Multi-source sensing based landfill leachate treatment process anomaly diagnosis method and system

Abstract

The present application belongs to the technical field of waste leachate treatment, and provides a waste leachate treatment process abnormality diagnosis method and system based on multi-source sensing, comprising: collecting sensing data of biochemical treatment and membrane separation units, calculating biochemical group characteristic values and membrane group characteristic values and constructing an operation state plane; performing grey correlation degree analysis on candidate disturbance factors, and dynamically dividing them into a main disturbance factor set and a secondary fluctuation factor set according to the correlation degree; fitting an operation constraint line with the main disturbance factor set as the target; fitting an early warning line with the secondary fluctuation factor set as the target, and combining the two curves to divide a normal working condition area and a warning attention area; and outputting a normal operation, risk warning or abnormality diagnosis conclusion according to the working condition landing point position and the tangential deflection direction. The present application dynamically quantifies the influence of external factors, constructs a hierarchical diagnosis boundary, and realizes progressive abnormality diagnosis from routine monitoring to early warning and then to precise disposal.

Description

Technical Field

[0001] This invention belongs to the field of landfill leachate treatment technology, specifically a method and system for abnormal diagnosis of landfill leachate treatment process based on multi-source sensing. Background Technology

[0002] Landfill leachate, with its complex composition and volatile water quality and quantity, is a significant source of water pollution. In the field of environmental protection technology, its treatment typically employs a multi-stage process combining biological treatment and membrane separation. With the development of Internet of Things (IoT) technology, multi-source sensors have been widely applied to leachate treatment systems, providing a rich data foundation for monitoring operational status. However, existing anomaly diagnosis methods still have significant shortcomings: First, the influence of external factors such as influent COD load, water temperature, and sludge concentration on system status changes dynamically with seasonal and water quality variations. Existing methods rely on fixed experience to artificially classify these factors as primary and secondary, lacking a dynamic quantitative screening mechanism, resulting in insufficient adaptability of diagnostic boundaries under different operating conditions. Second, due to the lack of dynamic impact assessment, existing methods mix all factors into the same diagnosis, weakening the decisive constraint of strong influencing factors and obscuring the cumulative early warning value of weak influencing factors, causing confusion between primary and secondary factors. In addition, existing technologies only set a single operational constraint boundary, adopting a binary judgment logic of either / or, which cannot distinguish between urgent anomalies and trend risks. It lacks an early warning line to provide risk indication in the early stages of status deterioration, and it is difficult to accurately identify specific types such as membrane fouling, biochemical shock, or system instability when anomalies occur, making it impossible for operation and maintenance personnel to take timely and differentiated measures.

[0003] Therefore, the present invention provides a method and system for diagnosing anomalies in landfill leachate treatment processes based on multi-source sensing. Summary of the Invention

[0004] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0005] The technical solution adopted by this invention to solve its technical problem is: a method for abnormal diagnosis of landfill leachate treatment process based on multi-source sensing, comprising the following steps:

[0006] Sensor data from the biochemical treatment and membrane separation units during leachate treatment were collected, and characteristic values ​​of the biochemical group and membrane group were calculated respectively. An operating status plane was then constructed.

[0007] Grey relational analysis was performed on the candidate disturbance factors and two sets of eigenvalues. The candidate disturbance factors were divided into a set of primary disturbance factors and a set of secondary fluctuation factors according to the degree of correlation.

[0008] Using the set of major disturbance factors as the target, the distribution range of two sets of feature values ​​corresponding to different combinations of major disturbance factors in historical normal operation is called to fit the operating constraint line under the action of major disturbance factors;

[0009] Using the set of minor fluctuation factors as the target, the edge operation data of the effluent water quality meeting the standards but close to the discharge limit during normal operation in history are called to fit the early warning line under the influence of minor fluctuation factors. Combined with the operation constraint line, the early warning attention area under the influence of fluctuation and the normal operating condition area under stable operation are divided in the state plane.

[0010] The diagnostic conclusion is output based on the landing point position of the current operating condition in the state plane. When the landing point is in the normal operating condition zone, the normal operation is output. When it is in the warning concern zone, the membrane fouling risk warning or biochemical shock risk warning is output according to the tangential deflection direction. When it exceeds the operating constraint line, the membrane fouling acceleration anomaly, biochemical shock anomaly or system instability anomaly is output according to the tangential deflection direction.

[0011] A multi-source sensing-based anomaly diagnosis system for landfill leachate treatment processes includes the following modules:

[0012] State plane construction module: Collect sensor data from the biochemical treatment and membrane separation units during the leachate treatment process, calculate the characteristic values ​​of the biochemical group and the membrane group respectively, and construct the operating state plane;

[0013] Primary and secondary disturbance classification module: Perform grey relational analysis on candidate disturbance factors and two sets of feature values, and classify the candidate disturbance factors into a primary disturbance factor set and a secondary fluctuation factor set according to the degree of correlation.

[0014] Run constraint fitting module: Taking the main disturbance factor set as the target, it calls the distribution range of two sets of feature values ​​corresponding to different combinations of main disturbance factors in the historical normal operation, and fits the running constraint line under the action of the main disturbance factors.

[0015] Early warning boundary partitioning module: Taking the set of minor fluctuation factors as the target, it calls the edge operation data of the effluent water quality that meets the standards but is close to the discharge limit during normal operation in history, fits to obtain the early warning line under the influence of minor fluctuation factors, and combines the operation constraint line to divide the early warning attention area under the influence of fluctuation and the normal operating condition area under stable operation in the state plane.

