A real-time monitoring method and system for sewage treatment data based on the Internet of Things

Abstract

The present invention provides a method and system for real-time monitoring of sewage treatment data based on the Internet of Things, which relates to the field of data processing technology. The method comprises: setting a sensor network at key nodes in the sewage treatment process, collecting a variety of raw data including water quality parameters, flow, temperature, pressure and equipment operation status in real time, and transmitting the various raw data to a cloud data center for pre-processing to generate processed data; according to the processed data, real-time monitoring of water quality change trends, identification of pollution events or equipment failures, and generation of real-time monitoring and analysis results. The present invention realizes real-time monitoring and intelligent analysis of sewage treatment data through the Internet of Things, effectively warns of pollution events or equipment failures, and automatically generates treatment plans, timely initiates emergency responses, and improves sewage treatment efficiency and safety.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a real-time monitoring method and system for sewage treatment data based on the Internet of Things. Background Art

[0002] Some traditional sewage treatment monitoring methods still rely mainly on manual inspections. Due to limitations on manpower and inspection cycles, the monitoring frequency is low and it is impossible to grasp the dynamic changes in the sewage treatment process in real time.

[0003] Although automated instruments can provide a certain degree of automated monitoring, their coverage is limited, and the frequency and accuracy of data updates often cannot meet the needs of real-time monitoring, resulting in untimely data acquisition.

[0004] Due to the untimely acquisition of data and low monitoring frequency, when abnormal water quality or equipment failure is found, the opportunity for treatment has been missed, resulting in reduced treatment efficiency. Manual inspections and limited automated instrument monitoring methods make the response to abnormal situations slow, making it difficult to take quick measures to avoid the expansion of pollution incidents or equipment failures. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a real-time monitoring method and system for sewage treatment data based on the Internet of Things, so as to ensure the continuity and stability of sewage treatment and improve the treatment efficiency and water quality standards.

[0006] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0007] In a first aspect, a real-time monitoring method for sewage treatment data based on the Internet of Things is provided, the method comprising:

[0008] A sensor network is set up at key nodes in the sewage treatment process to collect a variety of raw data including water quality parameters, flow, temperature, pressure and equipment operating status in real time, and transmit the various raw data to the cloud data center for pre-processing to generate processed data;

[0009] Based on the processed data, real-time monitoring of water quality trends can be performed to identify pollution events or equipment failures, and real-time monitoring and analysis results can be generated;

[0010] Based on the monitoring and analysis results, the prediction model is trained in combination with historical data to predict the changes and abnormal conditions of water quality indicators in the future and generate prediction results;

[0011] Compare the prediction results with the real-time monitoring data. When any data exceeds the preset threshold, the alarm scheme is triggered and an early warning signal is generated;

[0012] A sewage treatment plan is automatically generated based on early warning signals and real-time monitoring and analysis results. When an alarm is triggered, emergency response measures are automatically initiated according to the sewage treatment plan, including starting backup equipment and adjusting the treatment process to deal with potential pollution incidents or equipment failures.

[0013] Furthermore, a sensor network is set up at key nodes in the sewage treatment process to collect a variety of raw data including water quality parameters, flow, temperature, pressure and equipment operation status in real time, and transmit the various raw data to the cloud data center for pre-processing to generate processed data, including:

[0014] Deploy sensor networks at key nodes in the sewage treatment process to collect a variety of raw data such as water quality parameters, flow, temperature, pressure and equipment operating status in real time, and transmit the raw data to the cloud data center for storage in real time;

[0015] An initial detection threshold of outliers is set, and the initial detection threshold is used to perform outlier detection on a variety of raw data to obtain a preliminary detected outlier set;

[0016] Get the set of true outliers and calculate the error between the detected outliers and the true outliers under the current threshold;

[0017] Simulate gradient descent, calculate the gradient of the error between the outlier and the true outlier with respect to the threshold, and iteratively adjust the threshold according to the gradient direction. When the number of iterations is reached, the optimized outlier detection threshold is obtained.

[0018] The optimized outlier detection threshold is used to perform outlier detection on various raw data, and the data detected to be beyond the normal fluctuation range is smoothed to generate processed data.

[0019] Furthermore, the gradient calculation formula of the error between the outlier and the true outlier with respect to the threshold is:

[0020] ;

[0021] in, Indicates The threshold value at the iteration; Indicates The threshold value at the iteration; Indicates The learning rate at the iteration; represents the number of samples; Represents the index value; Indicates The true outlier value of samples; , Represents the parameters of the linear prediction function; Indicates The regularization parameter at the iteration; Indicates The amount by which the threshold changes at each iteration.

[0022] Furthermore, based on the processed data, water quality trends can be monitored in real time, pollution events or equipment failures can be identified, and real-time monitoring and analysis results can be generated, including:

[0023] Perform real-time analysis on the water quality data in the processed data to identify the changing trend of water quality parameters over time, including calculating the moving average and standard deviation of each water quality parameter to obtain the analysis results of the water quality changing trend;

[0024] Based on the analysis results of water quality change trends, by setting thresholds and pattern matching, data anomalies of pollution events or equipment failures are identified to generate real-time monitoring and analysis results.

[0025] Furthermore, based on the monitoring and analysis results, the prediction model is trained in combination with historical data to predict the changes and abnormalities of water quality indicators in the future and generate prediction results, including:

[0026] Integrate the real-time monitoring and analysis results with the historical water quality data set to obtain an integrated data set;

[0027] Analyze the integrated data set to identify the characteristics that affect changes in water quality indicators, including historical water quality parameters, timestamps, weather conditions, and equipment operating status;

[0028] The integrated data set is divided into training set and test set in chronological order, and the regression model is used as the prediction model;

[0029] The prediction model is trained using the training set, and during the training process, the prediction model parameters are adjusted to obtain a trained prediction model;

[0030] The trained prediction model is evaluated using the test set, and the prediction error of the trained prediction model on the test set is calculated using the quantile loss function to obtain a verified trained prediction model;

[0031] The current real-time monitoring data is input into the verified trained prediction model to predict the changes and abnormal conditions of water quality indicators in the future, and generate prediction results, including the predicted values ​​of water quality indicators and abnormal situation warnings in the future.

[0032] Furthermore, the calculation formula of the quantile loss function is:

[0033] ;

[0034] in, Indicated in quantile The actual observed value With the predicted value loss due to the difference between It represents the actual observed value of water quality index; It represents the predicted value of water quality index; represents the weight adjustment parameter; Represents the quantile, that is, the number between 0 and 1; represents a positive number; represents a constant term; represents the error sensitivity adjustment parameter; Represents the normalization parameter.

