A multifunctional integrated water quality online monitoring method, system and device

Abstract

This invention belongs to the field of water quality monitoring technology, specifically relating to a multifunctional integrated online water quality monitoring method, system, and device. This invention performs anomaly analysis and correlation analysis on multimodal water quality data collected by a sensor matrix. Through multiple verifications of anomaly and correlation data, the accuracy of anomaly analysis is improved, providing strong data support for the execution of subsequent response strategies. Simultaneously, based on the characteristics of the anomaly data, combined with the current device power consumption and weather forecasts for a preset future time period, different power consumption levels are selected to dynamically adjust the upper limits of data acquisition and upload frequencies for each sensor in the sensor matrix. This corrects the execution of the response strategy, reducing the likelihood of the device running out of power during periods of insufficient sunlight, and facilitating flexible adjustment of device power consumption to cope with insufficient sunlight conditions.

Description

Technical Field

[0001] This invention belongs to the field of water quality monitoring technology, specifically relating to a multifunctional integrated online water quality monitoring method, system, and device. Background Technology

[0002] The multifunctional integrated online water quality monitoring system combines detection and analysis. It can not only automatically detect water quality, but also analyze the detection results in real time. It has multiple functions such as water intake, pretreatment, detection, data acquisition and processing, control, remote monitoring system and alarm.

[0003] However, current integrated online water quality monitoring often relies on mains power, making it difficult to cover remote areas with inadequate infrastructure, such as the upper reaches of rivers. Although some monitoring systems are powered by solar energy, the power consumption of the equipment is difficult to adjust according to weather changes, and the equipment is prone to running out of power during periods of insufficient sunlight. How to flexibly adjust the power consumption of the equipment to cope with insufficient sunlight has become an urgent problem to be solved. Summary of the Invention

[0004] The purpose of this invention is to provide a multifunctional integrated online water quality monitoring method, system, and device. It can execute a response strategy based on multiple verifications of abnormal data and related data, and select different power consumption level strategies by combining the current power consumption of the device and the weather forecast results within a preset time period in the future to correct the execution of the response strategy, thereby facilitating flexible adjustment of device power consumption to cope with insufficient sunshine weather conditions.

[0005] The specific technical solution adopted by this invention is as follows:

[0006] A multifunctional integrated online water quality monitoring method, comprising:

[0007] Multimodal water quality data is collected through a sensor matrix, which includes various sensors and various types of water quality data.

[0008] Anomaly analysis is performed on multimodal water quality data to identify anomalous data.

[0009] For each abnormal data, determine whether there is any associated abnormal data and determine the abnormal status of the abnormal data. If there is associated abnormal data, it is determined to be a real abnormality and the first response strategy is executed. Otherwise, it is determined to be a suspected abnormality and the second response strategy is executed. The first response strategy includes increasing the acquisition frequency of the sensor corresponding to the abnormal data and determining whether to increase the data upload frequency. The second response strategy includes reviewing the abnormal data and re-determining the abnormal status.

[0010] Based on the type of data anomaly, a first reference weight is determined; based on the current device battery level, a second reference weight is determined; and based on the weather forecast results for the next preset time period, a third reference weight is determined.

[0011] Based on the first reference weight, the second reference weight, and the third reference weight, a power allocation strategy for dynamically adjusting the sensor acquisition frequency and data upload frequency is determined.

[0012] In a preferred embodiment, the specific steps for anomaly analysis of multimodal water quality data are as follows:

[0013] Time alignment and normalization were performed on the multimodal water quality data;

[0014] Establish a time sliding window, calculate the trend slope, and compare it with the historical trend slope benchmark to identify abnormal trends;

[0015] Establish a dynamic threshold interval based on historical data, compare the current data point with the dynamic threshold interval, identify candidate outliers, analyze the isolation of candidate outliers in the time series, and identify outliers.

[0016] By fusing abnormal trends and outliers, and calculating confidence levels, we can determine whether water quality data is abnormal.

[0017] In a preferred embodiment, the steps of establishing a dynamic threshold interval based on historical data, comparing the current data point with the dynamic threshold interval, identifying candidate outliers, analyzing the isolation of candidate outliers in the time series, and determining outliers are as follows:

[0018] Historical water quality data is acquired, and a dynamic threshold band is established. If the water quality data exceeds the dynamic threshold band, it is marked as a candidate outlier.

[0019] Based on the historical behavior patterns of data points, analyze the isolation degree of data points in their own parameter time series. If the isolation degree is greater than the preset isolation threshold, it is judged as an outlier.

[0020] In a preferred embodiment, for each abnormal data point, determining whether there are related abnormal data points and determining the abnormal state of the abnormal data involves the following specific steps:

[0021] Based on the preset associated database, determine whether there is associated data for the abnormal data. If associated data exists, identify whether the associated data is abnormal. If the associated data is abnormal, determine that the abnormal state of the abnormal data is a real abnormality and execute the first response strategy.