[0016] The graded anomaly diagnosis module outputs diagnostic conclusions based on the landing point position of the current operating condition in the state plane. When the landing point is in the normal operating condition zone, it outputs normal operation; when it is in the warning concern zone, it outputs membrane fouling risk warning or biochemical shock risk warning according to the tangential deflection direction; when it exceeds the operating constraint line, it outputs membrane fouling acceleration anomaly, biochemical shock anomaly or system instability anomaly according to the tangential deflection direction.

[0017] The beneficial effects of this invention are as follows: First, this invention uses grey relational analysis to dynamically quantify and rank the correlation strength between candidate disturbance factors and system state, overcoming the problem of the diagnostic boundary being disconnected from changes in operating conditions caused by the traditional method's reliance on fixed experience to classify primary and secondary factors. This allows the primary and secondary classification to adaptively adjust with changes in conditions such as season and water quality, significantly improving the adaptability and robustness of the diagnostic method under different operating scenarios. Second, this invention uses strong influencing factors to construct operating constraint lines and weak influencing factors to construct early warning lines, solving the dual defects of existing single diagnostic models that mix primary and secondary factors, resulting in the weakening of the constraint force of strong factors and the masking of the cumulative early warning value of weak factors. Third, this invention sets early warning lines and constraint lines simultaneously in the state plane, forming a three-level regional division of normal operating condition zone, early warning attention zone, and abnormal zone. Based on the direction of tangential deflection of the landing point, it accurately identifies specific abnormal types such as membrane fouling, biochemical shock, and system instability, overcoming the shortcomings of traditional binary judgment logic in lacking early warning and type identification capabilities. This achieves full-process hierarchical diagnosis from state monitoring to risk warning to abnormal handling, providing clear judgment of intervention timing and differentiated handling basis. Attached Figure Description

[0018] The invention will now be further described with reference to the accompanying drawings.

[0019] Figure 1 This is a flowchart of the steps of the method for abnormal diagnosis of landfill leachate treatment process based on multi-source sensing in this invention;

[0020] Figure 2 This is a logic diagram of the abnormal diagnosis method for landfill leachate treatment process based on multi-source sensing of the present invention.

[0021] Figure 3 This is a flowchart of the abnormal diagnosis system for landfill leachate treatment process based on multi-source sensing according to the present invention. Detailed Implementation

[0022] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0023] Example 1: Please refer to Figure 1-2 As shown in the embodiment of the present invention, the method for abnormal diagnosis of landfill leachate treatment process based on multi-source sensing includes the following steps:

[0024] Step 1: Collect sensor data from the biochemical treatment and membrane separation units during the leachate treatment process, calculate the characteristic values ​​of the biochemical group and the membrane group respectively, and construct the operating status plane;

[0025] In step one, the process of collecting sensor data from the biochemical treatment and membrane separation units during leachate treatment is as follows:

[0026] Dissolved oxygen, pH, temperature, and redox potential sensors are installed in the biochemical treatment unit of the landfill leachate treatment system, while pressure, flow, and conductivity sensors are installed in the membrane separation unit. All sensors continuously collect data at a uniform sampling frequency, and the collected data are time-series aligned with the same time marker so that each moment corresponds to a complete set of sensor readings.

[0027] In step one, the process of calculating the biochemical eigenvalues ​​and membrane eigenvalues ​​respectively is as follows:

[0028] For the biochemical treatment unit, the readings of the four sensors (dissolved oxygen, pH, temperature, and redox potential) at all times within a time window preceding the current time are arranged into a data matrix. Each column of the matrix corresponds to all readings of a sensor within a certain time period. Principal component analysis is performed on the matrix to calculate the covariance matrix of each sensor's data. The eigenvalues ​​and eigenvectors of the covariance matrix are then solved to obtain multiple principal components. The first principal component is the direction with the largest variance in the linear combination of the sensor data. The score of the first principal component at the end of the time window is taken as the eigenvalue of the biochemical group.

[0029] For the membrane separation unit, the same principal component analysis operation is used. The readings of the three sensors, pressure, flow rate and conductivity, are arranged into a data matrix for all times in the same time window before the current time. After principal component analysis, the score of the first principal component at the end of the time window is taken as the characteristic value of the membrane group.

[0030] This embodiment provides an example illustrating the calculation of biochemical and membrane characteristic values:

[0031] Based on the current time, a fixed-length time range is extracted backward. If the time window length is set to 30 minutes and the sensor sampling frequency is once per minute, the window contains sampling data for 30 times, including the current time and the previous 29 times.

[0032] Taking the biochemistry group as an example: the four sensors of dissolved oxygen, pH, temperature and redox potential each have 30 readings at 30 time points within the window, forming a data matrix of 30 rows (time points) and 4 columns (sensors);

[0033] The calculation process of principal component analysis on the matrix is ​​as follows: Calculate the covariance between each pair of the four sensors to obtain a 4x4 covariance matrix. Solve for the eigenvalues ​​and eigenvectors of the covariance matrix. There are four eigenvalues, with the largest eigenvalue being called the first eigenvalue. The corresponding eigenvector is a vector containing four coefficients, which correspond to the weights of dissolved oxygen, pH, temperature, and redox potential, respectively. This set of weights maximizes the variance of the value obtained after combining the data from the four sensors according to their weights, thus best reflecting the overall trend of the biochemical unit within the current window. Multiply this weight vector by the actual readings of the four sensors at the current moment within the window to obtain the biochemical group eigenvalue. Similarly, the pressure, flow rate, and conductivity sensors of the membrane separation unit are calculated using the same process to obtain the membrane group eigenvalue.

