Statistical and computational model-based ecological environment monitoring data analysis method and system

Abstract

The application relates to the technical field of environmental monitoring, and provides an ecological environment monitoring data analysis method and system based on statistics and a calculation model.The method comprises the following steps: performing trend correction on a value classified as abnormal external data to generate a time series subjected to trend correction; determining a first prediction period and a second prediction period based on the time series subjected to trend correction, inputting extracted abnormal related parameter values, meteorological condition data and data of at least one other preset parameter into a calculation model to obtain a risk value, and determining a risk level according to the risk value; and outputting an alarm signal when the risk level meets a preset alarm condition.The application performs trend correction on abnormal external data only, thereby distinguishing invalid data caused by equipment failure from real environmental changes, performing trend correction on the latter, retaining a real track of environmental state changes, and overcoming the defects of the prior art that cannot distinguish abnormal causes and uniformly processes.

Description

Technical Field

[0001] This invention belongs to the field of ecological environment monitoring technology, specifically relating to ecological environment monitoring data analysis methods and systems based on statistical and computational models. Background Technology

[0002] By deploying various sensors in key urban areas to continuously monitor environmental elements such as water, air, and noise, massive, multi-source, and high-dimensional real-time monitoring data can be obtained. Then, by performing efficient and specific analysis on this data, effective information can be extracted, and insights into environmental change trends and early warnings of potential pollution events can be obtained.

[0003] However, existing monitoring data analysis methods rely on preset statistical models or fixed thresholds to judge data anomalies in practical applications. They are quite sensitive to data noise and sudden outliers. A single extreme data point may cause the statistical results to be distorted, resulting in a bias in the judgment of the entire abnormal period and a high false alarm rate and false negative rate in the identification of abnormal events.

[0004] Furthermore, most mainstream data analysis architectures currently adopt a centralized processing model, which involves using terminal sensors to collect data and then transmitting the collected data to the cloud or data center for unified calculation. In this process, network transmission delays are inevitably introduced, causing the analysis results to lag behind the actual situation. This lag also prevents the system from performing real-time verification and fusion analysis of related information such as meteorological parameters, which to some extent weakens the ability to trace faults and respond immediately.

[0005] To address the aforementioned problems, this invention provides a method and system for analyzing ecological and environmental monitoring data based on statistical and computational models. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for analyzing ecological and environmental monitoring data based on statistical and computational models, in order to solve the problem that existing ecological and environmental monitoring data analysis methods are unable to take into account the impact of abnormal monitoring data during the screening of abnormal time points, resulting in poor accuracy of the screening results.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Methods for analyzing ecological and environmental monitoring data based on statistical and computational models include: Acquire the data to be evaluated, which includes a time series of at least one parameter to be evaluated; in response to acquiring the data to be evaluated, perform anomaly processing to identify and correct outlier values ​​in the time series, thereby generating a trend-corrected time series; perform a risk assessment based on the trend-corrected time series to determine the risk level; and output an alarm signal when the risk level meets a preset alarm condition. The risk assessment based on the trend-corrected time series includes: determining the first and second forecast periods based on the trend-corrected time series; extracting the values ​​of the parameters to be assessed within the first and second forecast periods as anomaly-related parameter values; inputting the anomaly-related parameter values, meteorological condition data, and data of at least one other preset parameter into the calculation model to obtain a risk value, and determining the risk level based on the risk value.

[0008] Preferably, in response to acquiring the data to be evaluated, performing anomaly processing includes: determining the normal value range based on a statistical model; and identifying values ​​in the time series that are outside the normal value range as abnormal values.

[0009] Preferably, the method further includes: selecting a normal time interval in the time series where all values ​​are within the normal value range to determine the average prediction difference of the parameter to be evaluated; and updating the normal value range based on the average prediction difference and a preset deviation threshold.