[0022] If the abnormal data has no associated data or the associated data of the abnormal data is not abnormal, the abnormal status of the abnormal data is determined to be a suspected abnormality, and the second response strategy is executed.

[0023] After executing the second response strategy, if the data collected upon review is still abnormal, the abnormal status will be upgraded from suspected abnormality to actual abnormality; if the data collected upon review is not abnormal, the abnormal status will be downgraded from suspected abnormality to no abnormality.

[0024] In a preferred embodiment, the first response strategy includes increasing the acquisition frequency of the sensor corresponding to the abnormal data and determining whether to increase the data upload frequency, with the specific steps as follows:

[0025] Increase the sampling frequency of sensors corresponding to water quality data that show real anomalies, and increase the sampling frequency of sensors corresponding to water quality data that are strongly correlated with such water quality data.

[0026] The decision to increase the data upload frequency is based on the type of abnormal data. If the abnormal data is a key indicator, the data upload frequency is increased; otherwise, it is not.

[0027] In a preferred embodiment, the second response strategy includes reviewing and collecting abnormal data and re-determining the abnormal state. The specific steps are as follows:

[0028] Establish a temporary time window, and increase the acquisition frequency of the sensors corresponding to the abnormal data within the temporary time window to obtain high-frequency verification data;

[0029] Perform anomaly analysis on the high-frequency review data again. If the data is identified as anomalous, it is determined to be a real anomaly; otherwise, it is considered as not being anomalous.

[0030] In a preferred embodiment, the steps for determining a first reference weight based on the type of data anomaly, a second reference weight based on the current device battery level, and a third reference weight based on weather forecasts for a preset future time period are as follows:

[0031] Based on the preset level of data anomaly types, a first reference weight is assigned, with higher-level data anomaly types assigned a higher first reference weight.

[0032] Based on the preset power range level, identify the power range corresponding to the current device power level and assign a second reference weight, wherein data in the high power range is assigned a higher second reference weight.

[0033] Based on weather forecasts, sunshine data is obtained. According to a pre-set sunshine level table, the sunshine level is determined and assigned a third reference weight. Higher sunshine levels are assigned a higher third reference weight.

[0034] In a preferred embodiment, the power allocation strategy for dynamically adjusting the sensor acquisition frequency and data upload frequency based on the first reference weight, the second reference weight, and the third reference weight is determined through the following steps:

[0035] Calculate the comprehensive evaluation value based on the first reference weight, the second reference weight, and the third reference weight;

[0036] Compare the comprehensive evaluation value with the threshold range to select the power consumption level;

[0037] Determine the dynamic adjustment strategy based on the power consumption level.

[0038] The present invention also provides a multifunctional integrated online water quality monitoring system, which uses the above-described multifunctional integrated online water quality monitoring method, including:

[0039] The data acquisition module is used to collect multimodal water quality data through a sensor matrix, wherein the sensor matrix includes multiple sensors and the multimodal water quality data includes multiple types of water quality data.

[0040] The anomaly analysis module is used to perform anomaly analysis on multimodal water quality data and identify abnormal data.

[0041] The correlation analysis module is used to determine whether there are related abnormal data for each abnormal data and to determine the abnormal status of the abnormal data. If there are related abnormal data, it is determined to be a real abnormality and the first response strategy is executed; otherwise, it is determined to be a suspected abnormality and the second response strategy is executed. The first response strategy includes increasing the collection frequency of the sensor corresponding to the abnormal data and determining whether to increase the data upload frequency. The second response strategy includes reviewing the abnormal data and re-determining the abnormal status.

[0042] The weighting analysis module is used to determine the first reference weight based on the type of data anomaly, the second reference weight based on the current device power level, and the third reference weight based on the weather forecast results for a preset future time period.

[0043] The power allocation module is used to determine a power allocation strategy for dynamically adjusting the sensor acquisition frequency and data upload frequency based on the first reference weight, the second reference weight, and the third reference weight.

[0044] And, a multifunctional integrated online water quality monitoring device, the monitoring device comprising:

[0045] At least one processor;

[0046] and a memory communicatively connected to the at least one processor;

[0047] The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the multifunctional integrated online water quality monitoring method according to any one of claims 1 to 8.