[0035] Furthermore, based on the early warning signals and real-time monitoring and analysis results, a sewage treatment plan is automatically generated. When an alarm is triggered, emergency response measures are automatically initiated according to the sewage treatment plan, including starting backup equipment and adjusting the treatment process to deal with potential pollution incidents or equipment failures, including:

[0036] Conduct real-time monitoring and analysis of early warning signals to determine the pollutant type, pollution scope, specific location, equipment failure type, and impact of the pollution incident;

[0037] Establish a library of various sewage treatment solutions, and determine the initial solution that matches the current pollution situation or equipment failure from the solution library based on the results of early warning signal analysis;

[0038] According to the initial scheme, the optimization target of the sewage treatment scheme is determined, and it is iteratively optimized through the momentum term. In each iteration, the scheme parameters are adjusted according to the gap between the actual treatment effect of the current scheme and the optimization target to generate a new scheme; after multiple iterations, the final sewage treatment scheme is generated;

[0039] According to the final sewage treatment plan, emergency response measures are automatically initiated, including starting backup equipment and adjusting the treatment process. During the implementation of emergency response measures, water quality parameters and equipment status are monitored in real time to deal with potential pollution incidents or equipment failures.

[0040] In the second aspect, a real-time monitoring system for sewage treatment data based on the Internet of Things includes:

[0041] Data acquisition module, used to collect key data in the sewage treatment process in real time, including water quality parameters, equipment operation status, treatment efficiency and energy consumption information;

[0042] The IoT communication module is used to transmit the key data of the data acquisition module to the cloud server;

[0043] The data processing and analysis module is used to receive and process the data from the IoT communication module, conduct real-time analysis to monitor abnormal conditions in the sewage treatment process, and obtain analysis results;

[0044] The early warning and response module is used to connect with the data processing and analysis module. When the analysis result reaches the preset alarm threshold, the early warning signal is automatically triggered and the emergency treatment measures are automatically started;

[0045] The remote monitoring and operation interface module is used to remotely monitor various data of the sewage treatment process in real time, automatically adjust treatment parameters, and view historical data and reports.

[0046] According to a third aspect, a computing device includes:

[0047] one or more processors;

[0048] The storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method described.

[0049] In a fourth aspect, a computer-readable storage medium stores a program, and when the program is executed by a processor, the method described is implemented.

[0050] The above solution of the present invention includes at least the following beneficial effects:

[0051] By setting up sensor networks at key nodes, a variety of raw data are collected in real time and transmitted to the cloud data center for processing, which improves the real-time and accuracy of data acquisition, allowing monitoring personnel to grasp the dynamic changes in the sewage treatment process at the first time. Combined with historical data training prediction models, it is possible to predict future changes and abnormalities in water quality indicators, improving the depth and breadth of data processing.

[0052] Real-time monitoring of water quality trends, identification of pollution events or equipment failures, and generation of real-time monitoring and analysis results enable monitoring personnel to quickly detect abnormalities and take appropriate measures. Combined with the prediction results, potential problems can be warned in advance, further reducing treatment risks and costs.

[0053] When any data exceeds the preset threshold, the alarm scheme is automatically triggered and an early warning signal is generated to ensure that the monitoring personnel can immediately know the abnormal situation. At the same time, according to the early warning signal and the results of real-time monitoring and analysis, the sewage treatment plan is automatically generated, and emergency response measures are automatically started, such as starting backup equipment and adjusting the treatment process, which improves the efficiency and accuracy of emergency response. Through long-term data accumulation and analysis, we can deeply understand the characteristics and laws of the sewage treatment process, and provide a scientific basis for optimizing the treatment process, improving treatment efficiency and water quality standards. At the same time, the application of predictive models also makes the sewage treatment process more intelligent and refined. The introduction of real-time monitoring and automated emergency response reduces the frequency of manual inspections and manual operations, and reduces operation and maintenance costs. At the same time, the early warning of potential problems through predictive models also reduces additional losses caused by equipment failures or pollution incidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a flow chart of a method for real-time monitoring of sewage treatment data based on the Internet of Things provided by an embodiment of the present invention.

[0055] Figure 2 It is a schematic diagram of a real-time monitoring system for sewage treatment data based on the Internet of Things provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0057] like Figure 1 As shown, an embodiment of the present invention proposes a real-time monitoring method for sewage treatment data based on the Internet of Things, and the method includes the following steps:

[0058] Step 11, setting up a sensor network at key nodes in the sewage treatment process to collect various raw data including water quality parameters, flow, temperature, pressure and equipment operation status in real time, and transmitting the various raw data to the cloud data center for pre-processing to generate processed data;

[0059] Step 12, based on the processed data, monitor the water quality change trend in real time, identify pollution events or equipment failures, and generate real-time monitoring and analysis results;

[0060] Step 13, based on the monitoring and analysis results, combined with historical data, the prediction model is trained to predict the changes and abnormal conditions of water quality indicators in the future and generate prediction results;

[0061] Step 14, compare the prediction results with the real-time monitoring data. When any data exceeds the preset threshold, the alarm scheme is triggered and an early warning signal is generated;

[0062] Step 15, automatically generate a sewage treatment plan based on the early warning signal and real-time monitoring and analysis results. When the alarm is triggered, automatically initiate emergency response measures based on the sewage treatment plan, including starting backup equipment and adjusting the treatment process to deal with potential pollution incidents or equipment failures.

[0063] In the embodiment of the present invention, the sensor network can fully cover each key node in the sewage treatment process, ensuring that a variety of raw data such as water quality parameters, flow, temperature, pressure and equipment operating status are collected without omission. By collecting data in real time, the dynamic changes in the sewage treatment process can be reflected immediately, providing a timely and accurate information basis for subsequent monitoring and analysis. By real-time monitoring of water quality change trends, water quality anomalies or equipment failures can be discovered in a timely manner, providing strong support for rapid response and processing. After identifying a pollution event or equipment failure, the problem can be quickly located, reducing troubleshooting time and improving the efficiency of problem handling.

[0064] By using historical data to train the prediction model, we can predict the changes and abnormalities of water quality indicators in the future, and provide a scientific basis for formulating countermeasures in advance. The prediction results can provide data support for decision-making optimization in the sewage treatment process, helping managers to formulate more reasonable and effective treatment plans.

