with the set performance threshold. If the ammonia nitrogen removal rate is detected to be lower than the preset threshold (e.g., lower than 85%) in two consecutive cycles (e.g., adjacent 2 hours or two control cycles), the system will determine that the operation performance is abnormal and enter the parameter weight adjustment trigger condition.

[0080] To ensure the timeliness and pertinence of weight correction, the system needs to call historical data with similar operating conditions (such as pH, temperature, and microalgae concentration range) to the current reaction stage, preferably constructing a time series data set covering the last ten effective cycles. Each cycle contains the real-time acquisition values of the target output variable (ammonia nitrogen removal rate) and the operating parameters involved in the fitting (ammonia nitrogen concentration, nitrite concentration, dissolved oxygen concentration, microalgae concentration, temperature, etc.).

[0081] For the time series data set, the exponential smoothing prediction method is used to calculate the short-term trend change of the ammonia nitrogen removal rate. Compared with simple average or linear regression, the exponential smoothing method can more accurately capture the weak trend in the time series and is suitable for short-cycle operating condition evaluation.

[0082] The calculation method of the exponential smoothing prediction value is as follows:

[0083] The predicted value at the current time = smoothing coefficient x current actual value + (1-smoothing coefficient) x predicted value at the last time.

[0084] Wherein, the smoothing coefficient is a weight factor between 0 and 1 (preferably 0.3-0.5 in this invention), which can be dynamically adjusted according to the system fluctuation degree.

[0085] The trend value of ammonia nitrogen removal rate obtained by this method is compared with the current measured value, and the difference between the two is the deviation factor, which is used to judge whether the actual performance of the system deviates seriously from the trend expectation.

[0086] When the absolute value of the deviation factor is greater than a certain set tolerance threshold (such as ±5%), the system triggers the weight correction program and enters the next step.

[0087] To improve the adaptability of the weighted multi-parameter fitting relationship, the application establishes a correlation matrix between the operating parameters and the ammonia nitrogen removal rate based on historical data, which is used to support the weight correction logic.

[0088] The specific operation is:

[0089] For the above ten cycles of historical data, the Pearson correlation coefficient between each operating parameter and the ammonia nitrogen removal rate is calculated , which is used to measure the degree of linear correlation between two variables, with a value range of -1 to +1. The closer the coefficient is to +1, the stronger the positive impact of the parameter on the ammonia nitrogen removal rate.

[0090] According to the Pearson coefficient, all operating parameters involved in the fitting are divided into three categories:

[0091] High correlation group (|r| ≥ 0.7);

[0092] Medium correlation group (0.4 ≤ |r|<0.7);

[0093] Low correlation group (|r|<0.4).

[0094] The fluctuation amplitude of each operating parameter in the current cycle is analyzed. The fluctuation amplitude can be represented by the ratio of the absolute value of the difference between the current value and the value of the last cycle to the mean value, which reflects the degree of dynamic change of the parameter in the current cycle.

[0095] Then, according to the correlation category of the parameter and its current fluctuation amplitude, the weight adjustment function is called for value correction. The function follows the following logic:

[0096] If the parameter belongs to the high correlation group and the fluctuation amplitude is large (more than its historical standard deviation) in the current cycle, the original weight value of the parameter is positively increased (for example, by 10%-20%);

[0097] If the parameter belongs to the low correlation group and the fluctuation amplitude is small (within ±10% of the historical mean), the weight value of the parameter is linearly decreased (for example, by 5%-10%);

[0098] The parameters in the medium correlation group are adjusted moderately according to the specific fluctuation amplitude, with smaller or no increase or decrease.

[0099] All the corrected weight values are normalized (for example, the sum is normalized to 1), and used in the weighted fitting function in the calculation of the control variables (light intensity, carbon source dosage rate, circulation ratio, reaction time) in the next cycle.

[0100] After adopting the new set of weights, the system will reconstruct the weighted fitting function to update the output values ​​of the control variables. This process can use the aforementioned nonlinear regression model structure without changing the original model architecture, only replacing the parameter weights.

[0101] Meanwhile, the revised weights and their effective period will be marked in the data records and participate in subsequent model evaluations, serving as part of the "sample labels" in the machine learning path for model iterative optimization.