[0048] The technical effects achieved by this invention are as follows:

[0049] This invention performs anomaly and correlation analysis on multimodal water quality data collected by a sensor matrix. Multiple verifications of the anomaly and correlation data improve the accuracy of the anomaly analysis, providing strong data support for the execution of subsequent response strategies. Simultaneously, based on the characteristics of the anomaly data, combined with the current device power consumption and weather forecasts for a preset time period, different power consumption levels are selected to dynamically adjust the upper limits of data acquisition and upload frequencies for each sensor in the sensor matrix. This corrects the execution of the response strategy, reducing the likelihood of the device running out of power during periods of insufficient sunlight, and facilitating flexible power consumption adjustments to cope with insufficient sunlight conditions. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0051] Figure 2 This is a schematic diagram of the system modules of the present invention;

[0052] Figure 3 This is a schematic diagram of the monitoring device of the present invention. Detailed Implementation

[0053] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0054] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0055] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in a preferred embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0056] Please see Figure 1 As shown, the present invention provides a multifunctional integrated online water quality monitoring method, comprising:

[0057] S1. Collect multimodal water quality data through a sensor matrix, wherein the sensor matrix includes multiple sensors and the multimodal water quality data includes multiple types of water quality data;

[0058] In step S1, during the online monitoring of water quality, data is first collected from the water body through the device's sensor matrix to acquire multimodal water quality data. The sensor matrix consists of various types of sensors, such as pH sensors, dissolved oxygen sensors, turbidity sensors, conductivity sensors, and temperature sensors. These sensors are integrated into the monitoring device and connected to the data acquisition unit wirelessly or via wired means. The multimodal water quality data is a collection of various water quality data collected by multiple types of sensors, including pH value, dissolved oxygen, turbidity, conductivity, and water temperature, thereby achieving comprehensive coverage and real-time acquisition of water quality indicators, ensuring the diversity and integrity of the data, and providing a basis for subsequent anomaly analysis.

[0059] S2. Perform anomaly analysis on multimodal water quality data to identify anomalous data;

[0060] In step S2, anomaly analysis is performed on the multimodal water quality data to identify abnormal data, enabling real-time monitoring and early warning of water quality status, and providing a basis for subsequent response strategies. The specific steps for anomaly analysis of the multimodal water quality data are as follows:

[0061] Time alignment and normalization were performed on the multimodal water quality data;

[0062] Establish a time sliding window, calculate the trend slope, and compare it with the historical trend slope benchmark to identify abnormal trends;

[0063] Establish a dynamic threshold interval based on historical data, compare the current data point with the dynamic threshold interval, identify candidate outliers, analyze the isolation of candidate outliers in the time series, and identify outliers.

[0064] By fusing abnormal trends and outliers, calculating confidence levels, and determining whether water quality data is abnormal, the system can effectively analyze these trends and calculate confidence levels.

[0065] Specifically, the multimodal water quality data is first preprocessed. Time alignment ensures that data from all sensors are aligned in timestamps, guaranteeing synchronization across timelines. Normalization transforms data with different dimensions and numerical ranges to the same scale, eliminating dimensional interference. A fixed-length window (e.g., the most recent 30 minutes) is established and allowed to slide over time, always analyzing data within the most recent period. Linear fitting is performed on the data points within this window to calculate the trend slope, representing the recent state of the water quality data (including rising, falling, and stable states). Historical slopes are calculated based on historical data (e.g., data from the same time period over the past week), and a normal slope range is statistically determined as a historical trend slope benchmark. For example, the slow rise in dissolved oxygen each morning due to the onset of photosynthesis serves as a historical trend slope benchmark. The current trend slope is compared with this historical trend slope benchmark to identify abnormal trends. The Z-Score method can be used to compare and determine abnormal trends. The expression for calculating the Z-Score is: In the formula, The standard score represents how many standard deviations a raw data point deviates from the mean of its dataset. This indicates the slope of the trend within the current time window. This represents the average historical slope. The standard deviation of the historical slope, if If the value exceeds the preset trend threshold (which can be 2), it is considered an abnormal trend. The trend threshold is set based on the probabilistic statistical characteristics of the standard normal distribution and extensive backtesting of historical water quality data in this monitoring scenario. When the trend threshold is set to 2, if... A value greater than 2 indicates that the current slope deviates from the historical average level by more than 95% of historical occurrences, which is a rare event and highly suspicious. When identifying outliers, historical water quality data is acquired and processed, a dynamic threshold interval is established, and the current data point is compared with the dynamic threshold interval. If a data point exceeds the dynamic threshold interval, it is initially marked as a candidate outlier. Then, it is analyzed whether the candidate outlier is isolated in its own time series. For example, if the candidate outlier is a high value, and it is surrounded by a series of equally high values ​​before and after it, then the candidate outlier may not be an anomaly, but rather an overall increase in values. If it is surrounded by low values ​​before and after it, making it isolated in the time series, then the candidate outlier is more likely to be a true anomaly. If the isolation degree of the candidate outlier is greater than the preset isolation threshold, it is determined to be an outlier. When calculating the isolation degree, n data points (5 points before and after) are taken as the center of the candidate outlier point to form a local window for calculating the isolation degree. The expression for calculating the isolation degree is: In the formula, Indicates the degree of isolation. Indicates the candidate outlier value. This represents the median of all data within the local window, excluding the candidate point itself. This represents the average absolute deviation of all data within the local window, excluding the candidate point itself. If the isolation degree is greater than a preset isolation threshold (which can be 3), it indicates that the point has high isolation, and the candidate outlier is determined to be an outlier. The isolation threshold is set based on the Wright criterion established in robust statistics. The Wright criterion originates from the characteristics of the normal distribution and is used to determine whether there are outliers in the data. Finally, the results of the outlier trend and outliers are quantified into a score on the same scale. The expression for quantifying the outlier trend is: In the formula, Indicates the strength of the abnormal trend. The standard score representing an abnormal trend, and the expression for quantifying outliers, are as follows: In the formula, Indicates the intensity of outliers. express, This represents the preset isolation threshold. Then, the intensity of abnormal trends and the intensity of outliers are fused. The fusion expression is: In the formula, Indicates the fusion score, Indicates the abnormal trend coefficient. Indicates the strength of the abnormal trend. Represents the outlier coefficient. Indicates the intensity of outliers. , and The confidence scores can be 0.6 and 0.4 respectively. The confidence score is calculated using the fusion score, and the expression for calculating the confidence score is: In the formula, Indicates the confidence level. The fusion score is represented by a confidence level that is compared with a preset confidence threshold (which can be 80%). If the confidence level is greater than the preset confidence threshold, the water quality data is determined to be abnormal; otherwise, the water quality data is determined to be non-abnormal.