[0065] By comparing the prediction results with the real-time monitoring data, data anomalies can be discovered in time and alarm solutions can be triggered to ensure that the problem is handled in a timely manner. The issuance of early warning signals can remind managers to take timely measures to reduce the risks and losses caused by abnormal water quality or equipment failure. According to the early warning signals and real-time monitoring and analysis results, the sewage treatment plan is automatically generated to achieve intelligent treatment and reduce human intervention. Automatically starting emergency response measures, such as starting backup equipment and adjusting treatment processes, can quickly deal with potential pollution incidents or equipment failures and ensure the continuity and stability of sewage treatment. Automated emergency response measures can handle problems more accurately and efficiently, and improve the quality and efficiency of sewage treatment.

[0066] In a preferred embodiment of the present invention, in step 11, a sensor network is set up at key nodes in the sewage treatment process to collect a variety of raw data including water quality parameters, flow, temperature, pressure and equipment operation status in real time, and transmit the various raw data to a cloud data center for preprocessing to generate processed data, which may include:

[0067] Step 112, deploying a sensor network at key nodes in the sewage treatment process to collect various raw data such as water quality parameters, flow, temperature, pressure and equipment operation status in real time, and transmitting the various raw data to a cloud data center for storage in real time;

[0068] Step 113, setting an initial detection threshold of outliers, and using the initial detection threshold to perform outlier detection on a variety of raw data to obtain a set of preliminarily detected outliers;

[0069] Step 114, obtaining a set of true outlier values, and calculating the error between the detected outlier values ​​and the true outlier values ​​under the current threshold;

[0070] Step 115, simulating gradient descent, calculating the gradient of the error between the outlier and the true outlier with respect to the threshold, and iteratively adjusting the threshold according to the gradient direction, and obtaining the optimized outlier detection threshold when the number of iterations is reached;

[0071] Step 116, using the optimized outlier detection threshold, performs outlier detection on various raw data, and performs smoothing processing on the data detected to be out of the normal fluctuation range to generate processed data.

[0072] In the embodiment of the present invention, according to the process flow and monitoring requirements of sewage treatment, key nodes where sensors need to be deployed are determined, such as water inlets, treatment units, water outlets, etc. Appropriate sensor types are selected according to monitoring parameters (water quality parameters, flow, temperature, pressure, equipment operating status, etc.) to ensure the accuracy and reliability of data collection. The selected sensors are installed at key nodes in the sewage treatment process according to predetermined positions and methods to form a sensor network. The sensor network is configured to enable real-time data collection and transmit the data to a cloud data center for storage via wired or wireless means.

[0073] Step 113: Analyze the normal fluctuation range of each monitoring parameter using historical data to provide a basis for setting the initial detection threshold. Set the initial detection threshold of each monitoring parameter based on the historical data analysis results, combined with expert experience and actual needs.

[0074] Step 114: annotate abnormal events in the sewage treatment process through manual inspection to form a set of true abnormal values. Align the manually annotated true abnormal values ​​with the original data collected by the sensor network to ensure data consistency.

[0075] Step 115, select an initial threshold (It may be a range, such as arrive ), as the starting point for outlier detection. Using the initial threshold Perform outlier detection on the original data to obtain a set of detected outliers. Compare with the true outlier set and calculate the error. For each threshold parameter, calculate the gradient of the error function with respect to the parameter. Select a learning rate, which determines the step size of the threshold adjustment in each iteration. Iterate and update the threshold according to the gradient direction and the learning rate. Re-perform outlier detection using the updated threshold and calculate the new error. If the error meets the predetermined convergence condition (such as reaching the maximum number of iterations), stop the iteration. When the iteration ends, the optimized outlier detection threshold is obtained. This implementation process automatically adjusts the threshold of outlier detection through the gradient descent algorithm to minimize the detection error, thereby obtaining more accurate outlier detection results.

[0076] In step 116, it is assumed that an optimized outlier detection threshold has been obtained through gradient descent.

[0077] Iterate over each data point in the original dataset , and the threshold For comparison. > , then mark is an outlier. Otherwise, mark It is a normal value.

[0078] The moving average method is selected as the smoothing technique. The fluctuation is reduced by calculating the average value within a certain window around the data point. For each data point marked as an outlier , find the surrounding data points. Calculate this The average of the data points , replace the original outliers with the average value; where, Indicates the first Data points The moving average of Indicates the window size of the moving average; express; Represents the index variable. Initialize a new data set to store the processed data. For each original data point , check if it is marked as an outlier. If it is an outlier, add the previously calculated smoothed value to the new data set. If it is not an outlier, directly add the original value Add to the new dataset. Before outputting the final processed data, some basic verification can be performed, such as checking whether the data still maintains the original trend and pattern, and whether the outliers have been effectively smoothed. Save the processed data to a file.

[0079] For example, pH sensors are deployed at key nodes such as the water inlet, regulating tank, biochemical tank and outlet of the sewage treatment process to collect pH data in real time and transmit it to the cloud data center. According to historical data, the initial detection threshold of pH is set to 6.5~8.5 (within the normal range). Through manual inspection, the actual abnormal pH events are marked to form a set of real abnormal values. The initial detection threshold is optimized using the gradient descent algorithm. After multiple iterations, it is found that when the pH detection threshold is set to 6.8~8.2, the error between the detected abnormal value and the real abnormal value is the smallest. The optimized pH detection threshold (6.8~8.2) is used to detect abnormal values ​​of pH data collected in real time. For data outside this range, the moving average method is used for smoothing to generate processed data. In this way, the authenticity of the data is retained and the impact of abnormal values ​​on data processing and analysis is eliminated.

[0080] By collecting and transmitting a variety of raw data in real time, as well as optimizing the outlier detection threshold, it is possible to more accurately identify abnormal situations in the sewage treatment process and improve the accuracy and efficiency of monitoring. The processed data provides strong support for the operational decisions of the sewage treatment plant. Based on these data, the treatment process can be adjusted in time and the equipment operating parameters can be optimized to ensure that the effluent water quality meets the standards and reduce operating costs. Through real-time monitoring and outlier detection, equipment failures or process abnormalities can be discovered in time, so that preventive measures can be taken to avoid the expansion of failures or accidents, and to ensure the continuity and stability of the sewage treatment process. The realization of the entire process relies on advanced technologies such as the Internet of Things, big data, and artificial intelligence, which have promoted the intelligent management process of the sewage treatment industry. This improves management efficiency.

[0081] In a preferred embodiment of the present invention, the gradient calculation formula of the error between the outlier value and the true outlier value with respect to the threshold is:

[0082] ;

[0083] in, Indicates The threshold value at the iteration; Indicates The threshold value at the iteration; Indicates The learning rate at the iteration; represents the number of samples; Represents the index value; Indicates The true outlier value of samples; , Represents the parameters of the linear prediction function; Indicates The regularization parameter at the iteration; Indicates The amount by which the threshold changes at each iteration.