[0034] In step one, a two-dimensional operating state plane is established with the biochemical group characteristic values ​​as the horizontal axis and the membrane group characteristic values ​​as the vertical axis. The two sets of characteristic values ​​calculated at each moment are marked as a coordinate point on the plane.

[0035] Understandably, the significance of step one lies in the following: This step constructs the characteristic values ​​of the biochemical unit and the membrane unit through unified sampling and time-series alignment of multiple source sensors, and maps the two to a two-dimensional operating state plane. In the existing technology, the data of each sensor are analyzed in isolation, which makes it difficult to comprehensively reflect the coordinated state changes of the biochemical unit and the membrane unit. This step reduces the high-dimensional sensor data to two quantifiable characteristic indicators, laying a unified data foundation for subsequent analysis of the influence of external factors and the division of state regions. It solves the problem that multi-source heterogeneous data is difficult to conduct correlation analysis under the same framework, and realizes the quantitative and visual representation of the overall operating state of the system.

[0036] Step 2: Perform grey relational analysis on the candidate disturbance factors and the two sets of feature values, and divide the candidate disturbance factors into the main disturbance factor set and the secondary fluctuation factor set according to the degree of correlation.

[0037] In step two, the candidate disturbance factors include four main categories: influent water quality (including influent COD load, influent ammonia nitrogen concentration, and influent pH, reflecting the total amount of organic pollutants, total amount of nitrogenous pollutants, and pH entering the system, respectively); operating conditions (including biological tank water temperature, mixed liquor sludge concentration, and deviation between the setpoint and actual dissolved oxygen values, reflecting the environmental temperature conditions of the biochemical reaction, the biomass level of activated sludge, and the execution deviation of aeration control, respectively); hydraulic fluctuations (including influent flow rate fluctuation rate and return flow rate change rate, reflecting the change in the amount of water entering the system per unit time and the change in the internal and external return flow ratio, respectively); and treatment effect (including the degree of closeness between effluent COD and discharge limits, and the degree of closeness between effluent ammonia nitrogen and discharge limits, reflecting the current removal margin of the system for major pollutants). For example, see Table 1 below.

[0038] Table 1: Classification and Explanation of Candidate Perturbation Factors;

[0039] category Disturbance factor symbol unit Physical meaning Influent water quality Influent COD load value COD_in kg / d Total amount of organic pollutants entering the system Influent water quality influent ammonia nitrogen concentration <![CDATA[NH3-N_in]]> mg / L Total amount of nitrogen pollutants entering the system Influent water quality Inlet water pH pH_in — pH of the system Operating conditions Biological pool water temperature T_bio ℃ Environmental temperature conditions for biochemical reactions Operating conditions Mixed liquor sludge concentration value MLSS g / L Biomass level of activated sludge Operating conditions Deviation between dissolved oxygen setpoint and actual value ΔDO mg / L Execution deviation of aeration control Hydraulic fluctuations Inflow Flow Fluctuation Q_fluct % Variation in influent volume per unit time Hydraulic fluctuations Rate of change of return flow R_change % Changes in internal and external reflux ratio Processing effects How close is the effluent COD to the discharge limit? COD_approach % System removal margin for organic pollutants Processing effects How close are the effluent ammonia nitrogen levels to the discharge limits? <![CDATA[NH3_approach]]> % The system's removal margin for nitrogen pollutants

[0040] In step two, the process of performing grey relational analysis on the candidate perturbation factors and the two sets of feature values ​​is as follows:

[0041] All biochemical eigenvalues ​​and membrane eigenvalues ​​recorded recently (e.g., within 7 days) were used as two reference sequences, while the monitoring data of all candidate perturbation factors within the same time period were used as comparison sequences.

[0042] Each set of comparison sequences corresponds one-to-one with two reference sequences in time, that is, the factor data and the eigenvalue data at the same time correspond to each other.

[0043] The parameters in the reference sequence and comparison sequence are standardized by subtracting the minimum value of the sequence from each value and then dividing by the difference between the maximum and minimum values ​​of the sequence.

[0044] For each candidate perturbation factor, the absolute value of the deviation between the candidate perturbation factor and the biochemical group feature value is calculated at each time step. After averaging all the absolute values ​​of deviation, the gray correlation degree between the candidate perturbation factor and the biochemical group feature value is obtained. Similarly, the gray correlation degree between the candidate perturbation factor and the membrane group feature value is calculated.

[0045] For any candidate perturbation factor, the grey correlation degree between the candidate perturbation factor and the biochemical group feature value, and the grey correlation degree between the candidate perturbation factor and the membrane group feature value are averaged to obtain the comprehensive correlation degree of the candidate perturbation factor, which characterizes the influence of the candidate perturbation factor on the overall state of the system.

[0046] Candidate disturbance factors are ranked from high to low according to their comprehensive correlation scores. The top half of the candidate disturbance factors are included in the main disturbance factor set, and the remaining candidate disturbance factors are included in the secondary volatility factor set.