[0066] Secondly, the multimodal water quality data were time-aligned and normalized. The specific steps are as follows:

[0067] Establish a unified timestamp sequence and resample the water quality data collected by each sensor to make all water quality data relatively aligned on the time axis;

[0068] Specifically, resampling uses linear interpolation to unify data from different sampling frequencies onto a 1-minute time series, ensuring data synchronization. Normalization uses the Z-score standardization method, where the standardization expression is: In the formula, This represents the standardized data value. Represents the original data value. This represents the mean of the dataset. The standard deviation of a dataset is primarily used to ensure that the data conforms to a standard normal distribution. and All calculations are based on historical datasets, which must include at least 30 days of continuous monitoring data, covering typical hydrological conditions (such as dry season and wet season), and the effective data ratio of the dataset must be ≥95%. Missing data are filled using linear interpolation. Take the arithmetic mean of the historical dataset and calculate it using the following expression: In the formula, The total number of historical data points. For the first One data value, Take the standard deviation of the historical dataset and calculate it using the following formula: μ and σ are automatically updated every 7 days to ensure that the normalized baseline adapts to data changes. During the update, a rolling window method is used to retain the latest 30 days of data.

[0069] Secondly, calculate the trend slope, the specific steps are as follows:

[0070] By fitting the short-term trend line of water quality data, the slope of the trend line is calculated.

[0071] Specifically, the least squares method is used to fit the linear trend line, where the expression for calculating the slope is: In the formula, Indicates the slope. Indicates a time index. This represents water quality data values. This indicates the total number of data points used for analysis.

[0072] Secondly, a dynamic threshold range is established based on historical data. The specific steps are as follows:

[0073] Acquire historical water quality data and establish dynamic threshold bands based on the acquired historical data;

[0074] Specifically, historical water quality data is first acquired. Based on a rolling window calculation, upper and lower thresholds are set to establish a dynamic threshold band. When establishing the dynamic threshold band, a day is divided into 24-hour segments, and an independent threshold interval is established for each hour. All data points belonging to the hour segment are extracted from the preprocessed historical data, and the percentiles of these data points are calculated. The upper threshold is the 95th percentile of the data for that hour, and the lower threshold is the 5th percentile of the data for that hour. The data extraction and calculation are repeated for 24 hours to obtain 24 pairs of upper and lower limit values, which are connected to form a dynamic threshold band. Thus, a dynamic threshold band is established based on the acquired historical data.

[0075] S3. For each abnormal data, determine whether there is any related abnormal data and determine the abnormal status of the abnormal data. If there is related abnormal data, it is determined to be a real abnormality and the first response strategy is executed. Otherwise, it is determined to be a suspected abnormality and the second response strategy is executed. The first response strategy includes increasing the acquisition frequency of the sensor corresponding to the abnormal data and determining whether to increase the data upload frequency. The second response strategy includes re-collecting the abnormal data and re-determining the abnormal status.