[0084] In the embodiment of the present invention, the initialization threshold , which can be a value based on experience, historical data, or preliminary analysis. Set the learning rate , which determines the step size of the threshold adjustment at each iteration. Determine the regularization parameter , used to control the smoothness of threshold changes or to prevent overfitting. Prepare sample data, including the true outlier value of each sample (This is annotated by experts.) =0, indicating that this is the first iteration. For each sample, calculate the current threshold The predicted outliers and the true outliers under The error between . The outlier value can be predicted by the linear prediction function To calculate. Calculate the error about the threshold The gradient of . This gradient represents the rate of change of the error under the current threshold, that is, how to adjust the threshold to reduce the error. The part is the derivative of the error with respect to the threshold, multiplied by This is because what we are looking for is the direction that reduces the error. is a regularization term used to prevent the threshold from changing too much or too quickly, where is the change in the threshold between the current iteration and the previous iteration. Based on the calculated gradient and learning rate, the threshold is updated .

[0085] Learning Rate The step size of the update is controlled, and the direction of the gradient points to the direction of error reduction. Check whether the conditions for stopping the iteration are met, such as reaching the maximum number of iterations. If the conditions are met, the iteration stops and the optimized threshold is output. The optimized threshold is applied to the outlier detection of the sensor network to improve the accuracy and reliability of the detection.

[0086] By iteratively adjusting the threshold, the final solution can be gradually approached, minimizing the error between the detected outliers and the true outliers. This means that the system can more accurately identify abnormalities in the sewage treatment process, reduce false positives and false negatives, and improve the reliability of monitoring. The learning rate in the formula is and the regularization parameter It can be dynamically adjusted according to the iterative process, so that the system can better adapt to different operating environments and data characteristics. This adaptability enables the system to adjust the threshold more quickly and maintain the accuracy of detection when faced with changes in the sewage treatment process (such as water quality fluctuations, equipment aging, etc.). By accurately identifying outliers, the system can more effectively allocate resources to deal with these abnormal situations, such as timely adjusting the treatment process, repairing equipment, or taking other countermeasures. This helps to reduce the waste of resources, improve overall operational efficiency, and reduce operating costs. The optimized outlier detection threshold can be used as an important component of the intelligent management system, combined with other monitoring parameters and control systems to achieve comprehensive intelligent management of the sewage treatment process. This helps to improve the efficiency and effectiveness of sewage treatment while reducing the cost and risk of manual intervention.

[0087] In a preferred embodiment of the present invention, the above step 12, based on the processed data, real-time monitoring of water quality change trends, identification of pollution events or equipment failures, and generation of real-time monitoring and analysis results may include:

[0088] Step 121, performing real-time analysis on the water quality data in the processed data to identify the changing trend of the water quality parameters over time, including calculating the moving average and standard deviation of each water quality parameter to obtain the analysis result of the water quality changing trend;

[0089] Step 122, based on the analysis results of the water quality change trend, by setting thresholds and pattern matching, data anomalies of pollution events or equipment failures are identified to generate real-time monitoring and analysis results.

[0090] In an embodiment of the present invention, pre-processed water quality data is obtained from a cloud data center, including time series data of various water quality parameters (such as pH value, dissolved oxygen, turbidity, etc.). For each water quality parameter, a moving window size is set (such as the past 1 hour, the past 24 hours, etc.). Within the moving window, the average value of the water quality parameter is calculated as the moving average value at the current time point. The moving average value can smooth data fluctuations and more clearly show the changing trend of water quality parameters. Within the same moving window, the standard deviation of each water quality parameter is calculated to measure the degree of discreteness of the data. The moving average value and standard deviation of each water quality parameter are displayed in a graphical form, such as a time series graph, a line graph, etc. By comparing the moving average value and standard deviation at different time points, the changing trend of the water quality parameters, such as rise, fall, fluctuation, etc., can be analyzed.

[0091] Step 122, collect time series data of various water quality parameters (such as pH value, dissolved oxygen, turbidity, etc.) from the historical database. Perform statistical analysis on the historical data of each water quality parameter, and calculate statistical quantities such as mean, standard deviation, maximum value, minimum value, etc. Based on the statistical quantities, set the normal fluctuation range, slight abnormal range, and severe abnormal range. Among them, represents the mean; Represents the standard deviation, for example:

[0092] Normal fluctuation range .

[0093] Slightly abnormal range .

[0094] Severe abnormal range .

[0095] Store the threshold range in the configuration file or database. Collect historical pollution events or equipment failure cases, including event time, event type, affected water quality parameters, etc. For each abnormal event, extract abnormal data features, such as the size, duration, and change trend of the abnormal value. Extract data features before the abnormal event, such as the fluctuation and change trend of water quality parameters, to establish an early warning model.

[0096] The extracted abnormal data features and change trends are organized into abnormal patterns and stored in the pattern library.

[0097] Abnormal patterns include but are not limited to: sudden increase / decrease, continuous increase / decrease, increased fluctuation, etc. As new abnormal events emerge, the abnormal pattern library is continuously updated and improved.

[0098] Acquire water quality data in real time from the sensor network or cloud data center, including the current value and historical data of each water quality parameter. For each water quality parameter, calculate the moving average and standard deviation at the current time point according to the set moving window size. Compare the calculated moving average and standard deviation with the set threshold range. If the moving average or standard deviation of a water quality parameter exceeds the set threshold range, an abnormal alarm is triggered. Match the water quality data that triggers the abnormal alarm with the abnormal pattern in the pattern library.

[0099] The similarity between real-time data and abnormal patterns is calculated using a similarity algorithm (cosine similarity). The formula for calculating the cosine similarity is: ;in, is the value of cosine similarity, which represents the real-time water quality data vector and the exception pattern vector The angle between The cosine of ; Represents real-time water quality data vector The elements; Represents an abnormal pattern vector The elements; Indicates the number of water quality parameters; Represents the index variable. The range of cosine value is [−1, 1]. =1, it means that the two vectors completely coincide with each other; when =0, it means the two vectors are orthogonal (i.e. the angle between them is 90 degrees); when =−1, it means that the two vectors are in opposite directions. In water quality anomaly detection, we usually focus on The value is close to 1 because it means that the real-time water quality data is very similar to the abnormal pattern and similar abnormal events may have occurred.