[0102] Through the aforementioned dynamic weight correction mechanism, this invention can promptly adjust the fitting strategy using a data-driven approach when abnormal fluctuations occur in system operation or the removal effect of target pollutants declines, demonstrating significant adaptive control capabilities and a fault response mechanism.

[0103] After correcting the weights of the fitted model (e.g., through grouping by Pearson correlation coefficient and function correction driven by fluctuation amplitude), the system reapplies the new set of weights to the weighted fitting relationship. Through fitting calculations, an updated set of control variables is obtained, including light intensity (in micromolar photons per square meter per second) and the inflow-outflow ratio (in dimensionless ratio).

[0104] The light intensity output value is limited to an effective physiological illuminance range of 30 to 80 micromolar photons per square meter per second, which is the optimal light range determined by the present invention based on microalgae experiments. The influent-to-effluent ratio is set as an influent flow rate to an effluent flow rate, preferably between 1:1.2 and 1:1.8, to ensure that the reaction solution has a reasonable residence time in the reaction tank and to avoid microalgae loss.

[0105] Lighting control includes two dimensions: light intensity and lighting mode.

[0106] Light intensity control: An adjustable voltage LED light source is used. The light source controller receives the target light value command to continuously adjust the light intensity. The controller is connected to a closed-loop feedback system, and the illuminance value is monitored in real time by a light sensor to ensure that the actual illuminance and the model output value are kept within ±5% of each other.

[0107] Lighting mode control: Generate a lighting curve based on the model output value, and set an intermittent lighting cycle (e.g., run for 20 minutes, rest for 5 minutes) in combination with the daily cycle and reaction requirements to simulate the natural day-night rhythm and reduce the risk of microalgae photoinhibition.

[0108] Through the above methods, this invention not only achieves precise control of light intensity, but also introduces the dimension of "light rhythm," enhancing the adaptability of microalgae to artificial lighting environments and their photosynthetic efficiency.

[0109] The adjustment of the water in and out ratio relies on two groups of independently operated peristaltic pumps or electromagnetic valve controlled flow devices:

[0110] The water in pump receives the target water in rate value output by the model before starting, and uses the flow sensor as the feedback reference during operation to realize closed loop flow control.

[0111] The water out pump or liquid discharge device automatically calculates the target water out rate according to the water in amount and the set ratio, and realizes discharge control in the form of timing on-off or variable frequency operation.

[0112] For example, if the water in and out ratio output by the current cycle model is 1:1.5, the system sets that for every 100 liters of water in, 150 liters of water out should be discharged. To avoid excessive fluctuation of the reaction liquid level caused by liquid discharge, the liquid discharge operation can be set to be pulse type, i.e. intermittent discharge for 5 times, and the interval is set to be between 2 to 3 minutes.

[0113] The water in and out operation is monitored by the liquid level probe, and if the liquid level drops more than the safety threshold, the water out will be temporarily paused, and a reminder will be sent.

[0114] The application specially designs a running feedback and parameter memory mechanism to judge whether the current control strategy achieves the optimization goal, and stores the effective strategy for subsequent calling.

[0115] The evaluation of the control result is based on the following two indexes:

[0116] Microalgae concentration growth rate: comparison of the microalgae concentration change (such as optical density OD value or cell number density) between the current cycle and the previous cycle;

[0117] Ammonia nitrogen concentration reduction rate: the difference between the water in and water out ammonia nitrogen concentration divided by the treatment time to form the denitrification efficiency index per unit time.

[0118] When the above two indexes are improved compared with the previous cycle, and the system runs stably (such as pH fluctuation less than ±0.3, and dissolved oxygen maintained within the set range), the system will mark the current control variable combination (including light intensity and water in and out ratio) as "optimization effective", and write it into the control strategy library.

[0119] After completing one running cycle, the system automatically preprocesses the collected data in the whole process of the cycle. Mainly including the following contents:

[0120] Running parameter normalization processing: including ammonia nitrogen concentration, nitrite concentration, dissolved oxygen concentration, microalgae concentration, temperature, pH value, etc., which are all interval scaled according to the historical maximum and minimum values, so that each parameter is mapped to the [0, 1] interval;

[0121] Target output value extraction: including the final ammonia nitrogen removal rate, microalgae biomass growth rate or other control target values of the cycle.