[0047] It should be noted that if there are extreme cases where candidate perturbation factors have the same comprehensive correlation score and cross the boundary between the main set and the secondary set, they will be uniformly classified into the main perturbation factor set.

[0048] For example, the classification of primary and secondary disturbance factors in grey relational analysis is illustrated in Table 2 below;

[0049] Table 2: Example of Grey Relational Analysis;

[0050] Candidate perturbation factor Correlation with biochemical eigenvalues Correlation with membrane feature values Overall Relevance Ranking Partition results Influent COD load value 0.852 0.734 0.793 1 Main disturbance factors Biological pool water temperature 0.821 0.698 0.760 2 Main disturbance factors Mixed liquor sludge concentration value 0.785 0.712 0.749 3 Main disturbance factors Inflow Flow Fluctuation 0.562 0.648 0.605 4 Secondary volatility factors Rate of change of return flow 0.498 0.534 0.516 5 Secondary volatility factors

[0051] Understandably, the significance of step two lies in the following: This step introduces grey relational analysis to dynamically quantify and rank the correlation strength between candidate perturbation factors and two sets of feature values, thereby adaptively dividing external factors into a set of primary perturbation factors and a set of secondary fluctuation factors. Existing technologies rely on fixed experience to classify primary and secondary factors, which cannot adapt to the dynamic migration of influence caused by seasonal and water quality changes. This step realizes real-time quantitative screening of factor influence, ensuring that strong influencing factors obtain decisive weight in diagnosis while retaining independent analysis channels for weak influencing factors, thus solving the problem of mismatch in diagnostic boundaries caused by static classification of primary and secondary factors.

[0052] Step 3: Using the set of main disturbance factors as the target, call the distribution range of two sets of feature values ​​corresponding to different combinations of main disturbance factors in the historical normal operation, and fit to obtain the operating constraint line under the action of the main disturbance factors;

[0053] In step three, "historical normal operation" means that all indicators of the effluent water quality of the leachate treatment system have been consistently up to standard, and the system has not experienced any abnormal alarms or undergone any unplanned equipment maintenance or process adjustments. The operating data during the historical normal operation period represents the normal working status of the system under conditions free from abnormal interference. The system extracts all data that meet the standards within the most recent period (such as the past three months).

[0054] In step three, the process of calling the distribution ranges of two sets of eigenvalues ​​corresponding to different combinations of major disturbance factors during normal operation is as follows:

[0055] The extracted historical normal operation data are grouped and categorized according to different combinations of values ​​of the main disturbance factors (such as influent COD load, biological treatment tank water temperature, and mixed liquor sludge concentration). For example, the influent COD load is divided into three intervals: low load, medium load, and high load; the biological treatment tank water temperature is divided into three intervals: low temperature, suitable temperature, and high temperature; and the mixed liquor sludge concentration is divided into three intervals: low concentration, medium concentration, and high concentration. The three interval combinations of each of the three factors form a total of twenty-seven different operating condition combinations. The data at each moment is assigned to the corresponding operating condition combination based on the actual values ​​of the three main disturbance factors at the current moment.

[0056] It should be noted that when grouping and classifying different combinations of disturbance factors, the values ​​corresponding to the cumulative frequencies of 33.3% and 66.7% can be used as dividing points according to the frequency distribution of the data, and the main disturbance factors can be divided into three intervals: low, medium and high. For example, the grouping and classification of disturbance factors can be explained in Table 3 below.

[0057] Table 3: Examples of disturbance factor interval division;

[0058] Main disturbance factors unit low range middle section High range Influent COD load value kg / d < 120 120 ~ 280 > 280 Biological pool water temperature ℃ < 18 18 ~ 28 > 28 Mixed liquor sludge concentration value g / L < 8 8 ~ 15 > 15

[0059] For each operating condition combination, the interval between the minimum and maximum values ​​of the biochemical group characteristic data at all times within the operating condition combination is taken as the distribution range of the biochemical group characteristic values ​​under the operating condition combination, and the interval between the minimum and maximum values ​​of the membrane group characteristic data is taken as the distribution range of the membrane group characteristic values. Each operating condition combination corresponds to a rectangular region on the state plane, and the rectangular region reflects all the states that the system can reach under the main disturbance factors of that group when operating normally.

[0060] In step three, the process of fitting the operating constraint lines under the influence of the main perturbation factors is as follows:

[0061] Connect the outer edges of the rectangular regions corresponding to all operating condition combinations to form an envelope. Specifically, on the state plane, for each possible horizontal axis direction (i.e., biochemical characteristic value), find the maximum value of the vertical axis direction (i.e., membrane characteristic value) in all rectangular regions covering the horizontal axis, and connect all the maximum value points to form an upper boundary curve; similarly, find all the minimum value points and connect them to form a lower boundary curve. For example, taking some operating condition combinations as examples, the rectangular region range corresponding to each combination on the state plane is shown in Table 4 below.

[0062] Table 4: Examples of the correspondence between working condition combinations and rectangular areas of the state plane;

[0063] Operating condition number Influent COD load water temperature sludge concentration Biochemical characteristic value range Membrane group characteristic value range 1 Low low temperature Low [45, 68] [120,185] 2 Low low temperature middle [52, 75] [135,210] 3 Low Suitable temperature Low [55, 80] [140,220] 4 middle Suitable temperature middle [72, 105] [180,280] 5 high Suitable temperature high [95, 135] [230,350] 6 high high temperature high [110,155] [260,400]

[0064] It should be noted that in practical applications, only the upper boundary curve is taken as the operating constraint line, because abnormalities in leachate treatment usually manifest as an increase in characteristic values ​​(membrane fouling causes membrane module characteristic values ​​to rise, and biochemical shock causes biochemical module characteristic values ​​to rise), while the lower boundary reflects measurement noise more than process constraints.