[0010] Preferably, in the time series, a normal time interval is selected where all values ​​are within the normal range, in order to determine the average prediction difference of the parameter to be evaluated, including: Extract the value of the parameter to be evaluated at the beginning of the normal time interval as the baseline value; extract multiple sampling points from the normal time interval; for each sampling point, calculate the degree of deviation between its value and the baseline value as the prediction difference; calculate the average of the prediction differences of multiple sampling points to obtain the average prediction difference.

[0011] Preferably, in response to acquiring the data to be evaluated, the anomaly handling further includes: determining the cause type of the abnormal value and classifying the abnormal value as abnormal collected data or abnormal external data; wherein, performing trend correction is performed on the value classified as abnormal external data.

[0012] Preferably, determining the cause type of abnormal values ​​includes: Based on the time point when the abnormal value appears in the time series, the rate of change of the value at each time point in the inspection interval with a second preset duration before that time point is calculated; when there is at least one time point in the inspection interval whose rate of change of value is greater than the preset rate of change threshold, the abnormal value is classified as abnormal collected data, otherwise it is classified as abnormal external data.

[0013] Preferably, performing trend correction includes: Identify whether there are anomalous data clusters containing anomalous external data in the time series; if anomalous data clusters exist, perform trend correction on the anomalous external data within the time period between the correction start point and correction end point determined by the anomalous data clusters.

[0014] This invention also provides an ecological environment monitoring data analysis system based on statistical and computational models, which applies the aforementioned ecological environment monitoring data analysis method based on statistical and computational models, including: The data acquisition module is used to acquire the data to be evaluated, which includes a time series of at least one parameter to be evaluated. The data analysis module is used to respond to the data to be evaluated acquired by the data acquisition module, perform anomaly processing to identify and correct outlier values ​​in the time series, thereby generating a trend-corrected time series, and perform risk assessment based on the trend-corrected time series to determine the risk level. The risk alarm module is used to output an alarm signal when the risk level determined by the data analysis module meets the preset alarm conditions. The normal value range update module is used to update the normal value range for the data analysis module to perform anomaly handling based on the normal time range in the time series.

[0015] Preferably, it further includes: The average prediction difference determination module is used to select a normal time interval in the time series where all values ​​are within the normal value range, in order to determine the average prediction difference of the parameter to be evaluated. The normal value range update module is used to update the normal value range based on the average prediction difference determined by the average prediction difference determination module and the preset deviation threshold.

[0016] Preferably, when performing anomaly handling, the data analysis module is also used to: determine the cause type of the abnormal value and classify the abnormal value into abnormal collected data or abnormal external data; wherein, the data analysis module performs trend correction on the values ​​classified as abnormal external data.

[0017] Beneficial effects 1. This invention determines the cause type of abnormal values ​​based on the rate of change of values, classifies them into abnormal collected data or abnormal external data, and performs trend correction only on abnormal external data, thereby distinguishing invalid data caused by equipment failure from real environmental changes. By performing trend correction on the latter, the true trajectory of environmental state changes is preserved, overcoming the shortcomings of existing technologies that cannot distinguish the cause of abnormality and process them uniformly.

[0018] 2. This invention analyzes the normal time interval of stable values, calculates the average prediction difference, and dynamically updates the normal value interval for identifying abnormal values ​​based on the average prediction difference. It can be adjusted according to the seasonal or long-term drift of environmental background values, avoiding the problem of increased misjudgment rate caused by using a fixed threshold after environmental changes.

[0019] 3. This invention inputs the values ​​of abnormal related parameters, meteorological conditions, and at least one other preset parameter data of the second and first prediction periods into the calculation model, making the output risk level and alarm signal more forward-looking and comprehensive, and providing more accurate and reliable early warning of ecological and environmental risks. Attached Figure Description

[0020] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system module diagram of the present invention. Detailed Implementation

[0021] 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.

[0022] Example 1 See Figure 1 This embodiment provides a method for analyzing ecological environment monitoring data based on statistical and computational models. The method includes the following steps: S1. Obtain the data to be evaluated; The data to be evaluated includes a time series of at least one parameter to be evaluated, such as dissolved oxygen concentration, pH value, or turbidity parameter from a water quality monitoring station.