[0122] Subsequently, the system constructs a training sample pair in the form of "input parameter vector + corresponding target output value", that is:

[0123] Input: current cycle normalized parameter vector (denoted as ) Output: corresponding actual target value (denoted as )

[0124] Form a sample pair ( , ) as an extended sample available for training in the current cycle.

[0125] To prevent the model from rising in computational complexity due to the continuous accumulation of training samples, the present application uses a sliding sample window mechanism to control the sample capacity. The sample window is set to a maximum capacity (e.g. the latest 30 cycles), and when a new sample is input, the oldest set of samples is automatically removed, so that the training data always remains within the latest and limited time range, improving the timeliness and convergence efficiency of the model.

[0126] To avoid interference from redundant or repetitive data in the model, the present application introduces a Euclidean distance calculation method in the parameter distribution space to evaluate the difference between the current new sample and the historical sample, and accordingly selects the new data with stronger representativeness for model updating.

[0127] The specific implementation steps are as follows:

[0128] Calculate the Euclidean distance between the new sample ( ) and each historical sample ( ) in the sample window, which is defined as: ; where n is the number of operating parameters,

[0129] Calculate the minimum distance between the new sample and the historical sample set, and compare it with the set threshold value, for example, if (indicating that the new sample has a significant difference from the historical sample in the parameter space), it is determined that the sample is a "high difference" sample.

[0130] Only the new sample determined to be high difference is included in the training data, avoiding the repeated use of samples that are too close to the model updating and reducing the risk of overfitting of the model.

[0131] The accepted new sample will be used to incrementally update the existing weighted multi-parameter fitting relationship. Unlike retraining, incremental updating does not reconstruct the entire model structure, but rather, through local adjustment, it achieves parameter correction and model performance enhancement based on the preservation of the existing fitting relationship.

[0132] The essence of the fitting relationship is a mapping function from input vector to output variable, which is generated by training historical data and expressed in the form of multiple weighted regression: ; wherein, to is the normalized input parameter, to is the fitting weight, and ε is the residual term.

[0133] Incremental update adopts the following mechanism:

[0134] Take the new sample (x, y) as input, calculate its prediction error under the existing fitting function, that is, the error = |predicted value-actual value|; If is higher than the set threshold (for example, 5%), adjust the weight corresponding to the key parameter in

[0135] by a certain proportion, and fine-tune by the least square criterion or gradient update method. If is lower than the set threshold, only update the residual term, keep the main weight structure unchanged, and reduce unnecessary model disturbance.

[0136] If is lower than the set threshold, only update the residual term, keep the main weight structure unchanged, and reduce unnecessary model disturbance.

[0137] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered within the protection scope of the present application.​