[0076] In step S3, after identifying the water quality data as abnormal, it is further determined whether there is related data and whether the related data is abnormal. Through correlation data analysis, the accuracy of anomaly detection is improved, and false alarms and false negatives are reduced. If there is abnormal related data, it is determined to be a real anomaly, and the first response strategy is executed; otherwise, it is determined to be a suspected anomaly, and the second response strategy is executed. Through differentiated response strategies, resource optimization and rapid response are achieved. Specifically, for each abnormal data, the steps for determining whether there is abnormal related data are as follows:

[0077] Based on the preset associated database, determine whether there is associated data for the abnormal data. If associated data exists, identify whether the associated data is abnormal. If the associated data is abnormal, determine that the abnormal state of the abnormal data is a real abnormality and execute the first response strategy.

[0078] If the abnormal data has no associated data or the associated data of the abnormal data is not abnormal, the abnormal status of the abnormal data is determined to be a suspected abnormality, and the second response strategy is executed.

[0079] After executing the second response strategy, if the data collected upon review is still abnormal, the abnormal status will be upgraded from suspected abnormality to actual abnormality; if the data collected upon review is not abnormal, the abnormal status will be downgraded from suspected abnormality to no abnormality.

[0080] Specifically, the association database is built based on historical data mining, and the Pearson correlation coefficient is used to calculate the correlation between parameters. When building the association database, continuous and complete historical data is obtained (the historical data period can be one year and cover all seasons). The Pearson correlation coefficient is calculated between all pairs of water quality data parameters. The expression for calculating the Pearson correlation coefficient is as follows: In the formula, This represents the Pearson correlation coefficient. Indicates sample size. This represents the value of the i-th data point in the first type of water quality data parameters. This represents the average value of the first type of water quality data parameter. This represents the value of the i-th data point in the second type of water quality data parameters. This represents the average value of the second type of water quality data parameter. The closer to 1, the stronger the linear relationship. If a strong correlation is found, all water quality data parameters and Pearson correlation coefficients are stored in a structured file to construct a correlation database. Using the current anomalous data as the primary parameter, the database searches for all correlation parameters defined as related to the primary parameter. These are the correlation data for the anomalous data, used to determine if they exist. If they do, the correlation data is further checked for abnormality. If both the anomalous data and its correlation data are abnormal, the anomalous data is considered a true anomaly, and a first response strategy is executed. If no correlation data exists, or if correlation data exists but is not abnormal, the anomalous data is considered a suspected anomaly, and a second response strategy is executed. The second response strategy further assesses the suspected anomaly. The first response strategy includes increasing the sampling frequency of the sensor corresponding to the anomalous data and determining whether to increase the data upload frequency. After determining a true anomaly, the sensor corresponding to the anomalous data with a true anomaly status is identified, and other parameters strongly correlated with the anomalous parameter are identified based on the correlation database. The sensors corresponding to the abnormal data have three preset acquisition frequencies: normal (once every 5 minutes), enhanced (once every minute), and emergency (once every 30 seconds). To increase the acquisition frequency of the sensor corresponding to abnormal data, the frequency can be increased from the normal to the enhanced level or vice versa. Based on common knowledge, parameters that have a significant impact on aquatic ecological health and public safety are considered key indicators. If the abnormal data is a key indicator, the data upload frequency for that abnormal data is increased; otherwise, it is not increased. This allows for rapid capture of the dynamic changes in abnormal data, facilitating more accurate analysis. The second response strategy for assessing the scale and trend of abnormal data includes reviewing and collecting the abnormal data, re-evaluating the abnormal state, and setting a temporary monitoring window with a preset duration during the review and collection. Based on engineering experience and statistical laws regarding the dynamic characteristics of typical water quality parameter changes, the length of this temporary monitoring window can be set to 5 minutes. Within this temporary window, only the collection frequency of the sensor corresponding to the suspected abnormal data is increased to obtain high-frequency review data. This high-frequency review data is then subjected to anomaly analysis again. If it is identified as abnormal data, it is determined to be a true anomaly; otherwise, it is considered as no anomaly.

[0081] Secondly, the first response strategy includes increasing the sampling frequency of the sensors corresponding to the abnormal data and determining whether to increase the data upload frequency. The specific steps are as follows:

[0082] Increase the sampling frequency of sensors corresponding to water quality data that show real anomalies, and increase the sampling frequency of sensors corresponding to water quality data that are strongly correlated with such water quality data.

[0083] Determine whether to increase the data upload frequency based on the type of abnormal data. If the abnormal data belongs to a key indicator, then increase the data upload frequency; otherwise, do not increase it.