[0023] A time series consists of a series of sampling points arranged in chronological order, containing both the time point and its corresponding value.

[0024] S2. Anomaly identification and classification; By performing statistical analysis on historical data, a normal value range is determined for the time series; the upper and lower limits of this normal value range represent the reasonable fluctuation range of the parameter to be evaluated under normal operating conditions.

[0025] The preferred statistical analysis method is to calculate its mean and standard deviation over a long period of time to construct a Gaussian distribution, or to determine its fluctuation boundary at a specific confidence level using the quantile method.

[0026] Values ​​outside the normal range in the time series will be identified as outliers. After identifying outliers, in order to distinguish their causes, the cause type of the outliers will be further determined, thereby separating the abnormal collected data from abnormal external data; Abnormal data refers to anomalous values ​​that do not reflect the true environmental conditions and are caused by internal problems in the data acquisition process, such as sensor malfunctions, communication interruptions, or system errors. Anomalous external data refers to anomalous values ​​that reflect real changes in the environmental state and are caused by real external environmental events such as pollution source emissions or sudden climate changes.

[0027] The specific process for determining the cause type of abnormal values ​​is as follows: Based on the time point when the abnormal value appears in the time series, a time period with a second preset duration before that time point is selected as the interval to be checked, preferably 5 minutes; calculate the rate of change of the value at each time point in the interval to be checked, where the rate of change of the value is specifically the difference between the values ​​of two adjacent sampling points divided by the time interval. Determine whether there is at least one time point within the interval to be checked where the rate of change of the value is greater than a preset rate of change threshold. This rate of change threshold is preset based on the historical stability of the parameter and the physical response characteristics of the sensor. If such a time point exists, it indicates that the data has experienced a sudden change or jump that does not conform to the laws of physics. Specifically, the sensor reading jumps from the low level of the normal range to the high level limit in an instant within a single sampling interval. This rate of change exceeds the limit reached by the environmental physical or chemical process. This situation is usually a typical feature of equipment failure. Therefore, this abnormal value is classified as abnormal data acquisition. Conversely, if all numerical change rates are within the threshold, the anomaly is considered to be the result of environmental factors changing in a physically possible way, either gradually or rapidly, and is classified as anomalous external data.

[0028] S3, Abnormal Data Handling; Values ​​classified as abnormal data collection are marked as invalid data and removed from subsequent risk assessment calculations, while an equipment maintenance work order is generated.

[0029] For numerical values ​​classified as anomalous external data, which are of significant analytical value because they reflect real environmental changes, trend correction is performed to generate trend-corrected time series. The trigger condition for trend correction is the detection of anomalous data clusters in the time series; specifically: When the duration is less than a third preset duration (preferably 30 minutes), and the number of abnormal external data points exceeds a preset threshold (e.g., more than 5 points), an abnormal data cluster is identified. If an abnormal data cluster is detected, trend correction is performed.

[0030] The trend correction process takes the time point when the first abnormal external data appears in the abnormal data cluster as the correction start point and the end point of the fourth preset time period after the correction start point as the correction end point. During the time period between the correction start point and the correction end point, a preset numerical smoothing method is used to process the abnormal external data in this period, thereby eliminating numerical noise caused by sudden changes that may interfere with subsequent analysis, forming a smooth curve that better reflects the inherent development trend of the environmental event, and providing mathematically easier-to-process input for subsequent risk assessment.

[0031] The preferred numerical smoothing method is linear interpolation, polynomial fitting, or exponential smoothing.

[0032] S4. Determining the risk assessment period and extracting parameters; After generating the trend-corrected time series, the first and second forecast periods are determined based on the trend-corrected time series.