[0065] The upper boundary curve represents the upper limit of the system's ability to maintain normal operation under different combinations of the main disturbance factors, i.e., the operating constraint line under the action of the main disturbance factors.

[0066] Understandably, the significance of step three lies in the following: This step targets the main set of disturbance factors, uses historical normal operation data to fit operational constraint lines, and constructs the upper limit boundary of normal system operation under the influence of strong factors; existing single threshold methods treat all factors together and cannot reflect the decisive constraint effect of strong factors on the system's operational capability; the constraint lines established in this step eliminate the interference of secondary factors, truly reflect the system's limit tolerance under the current combination of main disturbance factors, solve the problem of delayed diagnostic response and difficulty in identifying core causes when strong factors suddenly change, and provide an objective benchmark for anomaly judgment.

[0067] Step 4: Using the set of minor fluctuation factors as the target, call up the edge operation data of the effluent water quality that meets the standards but is close to the discharge limit during normal operation in history, fit to obtain the early warning line under the influence of minor fluctuation factors, and combine it with the operation constraint line to divide the early warning concern area under the influence of fluctuation and the normal operating condition area under stable operation in the state plane.

[0068] In step four, the selection criteria for edge operation data are as follows: While all effluent quality indicators of the leachate treatment system meet the standards within a given time period, at least one major effluent indicator (such as effluent COD or effluent ammonia nitrogen) reaches more than 85% of the discharge limit, meaning the margin between the actual effluent value and the discharge limit is less than 15%. This standard reflects that the system, although not yet exceeding the standard, is in an edge operation state close to the limit. All data segments meeting this edge state standard within the most recent period (e.g., the past three months) are extracted to obtain the edge operation data.

[0069] In step four, the process of fitting the early warning line under the influence of the minor volatility factor is as follows:

[0070] The extracted edge operation status data are grouped and categorized according to different combinations of values ​​of the secondary fluctuation factors (such as influent flow fluctuation rate, return flow change rate, and the degree to which effluent COD approaches the discharge limit). The interval division method for the secondary fluctuation factors is consistent with step three, that is, the values ​​corresponding to the cumulative frequencies of each factor in the edge operation dataset of 33.3% and 66.7% are taken as the dividing points, and each secondary fluctuation factor is divided into three intervals: low, medium, and high. The three interval combinations of the three secondary factors form a total of twenty-seven different edge operation condition combinations; the data at each moment is assigned to the corresponding edge operation condition combination based on the actual values ​​of the three secondary fluctuation factors at the current moment.

[0071] For each edge condition combination, the interval between the minimum and maximum values ​​of the biochemical group characteristic data at all times within the condition combination is taken as the distribution range of the biochemical group characteristic values ​​under the condition combination, and the interval between the minimum and maximum values ​​of the membrane group characteristic data is taken as the distribution range of the membrane group characteristic values. Each edge condition combination corresponds to a rectangular region on the state plane, and the rectangular region reflects the characteristic value distribution when the system is in the edge operating state under the conditions of the secondary fluctuation factor.

[0072] Connect the inner edges of the rectangular regions corresponding to all edge operating condition combinations to form an envelope. Specifically, on the state plane, for each possible horizontal axis direction (i.e., biochemical characteristic value), find the maximum value of the vertical axis direction (i.e., membrane characteristic value) in all rectangular regions of edge operating condition combinations covering that horizontal axis, and connect all the maximum value points to form an upper boundary curve. Since the edge operation data comes from the state of effluent water quality approaching the limit rather than exceeding the standard, this boundary curve is located inside the operation constraint line, that is, closer to the origin of the state plane. This upper boundary curve is the early warning line for the system to enter the state of effluent water quality approaching the discharge limit under different combinations of secondary fluctuation factors.

[0073] In step four, the process of dividing the state plane into the early warning concern zone under the influence of fluctuations and the normal operating condition zone under stable operation is as follows:

[0074] The operating constraint line and the early warning line are placed in the state plane. Since the operating constraint line is fitted by the main disturbance factor and the early warning line is fitted by the secondary fluctuation factor, and the edge operating data itself is the part of the normal operating data that is close to the limit, the early warning line is always located inside the operating constraint line.

[0075] The two curves divide the state plane into three regions: the region inside the early warning line and closer to the origin of the coordinate system is the normal operating condition region under stable operation, indicating that the system is in a stable operating state under the combined effect of the current main disturbance factor and the secondary fluctuation factor;

[0076] The annular area between the early warning line and the operational constraint line is the warning concern area under the influence of fluctuations, indicating that although the system has not yet exceeded the operational constraint line, it has entered a state close to the emission limit due to the cumulative influence of secondary fluctuation factors.