[0084] Specifically, when increasing the sampling frequency of sensors, the increase in the sampling frequency of the sensor corresponding to the water quality data that shows a real anomaly can be equivalent to the increase in the sampling frequency of the sensor corresponding to the water quality data that is strongly correlated with that water quality data. For example, if the sampling frequency of the sensor corresponding to the water quality data that shows a real anomaly is increased from the normal level to the enhanced level, the sampling frequency of the sensor corresponding to the water quality data that is strongly correlated with that water quality data is also increased from the normal level to the enhanced level.

[0085] Secondly, the second response strategy includes reviewing and collecting abnormal data to reassess the abnormal status. The specific steps are as follows:

[0086] Establish a temporary time window, and increase the acquisition frequency of the sensors corresponding to the abnormal data within the temporary time window to obtain high-frequency verification data;

[0087] Perform anomaly analysis on the high-frequency review data again. If the data is identified as anomalous, it is determined to be a real anomaly; otherwise, it is considered to be without anomaly.

[0088] Specifically, the steps for performing anomaly analysis on high-frequency verification data are similar to those for performing anomaly analysis on multimodal water quality data, and will not be repeated here.

[0089] S4. Determine the first reference weight based on the type of data anomaly, determine the second reference weight based on the current device power level, and determine the third reference weight based on the weather forecast results for the future preset time period.

[0090] In step S4, the data anomaly type, current device battery level, and weather forecast results for a preset future time period provide three dimensions of quantitative input for the subsequent power allocation strategy. This facilitates a balance between data accuracy, event urgency, and device sustainability. The specific steps are as follows:

[0091] Based on the preset level of data anomaly types, a first reference weight is assigned, with higher-level data anomaly types assigned a higher first reference weight.

[0092] Based on the preset power range level, identify the power range corresponding to the current device power level and assign a second reference weight, wherein data in the high power range is assigned a higher second reference weight.

[0093] Based on the weather forecast results, sunshine data is obtained. According to the preset sunshine level table, the sunshine level is determined and assigned a third reference weight. Among them, the higher the sunshine level, the higher the third reference weight is assigned.

[0094] Specifically, an anomaly classification table is established. Based on common knowledge, various water quality parameters are classified according to their impact on the ecological environment and public safety, into three levels: Level 1 (critical / high-risk), Level 2 (important), and Level 3 (general / indicative). Level 1 anomalies include toxicity and microbiological indicators, with a corresponding weight of 0.8. Level 2 anomalies include high concentrations of ammonia nitrogen and significant deviations from pH values, with a corresponding weight of 0.5. Level 3 anomalies include water temperature and conductivity, with a corresponding weight of 0.2. To determine the primary reference weight, the power range of the equipment is divided into several intervals, and a baseline weight is assigned to each interval. This can result in three power ranges: a high power range, a medium power range, and a high power range. The system is divided into three reference weights: a high battery range (80% to 100% with a weight of 0.8), a medium battery range (30% to 79% with a weight of 0.4), and a low battery range (0% to 29% with a weight of 0.1). A second reference weight is determined by using a communication module (e.g., 4G or 5G) to obtain weather forecasts for a future preset time period (up to 48 hours) from a meteorological service API. These forecasts are then mapped to a preset sunshine level table, where sunny / strong sunshine is assigned a weight of 0.7, cloudy / moderate sunshine a weight of 0.4, and overcast / weak sunshine a weight of 0.2. This determines a third reference weight.

[0095] S5. Based on the first reference weight, the second reference weight, and the third reference weight, determine the power consumption allocation strategy for dynamically adjusting the sensor acquisition frequency and data upload frequency.

[0096] In step S5, the three reference weights are integrated to output an executable, dynamic power allocation strategy, which improves adaptability. The specific steps are as follows:

[0097] Calculate the comprehensive evaluation value based on the first reference weight, the second reference weight, and the third reference weight;

[0098] Compare the comprehensive evaluation value with the threshold range to select the power consumption level;

[0099] Determine the dynamic adjustment strategy based on the power consumption level;