[0100] According to the similarity, identify whether there are similar abnormal data features. Combine the pattern matching results and real-time data features to further determine whether the abnormality is caused by pollution events or equipment failures. Auxiliary judgment can be made based on the type of abnormal events recorded in the abnormal pattern library, the affected water quality parameters and other information. When an abnormality is identified, a real-time monitoring and analysis result containing abnormal time, abnormal water quality parameters, abnormal type (pollution or failure), abnormal degree and other information is generated.

[0101] ;

[0102] Among them, the abnormal time can be a timestamp or a specific date and time; the abnormal water quality parameter is the name of the water quality parameter that triggers the abnormality; the abnormal type is "pollution" or "fault"; the degree of abnormality can be quantified according to the degree of exceeding the threshold or the similarity of pattern matching.

[0103] Suppose that in a sewage treatment plant, the sensor network monitors in real time that the moving average of the pH value in a certain period of time suddenly drops, and the standard deviation increases. According to the set threshold range, the system determines that the pH value has exceeded the normal fluctuation range and entered the slightly abnormal range. At the same time, the system finds through pattern matching that the characteristics of this sudden drop in pH value and increased fluctuation are similar to a pollution incident caused by a leak of acidic substances in history. Therefore, the system generates real-time monitoring and analysis results, indicating that there may be a pollution incident, and recommends immediate inspection and treatment.

[0104] By analyzing water quality data in real time, it is possible to quickly identify water quality trends and abnormal data, and improve monitoring efficiency. By setting reasonable thresholds and abnormal pattern libraries, potential pollution events or equipment failures can be discovered in advance, providing early warnings for timely measures. The generated real-time monitoring and analysis results provide managers with accurate and comprehensive water quality information, which helps to make more scientific decisions, such as adjusting treatment processes and repairing equipment. By timely discovering and processing abnormal data, it is possible to reduce operational risks caused by water quality problems, such as substandard effluent quality and equipment damage. In-depth analysis of water quality data and mining of valuable information can provide strong support for the optimized operation and scientific management of sewage treatment plants.

[0105] In a preferred embodiment of the present invention, the above step 13, based on the monitoring and analysis results, combines historical data to train a prediction model to predict the changes and abnormal conditions of water quality indicators in the future period of time, and generates prediction results, which may include:

[0106] Step 131, integrating the real-time monitoring and analysis results and the historical water quality data set to obtain an integrated data set;

[0107] Step 132, analyzing the integrated data set to determine the characteristics that affect the change of water quality indicators, including historical water quality parameters, timestamps, weather conditions, and equipment operating status;

[0108] Step 133, dividing the integrated data set into a training set and a test set in chronological order, and using the regression model as a prediction model;

[0109] Step 134, training the prediction model using the training set, and adjusting the prediction model parameters during the training process to obtain a trained prediction model;

[0110] Step 135, using the test set to evaluate the trained prediction model, using the quantile loss function to calculate the prediction error of the trained prediction model on the test set, so as to obtain a verified trained prediction model;

[0111] Step 136, input the current real-time monitoring data into the verified trained prediction model to predict the changes and abnormal conditions of water quality indicators in the future, and generate prediction results, including the predicted values ​​of water quality indicators in the future and abnormal condition warnings.

[0112] In an embodiment of the present invention, real-time monitoring and analysis results are obtained from a sensor network or cloud data center, including the current value, moving average, standard deviation, and abnormal alarm information of each water quality parameter. Historical water quality data sets are extracted from a database or storage system, including past water quality parameter values, timestamps, weather conditions, equipment operating status, etc. The real-time monitoring data and historical data are cleaned to remove duplicates, missing or abnormal values. The real-time monitoring data and historical data are aligned according to the timestamps and merged into an integrated data set. If there is a lack of real-time monitoring data at certain time points, interpolation methods can be used to fill in the gaps.

[0113] Step 132, based on domain knowledge and data analysis, determine the key features that affect the changes in water quality indicators. These features may include historical water quality parameters (such as pH, dissolved oxygen, turbidity, etc.), timestamps (used to capture seasonal or periodic changes), weather conditions (such as temperature, humidity, rainfall, etc.), and equipment operating status (such as the operating status of the pump, the cleanliness of the filter, etc.). Preprocess the selected features, such as standardization, normalization, encoding (such as converting categorical variables to numerical variables), etc., to make them suitable for model training. Use the Pearson correlation coefficient to calculate the linear correlation between the features. According to the absolute value of the correlation coefficient, filter out the features that are highly correlated with the changes in water quality indicators. The closer the absolute value of the correlation coefficient is to 1, the stronger the linear relationship between the two features. Draw a correlation heat map or scatter plot matrix to intuitively display the correlation between the features.

[0114] Step 133, divide the integrated data set into a training set and a test set in chronological order. Older data is used for training, and newer data is used for testing. Select a regression model for water quality prediction, such as linear regression. Use the training set to train the prediction model, and minimize the prediction error by adjusting model parameters (such as learning rate, number of trees, number of layers, etc.). Use the test set to evaluate the trained prediction model and calculate the quantile loss function. According to the evaluation results, tune the model, such as adjusting hyperparameters, adding or deleting features, changing the model structure, etc. After multiple iterations of training and evaluation, a verified trained prediction model is obtained.

[0115] Step 134, input the current real-time monitoring data (including water quality parameter values, timestamps, weather conditions, equipment operating status, etc.) into the verified trained prediction model. The model predicts the changes and abnormalities of water quality indicators in the future (such as hours, days or weeks) based on the input data.

[0116] The prediction results are presented to users in the form of charts, reports or alarms, including predicted values ​​of future water quality indicators, abnormal situation warnings (such as water quality predicted to exceed set thresholds) and recommended response measures.

[0117] Prediction models can provide early warning of abnormal water quality conditions, providing decision makers with enough time to take countermeasures, such as adjusting water treatment processes and increasing monitoring frequency. Prediction models can help us optimize resource allocation, such as increasing monitoring equipment or personnel input when water quality is predicted to deteriorate, and reducing resource input when water quality is stable. Decisions based on prediction results are more scientific, accurate, and efficient, and can reduce resource waste or environmental risks caused by blind decision-making. By real-time monitoring and prediction of water quality indicators, water resources can be better protected and sustainable use of water resources and environmental protection can be promoted.

[0118] In a preferred embodiment of the present invention, the calculation formula of the quantile loss function is:

[0119] ;

[0120] in, Indicated in quantile The actual observed value With the predicted value loss due to the difference between It represents the actual observed value of water quality index; It represents the predicted value of water quality index; represents the weight adjustment parameter; Represents the quantile, that is, the number between 0 and 1; represents a positive number; represents a constant term; represents the error sensitivity adjustment parameter; Represents the normalization parameter.