[0077] The area outside the operating constraint line indicates that the system has exceeded its normal operating capability under the constraints of the main disturbance factors and belongs to the abnormal range;

[0078] Understandably, the significance of step four lies in the following: This step targets the set of minor fluctuation factors, uses edge operation data to fit an early warning line, and divides the normal operating condition zone and the warning concern zone together with the operation constraint line in the state plane; the existing binary judgment logic lacks an early risk identification mechanism, and the cumulative effect of weak factors is ignored; this step adds an early warning line inside the constraint line, uses edge operation data to reflect the state of approaching the limit caused by the accumulation of weak factors, so that the system can issue targeted warnings in the early stage of state deterioration and before exceeding the limit, thus solving the problem of lacking trend risk warnings.

[0079] Step 5: Output diagnostic conclusions based on the landing point position of the current operating condition in the state plane. When the landing point is in the normal operating condition zone, output normal operation. When it is in the warning concern zone, output membrane fouling risk warning or biochemical shock risk warning according to the tangential deflection direction. When it exceeds the operating constraint line, output membrane fouling acceleration anomaly, biochemical shock anomaly or system instability anomaly according to the tangential deflection direction.

[0080] In step five, the process of outputting diagnostic conclusions based on the landing point position of the current operating condition in the state plane is as follows:

[0081] The biochemical and membrane eigenvalues ​​calculated at the current moment are used as the x-axis and y-axis, respectively, and the landing points are marked on the state plane.

[0082] When the landing point enters the warning concern zone or exceeds the operational constraint line, it is necessary to determine the tangential deflection direction of the landing point in order to determine the specific type of anomaly or risk. The calculation of the tangential deflection direction is as follows:

[0083] When the landing point is located within the warning concern area, calculate the tangential deflection direction of the landing point relative to the early warning line; connect the landing point location to the point on the early warning line closest to the landing point, and calculate the angle between the direction of the connecting line and the tangential direction of the early warning line at that point; if the angle is biased towards the side where the membrane module characteristic value increases, that is, the deviation of the landing point relative to the warning line is mainly reflected in the increase of the membrane module characteristic value, then it is judged as membrane fouling risk; if the angle is biased towards the side where the biochemical module characteristic value increases, that is, the deviation of the landing point relative to the warning line is mainly reflected in the increase of the biochemical module characteristic value, then it is judged as biochemical shock risk.

[0084] When the landing point exceeds the operating constraint line, calculate the tangential deflection direction of the landing point relative to the operating constraint line, using the same calculation method as above; if the deviation of the landing point relative to the operating constraint line is mainly in the direction of the increase in membrane group characteristic values, it is determined to be an accelerated membrane fouling anomaly; if it is mainly in the direction of the increase in biochemical group characteristic values, it is determined to be a biochemical shock anomaly; if both groups of characteristic values ​​increase significantly and the deviation direction is significantly deflected in the positive directions of both coordinate axes, it is determined to be a system instability anomaly.

[0085] The criteria for determining the main direction of increase in membrane group eigenvalues ​​are: landing point direction angle less than 22.5 degrees; the criteria for determining the main direction of increase in biochemical group eigenvalues ​​are: landing point direction angle greater than 67.5 degrees; if the standardized deviation factor of both groups of eigenvalues ​​exceeds 3, and the landing point direction angle is between 22.5 degrees and 67.5 degrees, that is, both directional components are significant and comparable, it is determined to be a system instability anomaly. For example, see Table 5 below.

[0086] Table 5: Examples of criteria for determining the direction of tangential deflection;

[0087] Determine the scene Direction angle range Judgment result θ < 22.5° The main direction is the increase in membrane eigenvalues. Membrane fouling risk / accelerated membrane fouling anomaly 22.5° ≤ θ ≤ 67.5° Both sets of eigenvalues ​​showed a significant increase. System instability anomaly (must simultaneously satisfy the standardization deviation factor of both sets of eigenvalues ​​exceeding 3) θ > 67.5° The main direction is the increase in biochemical eigenvalues. Biochemical shock risk / biochemical shock anomaly

[0088] It should be noted that when the landing point is within the early warning line, it indicates that the system is in a stable operating state under the influence of the current major disturbance factor and minor fluctuation factor, and the diagnostic conclusion output is normal operation. When the landing point is between the early warning line and the operating constraint line, it indicates that although the system has not exceeded the constraint range of the major disturbance factor, it has been affected by the cumulative impact of the minor fluctuation factor and has entered a state where the effluent water quality is close to the discharge limit. At this time, the corresponding level of risk warning is output according to the direction of tangential deflection: if the tangential deflection is mainly in the direction of the increase of membrane module characteristic value, a membrane fouling risk warning is output, and it is recommended to increase the observation frequency of membrane system operation data; if the tangential deflection is mainly in the direction of the increase of biochemical module characteristic value, a biochemical shock risk warning is output, and it is recommended to pay attention to changes in influent water quality or operating parameters of the biological treatment tank; when the landing point exceeds the operating constraint line, it indicates that the major disturbance factor has made it impossible for the system to maintain a normal operating state. Based on the direction of tangential deflection, the specific anomaly type and handling recommendations are output: If the deviation is mainly in the direction of an increase in membrane module characteristic values, it is determined to be an accelerated membrane fouling anomaly, and it is recommended to arrange membrane cleaning or reduce the permeate flux; if the deviation is mainly in the direction of an increase in biochemical group characteristic values, it is determined to be a biochemical shock anomaly, and it is recommended to check the influent water quality or adjust the carbon source dosage; if both sets of characteristic values ​​deviate significantly along the positive directions of both coordinate axes at the same time, it is determined to be a system instability anomaly, and it is recommended to start emergency circulation and investigate the influent source.