[0100] Specifically, influence coefficients are assigned to the first, second, and third reference weights, and a comprehensive evaluation value is calculated. The expression for calculating the comprehensive evaluation value is as follows: In the formula, This represents the overall evaluation value. Indicates the first reference weight. Indicates the second reference weight. Indicates the third reference weight. This represents the first level of influence. This represents the second level of influence. This represents the third level of influence. , It can be 0.5. It can be 0.3. The threshold can be set to 0.2. The comprehensive evaluation value is compared with the threshold range to select a power consumption level strategy. Three threshold ranges can be set: a first threshold range (greater than or equal to 0.7), a second threshold range (greater than or equal to 0.4 and less than 0.7), and a third threshold range (less than 0.4), corresponding to the first, second, and third power consumption level strategies, respectively. The threshold ranges can be set based on the distribution of historical thresholds, with the range boundaries set at key quantiles. When the comprehensive evaluation value falls within the first threshold range, the first power consumption level strategy is selected, corresponding to increasing the upper limit of sensor acquisition frequency and data upload frequency to the highest level. The upper limit of sensor acquisition frequency is once every 30 seconds, and the upper limit of data upload frequency is... The upper limit of the data transmission frequency is once every 1 minute, which facilitates high-frequency data acquisition and uploading to ensure real-time monitoring. When the comprehensive evaluation value falls into the second threshold range, the second power consumption level strategy is selected, which corresponds to maintaining or appropriately reducing the upper limit of the sensor acquisition frequency and data upload frequency to a medium level. Specifically, the upper limit of the sensor acquisition frequency is once every 5 minutes, and the upper limit of the data upload frequency is once every 10 minutes, which helps to balance energy consumption and data quality. When the comprehensive evaluation value falls into the third threshold range, the third power consumption level strategy is selected, which corresponds to adjusting the upper limit of the sensor acquisition frequency and data upload frequency to a low level. Specifically, the upper limit of the sensor acquisition frequency is once every 15 minutes, and the upper limit of the data upload frequency is once every 1 hour, which helps to maintain basic monitoring.

[0101] Please see Figure 2 A multifunctional integrated online water quality monitoring system, using the aforementioned multifunctional integrated online water quality monitoring method, includes:

[0102] The data acquisition module is used to collect multimodal water quality data through a sensor matrix, wherein the sensor matrix includes multiple sensors and the multimodal water quality data includes multiple types of water quality data.

[0103] The anomaly analysis module is used to perform anomaly analysis on multimodal water quality data and identify abnormal data.

[0104] The correlation analysis module is used to determine whether there are related abnormal data for each abnormal data and to determine the abnormal status of the abnormal data. If there are related abnormal data, it is determined to be a real abnormality and the first response strategy is executed; otherwise, it is determined to be a suspected abnormality and the second response strategy is executed. The first response strategy includes increasing the collection frequency of the sensor corresponding to the abnormal data and determining whether to increase the data upload frequency. The second response strategy includes reviewing the abnormal data and re-determining the abnormal status.

[0105] The weighting analysis module is used to determine the first reference weight based on the type of data anomaly, the second reference weight based on the current device power level, and the third reference weight based on the weather forecast results for a preset future time period.

[0106] The power allocation module is used to determine a power allocation strategy for dynamically adjusting the sensor acquisition frequency and data upload frequency based on the first reference weight, the second reference weight, and the third reference weight.

[0107] In the above, the data acquisition module is responsible for collecting multimodal water quality data through the sensor matrix. After data acquisition, the anomaly analysis module will identify anomalies in the collected data and determine whether there are other data related to the abnormal data. If there is related data, the correlation analysis module will further check whether these data are also abnormal. If the related data is also abnormal, it is determined to be a real anomaly and triggers the first response strategy. The first response strategy will increase the acquisition frequency of the corresponding sensor and dynamically determine whether to increase the data upload frequency. Otherwise, the second response strategy will be activated. The second response strategy will review the suspected anomaly to determine whether the suspected anomaly is a real anomaly or no anomaly. In addition, the weight analysis module assigns a first reference weight, a second reference weight, and a third reference weight according to the anomaly type, the current power of the device, and the future weather forecast. The power allocation module integrates the three reference weights and dynamically optimizes the power allocation strategy for sensor acquisition and data upload.

[0108] Please see Figure 3 A multifunctional integrated online water quality monitoring device, characterized in that: the monitoring device includes:

[0109] At least one processor;

[0110] and a memory communicatively connected to the at least one processor;

[0111] The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the multifunctional integrated online water quality monitoring method according to any one of claims 1 to 8.

[0112] The processor of the aforementioned monitoring device can be a high-performance central processing unit (CPU) or graphics processing unit (GPU), and the memory can include storage devices such as random access memory (RAM), read-only memory (ROM), solid-state drive (SSD), or hard disk drive. In addition, the monitoring device may also include an arithmetic unit, input devices, output devices, and a network interface. The arithmetic unit can be a logic unit used to perform various arithmetic and logical operations to assist the processor in completing complex data processing tasks. Input devices can include keyboards, mice, touch screens, etc., used to receive user input instructions and data. Output devices can include displays, printers, etc., used to display processing results and output reports. The network interface is used to enable network communication between the monitoring device and other systems or devices for data exchange and remote monitoring.