[0121] In the embodiment of the present invention, all parameters of the quantile loss function are clarified and initialized. is the quantile parameter, which ranges from 0 to 1. It determines which position in the predicted distribution is of interest, such as the median ( = 0.5) or other percentiles. Weight adjustment parameter used to control the sensitivity of the loss function to prediction error. A smaller value will make the loss function more sensitive to the prediction error, and vice versa. Is a very small positive number used to prevent the denominator from being zero during calculations and to ensure numerical stability. Set to a number very close to zero. It is a constant term that provides a base value for the loss function and can affect the overall offset of the loss function. is the error sensitivity adjustment parameter, which is used to further adjust the impact of the prediction error on the loss function value. Values ​​amplify the impact of forecast errors.

[0122] It is a normalization parameter, which is used to normalize the prediction error to avoid excessive fluctuations in the loss function value due to differences in water quality index values.

[0123] Calculate the actual observations With the predicted value The absolute difference between . This difference reflects the degree of deviation between the predicted value and the actual value. Calculate the quantile deviation term, which reflects the difference between the position of the predicted value relative to the actual value in the distribution and the quantile of interest Here we use the absolute value function and To ensure that the denominator is not zero, and when When close to zero, the deviation term can transition smoothly.

[0124] Substitute the quantile deviation term into the calculation formula of the weight factor to obtain the weight factor The weight factor is a number between 0 and 1 that dynamically adjusts the weight of the loss based on the quantile and the size of the prediction error. When the prediction error is consistent with the quantile deviation, the weight factor is close to 1; when the prediction error is opposite to the quantile deviation, the weight factor is close to 0. In order to normalize the prediction error, the normalized error term is calculated Finally, substitute each item into the calculation formula of the quantile loss function to get the loss value This loss value reflects the The loss caused by the difference between the actual observed value and the predicted value under the condition of water quality index is considered comprehensively, including the size, direction, quantile of the prediction error and the size difference of the water quality index value.

[0125] By adjusting the quantile parameter , the quantile loss function can adapt to different prediction requirements. For example, when = 0.5, the loss function focuses on the median prediction, which is very useful for predicting the medium level of water quality indicators. Values ​​can be used to obtain prediction results of different quantiles to meet different decision-making needs. Weight adjustment parameters Controls the sensitivity of the loss function to prediction errors. When it is larger, the loss function is more sensitive to the prediction error, making the model more focused on reducing the error during training. This dynamic weight adjustment helps the model better fit the data and improves the prediction accuracy. Error sensitivity adjustment parameter This allows further tuning of the impact of prediction error on the loss function. By increasing The value of can amplify the impact of prediction error, making the model more sensitive to the error, thus prompting the model to work harder to reduce the error during training. The introduction of helps to avoid numerical problems caused by differences in water quality index values. Through normalization, it can ensure that the loss function has relatively stable performance under different water quality index values. Normalization also helps to improve the generalization ability of the model. By standardizing the measurement of prediction errors, the model can better adapt to different ranges of water quality index values, thereby improving the accuracy and stability of predictions.

[0126] In a preferred embodiment of the present invention, the above step 14, comparing the prediction result with the real-time monitoring data, triggering an alarm scheme and generating an early warning signal when any data exceeds a preset threshold, may include:

[0127] Get the predicted values ​​of water quality indicators from the prediction model. These predicted values ​​are based on historical data, time series analysis, machine learning models, etc. Get the actual observed values ​​of current water quality indicators in real time from the water quality monitoring system. Set reasonable thresholds for each water quality indicator according to water quality management standards or actual needs. These thresholds can be upper limits, lower limits, or both upper and lower limits. Create a variable or flag to record the current alarm status, which is initially set to "no alarm". Traverse the prediction results and real-time monitoring data to ensure that each pair of predicted values ​​and actual observed values ​​are compared. For each water quality indicator, compare its predicted value and actual observed value with the set threshold. If the value of any data exceeds the preset threshold (whether it is the predicted value or the actual observed value), enter the alarm judgment process.

[0128] In the comparison judgment, if it is confirmed that there is data exceeding the threshold and the alarm conditions are met (such as exceeding a certain time continuously, exceeding a certain degree, etc.), the alarm is triggered. Update the alarm status to "alarm", and record the alarm time, water quality indicators, values ​​exceeding the threshold, etc. Generate corresponding early warning signals according to the preset alarm scheme. This can be email, SMS notification, system log record, sound and light alarm, etc. Notify the early warning signal to relevant personnel, such as water quality management personnel, emergency response team, etc., in time so that they can take timely measures. After the alarm is triggered, continue to monitor the changes in water quality indicators in real time to assess whether the alarm situation continues or worsens. Record all relevant information of the alarm event in detail, including alarm time, water quality indicators, predicted values, actual observed values, degree of exceeding the threshold, alarm processing process, etc. When the water quality indicators return to the normal range and remain stable for a period of time, the alarm status can be lifted and the relevant personnel can be notified.

[0129] In a preferred embodiment of the present invention, the above step 15 automatically generates a sewage treatment plan based on the early warning signal and the real-time monitoring and analysis results. When an alarm is triggered, emergency response measures are automatically initiated according to the sewage treatment plan, including starting backup equipment and adjusting the treatment process to deal with potential pollution incidents or equipment failures, which may include:

[0130] Step 151, real-time monitoring and analysis of the early warning signal to determine the pollutant type, pollution range, specific location and equipment failure type and impact of the pollution event;

[0131] Step 152, establishing a library containing multiple sewage treatment schemes, and determining an initial scheme matching the current pollution situation or equipment failure from the scheme library according to the early warning signal analysis results;

[0132] Step 153, according to the initial scheme, determine the optimization target of the sewage treatment scheme, and perform iterative optimization through the momentum term. In each iteration, adjust the scheme parameters according to the gap between the actual treatment effect of the current scheme and the optimization target to generate a new scheme; after multiple iterations, generate the final sewage treatment scheme;

[0133] Step 154, automatically start emergency response measures according to the final sewage treatment plan, including starting backup equipment and adjusting the treatment process. During the execution of emergency response measures, water quality parameters and equipment status are monitored in real time to deal with potential pollution incidents or equipment failures.

[0134] In an embodiment of the present invention, an early warning signal containing pollution event or equipment failure information is received from an early warning system. The type of pollutant is determined using sensor data, historical pollution event records, etc. The spread range and speed of pollutants are evaluated by combining real-time monitoring data and pollution diffusion models. The specific location of the pollution event is accurately located using technologies such as GPS and sensor networks. The equipment failure signal is analyzed, and the type of failure and the degree of impact are determined by combining equipment maintenance records and fault diagnosis algorithms. The analysis results are organized into a report, including detailed information such as pollutant type, pollution range, specific location, equipment failure type and degree of impact.