[0089] Understandably, the significance of step five lies in the following: This step outputs a graded diagnostic conclusion based on the position of the current operating condition point in the state plane and its tangential deflection direction relative to the warning line or constraint line. Existing technologies lack the ability to identify abnormal types and can only provide general alarms. This step constructs a three-level progressive diagnostic logic of normal operation—early warning—abnormal handling, and can accurately distinguish abnormal types and their severity levels such as membrane fouling, biochemical shock, and system instability based on the deflection direction. This solves the problems of vague handling suggestions and low operation and maintenance efficiency, and realizes a leap from single judgment to graded and accurate diagnosis.

[0090] Example 2: Please refer to Figure 3 As shown in the embodiment of the present invention, the landfill leachate treatment process anomaly diagnosis system based on multi-source sensing includes the following modules:

[0091] State plane construction module: Collect sensor data from the biochemical treatment and membrane separation units during the leachate treatment process, calculate the characteristic values ​​of the biochemical group and the membrane group respectively, and construct the operating state plane;

[0092] Primary and secondary disturbance classification module: Perform grey relational analysis on candidate disturbance factors and two sets of feature values, and classify the candidate disturbance factors into a primary disturbance factor set and a secondary fluctuation factor set according to the degree of correlation.

[0093] Run constraint fitting module: Taking the main disturbance factor set as the target, it calls the distribution range of two sets of feature values ​​corresponding to different combinations of main disturbance factors in the historical normal operation, and fits the running constraint line under the action of the main disturbance factors.

[0094] Early warning boundary partitioning module: Taking the set of minor fluctuation factors as the target, it calls the edge operation data of the effluent water quality that meets the standards but is close to the discharge limit during normal operation in history, fits to obtain the early warning line under the influence of minor fluctuation factors, and combines the operation constraint line to divide the early warning attention area under the influence of fluctuation and the normal operating condition area under stable operation in the state plane.

[0095] The graded anomaly diagnosis module outputs diagnostic conclusions based on the landing point position of the current operating condition in the state plane. When the landing point is in the normal operating condition zone, it outputs normal operation; when it is in the warning concern zone, it outputs membrane fouling risk warning or biochemical shock risk warning according to the tangential deflection direction; when it exceeds the operating constraint line, it outputs membrane fouling acceleration anomaly, biochemical shock anomaly or system instability anomaly according to the tangential deflection direction.

[0096] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for diagnosing abnormality in landfill leachate treatment process based on multi-source sensing, characterized in that: Includes the following steps: Sensor data from the biochemical treatment and membrane separation units during leachate treatment were collected, and characteristic values ​​of the biochemical group and membrane group were calculated respectively. An operating status plane was then constructed. Grey relational analysis was performed on the candidate disturbance factors and two sets of eigenvalues. The candidate disturbance factors were divided into a set of primary disturbance factors and a set of secondary fluctuation factors according to the degree of correlation. Using the set of major disturbance factors as the target, the distribution range of two sets of feature values ​​corresponding to different combinations of major disturbance factors in historical normal operation is called to fit the operating constraint line under the action of major disturbance factors; Using the set of minor fluctuation factors as the target, the edge operation data of the effluent water quality meeting the standards but close to the discharge limit during normal operation in history are called to fit the early warning line under the influence of minor fluctuation factors. Combined with the operation constraint line, the early warning attention area under the influence of fluctuation and the normal operating condition area under stable operation are divided in the state plane. The diagnostic conclusion is output based on the landing point position of the current operating condition in the state plane. When the landing point is in the normal operating condition zone, the normal operation is output. When it is in the warning concern zone, the membrane fouling risk warning or biochemical shock risk warning is output according to the tangential deflection direction. When it exceeds the operating constraint line, the membrane fouling acceleration anomaly, biochemical shock anomaly or system instability anomaly is output according to the tangential deflection direction.

2. The multi-source sensing based landfill leachate treatment process anomaly diagnosis method according to claim 1, characterized in that: The method for calculating biochemical eigenvalues ​​is as follows: The readings of dissolved oxygen, pH, temperature, and redox potential sensors within the previous time window are arranged into a biochemical data matrix. Principal component analysis is performed on the biochemical data matrix to calculate the covariance matrix and solve for the eigenvalues ​​and eigenvectors. The eigenvector corresponding to the largest eigenvalue is taken as the biochemical weight vector. The biochemical weight vector is multiplied by the sensor reading at the end of the time window to obtain the biochemical group eigenvalues.

3. The multi-source sensing based landfill leachate treatment process anomaly diagnosis method according to claim 1, characterized in that: The method for calculating the eigenvalues ​​of the membrane module is as follows: The readings of the pressure sensor, flow sensor, and conductivity sensor within the same time window before the current moment are arranged into a membrane data matrix. Principal component analysis is performed on the membrane data matrix to calculate the covariance matrix and solve for the eigenvalues ​​and eigenvectors. The eigenvector corresponding to the largest eigenvalue is taken as the membrane weight vector. The membrane weight vector is multiplied by the sensor reading at the end of the time window to obtain the membrane group eigenvalues.