Claims

1. A leachate treatment method based on microalgae coupled nitrification process, characterized in that: include: The leachate was introduced into a reaction environment with controllable light conditions to establish suitable initial conditions for microalgae proliferation and ammonia oxidation reaction. During the reaction, operating parameters such as ammonia nitrogen concentration, nitrite concentration, dissolved oxygen concentration, pH value, temperature, and microalgae concentration are obtained to form real-time operating data; Based on the acquired real-time operational data, the optimal combination of light intensity, carbon source acceleration rate, cycle ratio, and reaction time is calculated through the established weighted multi-parameter fitting relationship, including: The obtained ammonia nitrogen concentration, nitrite concentration, dissolved oxygen concentration, microalgae concentration and temperature were normalized; each normalized parameter was assigned a value according to a preset weight, and a weighted multi-parameter fitting function was constructed. The fitting function used the target control variable as the output, and the optimal combination scheme was selected by using the principle of minimum deviation. The construction of the weighted multi-parameter fitting function, with the target control variable as the output, and the selection of the optimal combination scheme using the minimum deviation principle, includes: substituting each weight parameter into the multivariate nonlinear fitting model, training the fitting model using a stepwise regression algorithm, and outputting the predicted values ​​of light intensity, carbon source acceleration rate, cycle ratio, and reaction time in the control variable group; For each set of predicted values, calculate the deviation between the objective function value in the previous cycle and the current predicted objective function value. The objective function includes the pollutant removal rate and the microalgae growth rate. Select the set of control variables with the smallest deviation value as the optimal output scheme. When monitoring results show that the ammonia nitrogen removal rate is lower than the preset threshold for two consecutive cycles, historical operating data is retrieved for trend analysis, and the weight parameters in the weighted multi-parameter fitting relationship are dynamically corrected based on time series prediction methods to update the operating control strategy, including: The Pearson correlation coefficient between each operating parameter and the ammonia nitrogen removal rate was calculated over ten historical cycles. A correlation matrix between the parameters and the target performance indicators was constructed, and the parameters were divided into three groups: high correlation, medium correlation, and low correlation. The fluctuation range of each operating parameter in the current period is combined with its correlation category, and a weight adjustment function is applied to correct the value. In the high correlation group, the parameters with large fluctuation ranges are given priority to increase their weights; in the case of small fluctuation ranges in the current period, the weights of the parameters in the low correlation group are linearly decreased. The modified weight set is used to reconstruct the multi-parameter fitting relationship, which is used to update the next cycle control output of light intensity, carbon source acceleration rate, cycle ratio and reaction time. The lighting conditions and the ratio of water inlet to outlet are controlled according to the revised weighting parameters; At the end of each running cycle, the processed leachate is output, and all running data of this cycle is used to update the weighted multi-parameter fitting relationship. Specifically, this includes: forming training sample pairs by combining the normalized running parameters collected in the current running cycle and their corresponding target output values, and constructing a sliding sample window together with the historical samples retained in the most recent several cycles; comparing the Euclidean distance between the new samples and the historical samples in the parameter distribution space, and selecting samples with high differences to be added to the weighted multi-parameter fitting relationship first.

2. The leachate treatment method based on microalgae coupled nitrification process according to claim 1, characterized in that: The process of introducing the leachate into a light-controlled reaction environment to establish suitable initial conditions for microalgae proliferation and ammonia oxidation includes: The leachate was pretreated through a coarse screen and then introduced into the initial reaction unit, where the pH was adjusted to 6.8–7.5 using a buffer solution. Based on the influent ammonia nitrogen concentration and total organic carbon value, the required carbon-nitrogen ratio is calculated and the amount of external carbon source added is determined. After the carbon source is added, a pre-illumination period is set so that the light intensity in the reaction environment gradually increases to the target illuminance.

3. The leachate treatment method based on microalgae coupled nitrification process according to claim 1, characterized in that: When the monitoring results show that the ammonia nitrogen removal rate is lower than the preset threshold for two consecutive cycles, the process of calling historical operating data for trend analysis includes: Record the actual measured values ​​of ammonia nitrogen removal rate in two consecutive operating cycles, and determine whether both are below the set threshold. If so, call up historical operating data with environmental conditions similar to the reaction stage to form a time series dataset. Trend analysis was performed on the time series dataset. The exponential smoothing forecasting method was used to calculate the short-term trend change value of ammonia nitrogen removal rate. The deviation between the predicted trend and the current measured value was used as the trigger factor for adjusting the fitting weight. The weights of each operating parameter in the current weighted multi-parameter fitting relationship are dynamically adjusted to update the calculation path of the control variable for the next cycle.

4. The leachate treatment method based on microalgae coupled nitrification process according to claim 1, characterized in that: The control of illumination conditions and influent / outfluent ratio based on the corrected weight parameters includes: determining the target illumination intensity and leachate influent / outfluent ratio based on the output of the weighted fitting function after weight correction; adjusting the power supply time of the light source and the irradiation interval to match the target illumination intensity; and setting the influent and effluent rates by adjusting the running time and start / stop frequency of the peristaltic pump.

5. The leachate treatment method based on microalgae coupled nitrification process according to claim 1, characterized in that: The step of using all the running data of the current cycle to update the weighted multi-parameter fitting relationship includes: updating the weighted multi-parameter fitting relationship by using incremental learning, and using the data of the new cycle as the input of extended samples.

Images