[0135] Step 152, collect and organize a variety of sewage treatment solutions, including treatment methods for different pollutants, different pollution levels, and different equipment failures. Number and classify each solution for quick retrieval and matching. According to the analysis results of step 151, retrieve the initial solution that best matches the current pollution situation or equipment failure from the solution library. Consider the characteristics of the pollutants, the difficulty of treatment, the equipment status and other factors to select a suitable initial solution.

[0136] Step 153, based on factors such as pollution conditions, equipment conditions, and treatment costs, determine the optimization objectives of the sewage treatment scheme, such as minimizing treatment time, maximizing treatment efficiency, minimizing costs, etc. Initialize scheme parameters, such as treatment agent dosage, treatment time, equipment operating parameters, etc. Use momentum terms for iterative optimization, that is, adjust scheme parameters to generate new schemes based on the gap between the actual treatment effect of the current scheme and the optimization goal. After each iteration, evaluate the treatment effect of the new scheme. If the optimization goal is met or the upper limit of the number of iterations is reached, stop the iteration; otherwise, continue the iteration. After multiple iterations, generate a final sewage treatment scheme that meets the optimization goal. Compare the final scheme with the initial scheme to evaluate the optimization effect.

[0137] Step 154, according to the final sewage treatment plan, automatically start the emergency response measures, including starting the backup equipment, adjusting the treatment process, etc. Ensure that all relevant equipment is in normal working condition and operates according to the plan requirements. During the implementation of the emergency response measures, monitor the water quality parameters and equipment status in real time to ensure that the treatment effect meets expectations. If abnormal conditions are found or the treatment effect is not good, adjust the plan or take other measures in time. When the pollution incident is properly handled and the water quality returns to normal, stop the emergency response measures. Summarize and evaluate the treatment process to provide a reference for future pollution incident handling.

[0138] Suppose one day, the early warning system receives a signal indicating that the water quality in a certain area has suddenly deteriorated, the dissolved oxygen content has dropped sharply, and there is an odor. The system immediately uses sensor data and historical pollution event records for analysis and determines that the pollutants are organic pollutants (such as organic matter in domestic sewage). Combining real-time monitoring data and pollution diffusion models, the system assesses that pollutants are spreading downstream at a speed of X kilometers per hour, and the possible impact range is up to Y square kilometers. Using GPS and sensor networks, the system accurately locates the source of pollution as a sewer leak in a nearby community. At the same time, the equipment monitoring system reports that an aerator has a fault, the fault type is motor overheating, and the impact is medium. These analysis results are compiled into a report, including detailed information such as pollutant type (organic pollutant), pollution range (Y square kilometers), specific location (sewer in a community), equipment fault type (aerator motor overheating) and impact (medium), and sent to the central control room.

[0139] After receiving the report, the intelligent system immediately retrieves the initial solution that best matches the current pollution situation from the pre-established sewage treatment solution library. The solution library contains a variety of treatment methods for different pollutants, different pollution levels, and different equipment failures, and each solution is numbered and classified. According to the pollutant type (organic pollutant), pollution range (Y square kilometers) and equipment failure type (aerator motor overheating) in the report, the system selects an initial solution that includes increasing aeration volume, adding biological agents, adjusting treatment processes, and other steps. At the same time, the system also considers factors such as the characteristics of the pollutants (easy to biodegrade), the difficulty of treatment (medium) and the condition of the equipment (partial failure of the aerator). The system determines the optimization objectives of the sewage treatment solution based on factors such as pollution conditions, equipment conditions, and treatment costs, such as minimizing treatment time, maximizing treatment efficiency, and minimizing costs. Then, the system initializes the solution parameters, such as the proportion of aeration volume increase, the amount of biological agents added, and the adjustment method of the treatment process. Next, the system uses the momentum term for iterative optimization. In each iteration, the system adjusts the scheme parameters to generate a new scheme based on the gap between the actual treatment effect of the current scheme (such as the rate of improvement of dissolved oxygen content, the removal rate of organic pollutants, etc.) and the optimization target. After multiple iterations, the system generates a final sewage treatment scheme that meets the optimization target. While maintaining the treatment efficiency, this scheme also reduces the treatment cost and shortens the treatment time.

[0140] Through automated monitoring, analysis, and solution generation, the emergency response time is shortened, and the spread and harm of pollution incidents are reduced. Through iterative optimization algorithms, the treatment plan can be dynamically adjusted according to actual conditions to ensure the treatment effect. By accurately controlling parameters such as reagent dosage and treatment time, the treatment cost is reduced and the economic benefit is improved. By real-time monitoring of equipment status and water quality parameters, equipment failures and abnormal conditions can be discovered and handled in a timely manner, enhancing the stability and reliability of the system. The automated and intelligent sewage treatment solution generation and execution process improves the management and technical level of sewage treatment, and provides strong support for future pollution incident handling.

[0141] like Figure 2 As shown, an embodiment of the present invention further provides a real-time monitoring system for sewage treatment data based on the Internet of Things, comprising:

[0142] Data acquisition module, used to collect key data in the sewage treatment process in real time, including water quality parameters, equipment operation status, treatment efficiency and energy consumption information;

[0143] The IoT communication module is used to transmit the key data of the data acquisition module to the cloud server;

[0144] The data processing and analysis module is used to receive and process the data from the IoT communication module, conduct real-time analysis to monitor abnormal conditions in the sewage treatment process, and obtain analysis results;

[0145] The early warning and response module is used to connect with the data processing and analysis module. When the analysis result reaches the preset alarm threshold, the early warning signal is automatically triggered and the emergency treatment measures are automatically started;

[0146] The remote monitoring and operation interface module is used to remotely monitor various data of the sewage treatment process in real time, automatically adjust treatment parameters, and view historical data and reports.