[0113] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0114] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. A multifunctional integrated online water quality monitoring method, characterized in that: include: Multimodal water quality data is collected through a sensor matrix, which includes various sensors and various types of water quality data. Time alignment and normalization were performed on the multimodal water quality data; Establish a time sliding window, calculate the trend slope, and compare it with the historical trend slope benchmark to identify abnormal trends; Historical water quality data is acquired, and a dynamic threshold band is established. If the water quality data exceeds the dynamic threshold band, it is marked as a candidate outlier. Based on the historical behavior pattern of the data points, the isolation degree of the data points in their own parameter time series is analyzed. If the isolation degree is greater than the preset isolation threshold, it is determined to be an outlier. Weighted fusion of abnormal trends and outliers is performed to calculate confidence levels and determine whether water quality data is abnormal. For each abnormal data, determine whether there is any related abnormal data and determine the abnormal status of the abnormal data. If there is related abnormal data, it is determined to be a real abnormality and the first response strategy is executed; otherwise, it is determined to be a suspected abnormality and the second response strategy is executed. The first response strategy includes: increasing the sampling frequency of sensors corresponding to water quality data showing genuine anomalies, and increasing the sampling frequency of sensors corresponding to water quality data that are strongly correlated with such data; determining whether to increase the data upload frequency based on the type of anomaly data—if the anomaly data belongs to a key indicator, then increase the data upload frequency, otherwise not; the second response strategy includes: establishing a temporary time window, increasing the sampling frequency of sensors corresponding to the anomaly data within the temporary time window to obtain high-frequency verification data; performing anomaly analysis again on the high-frequency verification data, and if it is identified as anomaly data, then determining that the anomaly data is a genuine anomaly, otherwise it is considered as no anomaly; Based on the type of data anomaly, a first reference weight is determined; based on the current device power consumption, a second reference weight is determined; and based on the weather forecast results for a future preset time period, a third reference weight is determined. Based on the first, second, and third reference weights, a power consumption allocation strategy for dynamically adjusting the sensor acquisition frequency and data upload frequency is determined. When calculating isolation degree, n data points are taken before and after the candidate outlier to form a local window for calculating isolation degree. The expression for calculating isolation degree is as follows: In the formula, Indicates the degree of isolation. Indicates the candidate outlier value. This represents the median of all data within the local window, excluding the candidate point itself. It represents the average absolute deviation of all data within the local window, excluding the candidate points themselves. The confidence level is calculated using the following expression: In the formula, Indicates the confidence level. The fusion score is calculated by comparing the confidence level with a preset confidence threshold. If the confidence level is greater than the preset confidence threshold, the water quality data is determined to be abnormal; otherwise, the water quality data is determined to be non-abnormal.

2. The multifunctional integrated online water quality monitoring method according to claim 1, characterized in that: The steps for determining a first reference weight based on the type of data anomaly, a second reference weight based on the current device battery level, and a third reference weight based on weather forecasts for a preset future time period are as follows: First reference weight is assigned based on the preset level of the data anomaly type; second reference weight is assigned based on the preset battery level range to identify the battery range corresponding to the current device battery level; and third reference weight is assigned based on the sunshine data obtained from the weather forecast and the sunshine level determined according to a preset sunshine level table.

3. The multifunctional integrated online water quality monitoring method according to claim 1, characterized in that: The power allocation strategy for dynamically adjusting the sensor acquisition frequency and data upload frequency based on the first reference weight, the second reference weight, and the third reference weight is determined in the following steps: Calculate a comprehensive evaluation value based on the first reference weight, the second reference weight, and the third reference weight; compare the comprehensive evaluation value with a threshold range to select a power consumption level; and determine the dynamic adjustment strategy based on the power consumption level.

4. A monitoring system for a multifunctional integrated online water quality monitoring method according to any one of claims 1-3, characterized in that: include: The data acquisition module is used to collect multimodal water quality data through a sensor matrix, wherein the sensor matrix includes multiple sensors and the multimodal water quality data includes multiple types of water quality data. The anomaly analysis module is used to perform anomaly analysis on multimodal water quality data and identify abnormal data. The correlation analysis module is used to determine whether there are related abnormal data for each abnormal data and to determine the abnormal status of the abnormal data. If there are related abnormal data, it is determined to be a real abnormality and the first response strategy is executed; otherwise, it is determined to be a suspected abnormality and the second response strategy is executed. The first response strategy includes increasing the collection frequency of the sensor corresponding to the abnormal data and determining whether to increase the data upload frequency. The second response strategy includes reviewing the abnormal data and re-determining the abnormal status. The weighting analysis module is used to determine the first reference weight based on the type of data anomaly, the second reference weight based on the current device power level, and the third reference weight based on the weather forecast results for a preset future time period. The power allocation module is used to determine a power allocation strategy for dynamically adjusting the sensor acquisition frequency and data upload frequency based on the first reference weight, the second reference weight, and the third reference weight.

5. A multifunctional integrated online water quality monitoring device, characterized in that: The monitoring device includes: At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the multifunctional integrated online water quality monitoring method according to any one of claims 1 to 3.

Images