4. The multi-source sensing based landfill leachate treatment process anomaly diagnosis method of claim 1, wherein: The process of grey relational analysis is as follows: Record biochemical characteristic value sequences and membrane characteristic value sequences, and use the monitoring data sequences of each candidate perturbation factor within the same time period as comparison sequences; Calculate the absolute value of the deviation between each candidate perturbation factor and the biochemical group feature value sequence at each time step, and average all the absolute values ​​of deviation to obtain the correlation degree with the biochemical group feature value; The absolute value of the deviation between each candidate perturbation factor and the sequence of membrane feature values ​​is calculated at each time step, and the correlation between all the absolute values ​​of deviation and the membrane feature values ​​is obtained by averaging all the absolute values ​​of deviation.

5. The multi-source sensing based landfill leachate treatment process anomaly diagnosis method according to claim 4, characterized in that: The correlation between each candidate perturbation factor and the biochemical group characteristic value and the correlation between each candidate perturbation factor and the membrane group characteristic value are averaged to obtain the comprehensive correlation of each candidate perturbation factor. All candidate perturbation factors are sorted from high to low according to the comprehensive correlation, and the candidate perturbation factors are divided into the main perturbation factor set and the secondary fluctuation factor set according to the sorting results.

6. The method for abnormal diagnosis of landfill leachate treatment process based on multi-source sensing according to claim 1, characterized in that: The process of fitting the constraint line is as follows: Historical normal operation data are grouped and categorized according to different combinations of the main disturbance factors. Each combination corresponds to a rectangular region on the state plane. The horizontal axis of the rectangular region ranges from the minimum to the maximum value of the biochemical group characteristic value under that combination, and the vertical axis ranges from the minimum to the maximum value of the membrane group characteristic value under that combination. The upper boundaries of all rectangular regions are connected to form an envelope, which is used as the operating constraint.

7. The method for abnormal diagnosis of landfill leachate treatment process based on multi-source sensing according to claim 1, characterized in that: The process of fitting the early warning line is as follows: Data from periods when effluent indicators meet standards but at least one effluent indicator monitoring value is close to the discharge limit are extracted as marginal operation data. The marginal operation data are grouped and classified according to different combinations of minor fluctuation factors. Each combination corresponds to a rectangular region on the state plane. The horizontal axis of the rectangular region ranges from the minimum to the maximum value of the biochemical group characteristic value under the combination, and the vertical axis ranges from the minimum to the maximum value of the membrane group characteristic value under the combination. The inner edges of all rectangular regions are connected to form an envelope, and the envelope is taken as the early warning line.

8. The method for abnormal diagnosis of landfill leachate treatment process based on multi-source sensing according to claim 7, characterized in that: The operating constraint line and the early warning line are placed together in the state plane. The early warning line is located inside the operating constraint line. The area inside the early warning line is the normal operating condition area, and the area between the early warning line and the operating constraint line is the warning concern area.

9. The method for abnormal diagnosis of landfill leachate treatment process based on multi-source sensing according to claim 1, characterized in that: The method for calculating and determining the tangential deflection direction is as follows: When the landing point is located within the early warning concern zone, the point closest to the landing point is determined on the early warning line. The direction of the line connecting the landing point and the nearest point is obtained. The angle between the direction of the line and the tangent direction of the early warning line at the nearest point is calculated. If the angle deviates towards the positive direction of the membrane module characteristic value coordinate axis, it is determined as membrane fouling risk. If the angle deviates towards the positive direction of the biochemical group characteristic value coordinate axis, it is determined as biochemical shock risk. When the landing point exceeds the operating constraint line, the point closest to the landing point is determined on the operating constraint line. The tangential deflection direction is calculated. If the increase in membrane module characteristic value is the main deflection direction, it is determined as membrane fouling acceleration anomaly. If the increase in biochemical group characteristic value is the main deflection direction, it is determined as biochemical shock anomaly. If the deflection is simultaneously towards the positive directions of both coordinate axes and both sets of characteristic values ​​exceed the corresponding thresholds, it is determined as system instability anomaly.

10. An anomaly diagnosis system for landfill leachate treatment process based on multi-source sensing, characterized in that: The system is also used to perform the diagnostic method as described in any one of claims 1-9, and includes the following modules: State plane construction module: Collect sensor data from the biochemical treatment and membrane separation units during the leachate treatment process, calculate the characteristic values ​​of the biochemical group and the membrane group respectively, and construct the operating state plane; Primary and secondary disturbance classification module: Perform grey relational analysis on candidate disturbance factors and two sets of feature values, and classify the candidate disturbance factors into a primary disturbance factor set and a secondary fluctuation factor set according to the degree of correlation. Run constraint fitting module: Taking the main disturbance factor set as the target, it calls the distribution range of two sets of feature values ​​corresponding to different combinations of main disturbance factors in the historical normal operation, and fits the running constraint line under the action of the main disturbance factors. Early warning boundary partitioning module: Taking the set of minor fluctuation factors as the target, it calls the edge operation data of the effluent water quality that meets the standards but is close to the discharge limit during normal operation in history, fits to obtain the early warning line under the influence of minor fluctuation factors, and combines the operation constraint line to divide the early warning attention area under the influence of fluctuation and the normal operating condition area under stable operation in the state plane. The graded anomaly diagnosis module outputs diagnostic conclusions based on the landing point position of the current operating condition in the state plane. When the landing point is in the normal operating condition zone, it outputs normal operation; when it is in the warning concern zone, it outputs membrane fouling risk warning or biochemical shock risk warning according to the tangential deflection direction; when it exceeds the operating constraint line, it outputs membrane fouling acceleration anomaly, biochemical shock anomaly or system instability anomaly according to the tangential deflection direction.

Images