[0147] It should be noted that the system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0148] The embodiment of the present invention further provides a computing device, comprising: a processor, a memory storing a computer program, wherein when the computer program is executed by the processor, the method described above is executed. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0149] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A real-time monitoring method for sewage treatment data based on the Internet of Things, characterized in that: The method comprises: A sensor network is set up at key nodes in the sewage treatment process to collect a variety of raw data including water quality parameters, flow, temperature, pressure and equipment operating status in real time, and transmit the various raw data to the cloud data center for pre-processing to generate processed data; Based on the processed data, real-time monitoring of water quality trends can be performed to identify pollution events or equipment failures, and real-time monitoring and analysis results can be generated; Based on the monitoring and analysis results, the prediction model is trained in combination with historical data to predict the changes and abnormal conditions of water quality indicators in the future and generate prediction results; Compare the prediction results with the real-time monitoring data. When any data exceeds the preset threshold, the alarm scheme is triggered and an early warning signal is generated; Automatically generate sewage treatment plans based on early warning signals and real-time monitoring and analysis results. When an alarm is triggered, automatically initiate emergency response measures based on the sewage treatment plan, including starting backup equipment and adjusting treatment processes to deal with potential pollution incidents or equipment failures; Among them, a sensor network is set up at key nodes in the sewage treatment process to collect a variety of raw data including water quality parameters, flow, temperature, pressure and equipment operation status in real time, and transmit a variety of raw data to the cloud data center for pre-processing to generate processed data, including: Deploy sensor networks at key nodes in the sewage treatment process to collect a variety of raw data such as water quality parameters, flow, temperature, pressure and equipment operating status in real time, and transmit the raw data to the cloud data center for storage in real time; An initial detection threshold of outliers is set, and the initial detection threshold is used to perform outlier detection on a variety of raw data to obtain a preliminary detected outlier set; Get the set of true outliers and calculate the error between the detected outliers and the true outliers under the current threshold; Simulate gradient descent, calculate the gradient of the error between the outlier and the true outlier with respect to the threshold, and iteratively adjust the threshold according to the gradient direction. When the number of iterations is reached, the optimized outlier detection threshold is obtained. The optimized outlier detection threshold is used to perform outlier detection on various raw data, and the data detected to be beyond the normal fluctuation range is smoothed to generate processed data.

2. The method for real-time monitoring of sewage treatment data based on the Internet of Things according to claim 1 is characterized in that: The gradient calculation formula of the error between the outlier and the true outlier with respect to the threshold is: ； in, Indicates The threshold value at the iteration; Indicates The threshold value at the iteration; Indicates The learning rate at the iteration; represents the number of samples; Represents the index value; Indicates The true outlier value of samples; , Represents the parameters of the linear prediction function; Indicates The regularization parameter at the iteration; Indicates The amount by which the threshold changes at each iteration.

3. The method for real-time monitoring of sewage treatment data based on the Internet of Things according to claim 2 is characterized in that: Based on the processed data, water quality trends are monitored in real time, pollution events or equipment failures are identified, and real-time monitoring and analysis results are generated, including: Perform real-time analysis on the water quality data in the processed data to identify the changing trend of water quality parameters over time, including calculating the moving average and standard deviation of each water quality parameter to obtain the analysis results of the water quality changing trend; Based on the analysis results of water quality change trends, by setting thresholds and pattern matching, data anomalies of pollution events or equipment failures are identified to generate real-time monitoring and analysis results.

4. The method for real-time monitoring of sewage treatment data based on the Internet of Things according to claim 3 is characterized in that: Based on the monitoring and analysis results, combined with historical data, the prediction model is trained to predict the changes and abnormal conditions of water quality indicators in the future and generate prediction results, including: Integrate the real-time monitoring and analysis results with the historical water quality data set to obtain an integrated data set; Analyze the integrated data set to identify the characteristics that affect changes in water quality indicators, including historical water quality parameters, timestamps, weather conditions, and equipment operating status; The integrated data set is divided into training set and test set in chronological order, and the regression model is used as the prediction model; The prediction model is trained using the training set, and during the training process, the prediction model parameters are adjusted to obtain a trained prediction model; The trained prediction model is evaluated using the test set, and the prediction error of the trained prediction model on the test set is calculated using the quantile loss function to obtain a verified trained prediction model; The current real-time monitoring data is input into the verified trained prediction model to predict the changes and abnormal conditions of water quality indicators in the future, and generate prediction results, including the predicted values ​​of water quality indicators and abnormal situation warnings in the future.

5. The method for real-time monitoring of sewage treatment data based on the Internet of Things according to claim 4 is characterized in that: The calculation formula of the quantile loss function is: ； in, Indicated in quantile The actual observed value With the predicted value the loss resulting from the difference between It represents the actual observed value of water quality index; It represents the predicted value of water quality index; represents the weight adjustment parameter; Represents the quantile, that is, the number between 0 and 1; represents a positive number; represents a constant term; represents the error sensitivity adjustment parameter; Represents the normalization parameter.

6. The method for real-time monitoring of sewage treatment data based on the Internet of Things according to claim 5 is characterized in that: Automatically generate sewage treatment plans based on early warning signals and real-time monitoring and analysis results. When an alarm is triggered, automatically initiate emergency response measures based on the sewage treatment plan, including starting backup equipment and adjusting the treatment process to deal with potential pollution incidents or equipment failures, including: Conduct real-time monitoring and analysis of early warning signals to determine the pollutant type, pollution scope, specific location, equipment failure type, and impact of the pollution incident; Establish a library of various sewage treatment solutions, and determine the initial solution that matches the current pollution situation or equipment failure from the solution library based on the results of early warning signal analysis; According to the initial scheme, the optimization target of the sewage treatment scheme is determined, and it is iteratively optimized through the momentum term. In each iteration, the scheme parameters are adjusted according to the gap between the actual treatment effect of the current scheme and the optimization target to generate a new scheme; after multiple iterations, the final sewage treatment scheme is generated; According to the final sewage treatment plan, emergency response measures are automatically initiated, including starting backup equipment and adjusting the treatment process. During the implementation of emergency response measures, water quality parameters and equipment status are monitored in real time to deal with potential pollution incidents or equipment failures.

7. A real-time monitoring system for sewage treatment data based on the Internet of Things, the system implementing the method according to any one of claims 1 to 6, characterized in that: include: Data acquisition module, used to collect key data in the sewage treatment process in real time, including water quality parameters, equipment operation status, treatment efficiency and energy consumption information; The IoT communication module is used to transmit the key data of the data acquisition module to the cloud server; The data processing and analysis module is used to receive and process the data from the IoT communication module, conduct real-time analysis to monitor abnormal conditions in the sewage treatment process, and obtain analysis results; The early warning and response module is used to connect with the data processing and analysis module. When the analysis result reaches the preset alarm threshold, the early warning signal is automatically triggered and the emergency treatment measures are automatically started; The remote monitoring and operation interface module is used to remotely monitor various data of the sewage treatment process in real time, automatically adjust treatment parameters, and view historical data and reports.

8. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method as claimed in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program, which, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images