[0033] The first prediction period is defined as the period from the current moment to a future period with a fifth preset duration, preferably 30 minutes in the future, so as to use existing data trends to predict the evolution of risks in the short term. The second forecast period is defined as the period from the starting point of the aforementioned trend correction process to the current moment, used to retrospectively analyze the entire process and cumulative impact of the anomalous event from its occurrence to the present.

[0034] Once the time period is determined, the system automatically extracts the values ​​of the parameters to be evaluated within the first and second prediction periods. The predicted value sequence, calculated by extrapolation or trend extension based on the latest trend of the time series after trend correction, is used as the value for the first prediction period. The value for the second prediction period consists of the actual measured values ​​in the historical records and the trend-corrected values. These sets of values ​​are unified as the values ​​of the anomaly-related parameters for subsequent risk assessment calculations.

[0035] Furthermore, the anomaly-related parameter values ​​are a comprehensive dataset used for risk assessment calculations, which includes historical and corrected values ​​for the second forecast period, as well as future forecast values ​​for the first forecast period.

[0036] S5. Risk Assessment and Alarm; While processing the values ​​of abnormal parameters, the meteorological conditions data at the current moment and the data of at least one other preset parameter are acquired and input. The meteorological conditions data include wind speed, wind direction, temperature and air pressure. The other preset parameters are other environmental monitoring parameters that are strongly correlated with the parameter to be evaluated. For example, when evaluating dissolved oxygen in water, chemical oxygen demand and turbidity can be introduced as other preset parameters.

[0037] These multi-dimensional data are input into the preset risk assessment processing unit, and the input parameter values ​​are weighted and comprehensively evaluated according to the preset association rules and calculation logic.

[0038] Among them, the association rules and calculation logic are pre-built based on a large amount of historical environmental event data and expert knowledge base, which can quantify the correlation strength between different parameter combinations and specific risk events, thereby outputting a comprehensive risk value.

[0039] According to the preset risk classification rules, for example, the comprehensive risk value range of 0 to 100 is divided into four levels: low, medium, high, and extremely high. The calculated comprehensive risk value is determined as the corresponding risk level. It is judged whether the risk level meets the preset alarm conditions, such as the risk level being high or extremely high. If the conditions are met, an alarm signal is immediately output, such as through an audible and visual alarm, sending a text message, or pushing alarm information to the central control room.

[0040] Furthermore, the risk assessment processing unit can be viewed as a computational model used to quantify environmental risks by integrating multidimensional data, and its specific definition is as follows:

[0041] In the formula, The comprehensive risk value is a scalar value representing the final quantified environmental risk level, calculated by weighting input data from multiple dimensions. This represents the weight vector, which consists of weight coefficients corresponding to the three types of input data. The values ​​of each element are pre-calibrated by historical data and expert knowledge base, reflecting the contribution of different parameters to the final risk. This represents a transformation function, which means transforming the original input vector with different physical units and dimensions, such as... Normalization functions mapped to a uniform, comparable scale, such as minimum and maximum normalization functions. ; The input vector represents a vector containing the values ​​of the anomaly-related parameters, meteorological conditions, and other related parameters. The input value represents the normalization function, which means that it represents the value of a single element in the input vector during the normalization process. These represent the normalization boundary values, signifying the historical minimum and maximum values ​​of the parameters used in the normalization calculation. They are used to scale the original data to the [0,1] interval. The superscript indicates the normalization boundary value. This represents the transpose operation of a vector, with subscripts... Represents the first element in a vector or time series. Each element or data point.

[0042] S6. Dynamic update of normal value range; To adapt to seasonal environmental changes, long-term trend changes, or performance drift of the sensor itself, a feedback mechanism is also included to dynamically update the normal value range. This is achieved by determining the average prediction difference of the parameter to be evaluated. The specific steps are as follows: In the time series, select several normal time intervals in which all values ​​are within the current normal value range and the duration is greater than the first preset duration; For each selected normal time interval, the value of the parameter to be evaluated at its starting point is extracted and defined as the baseline value of the interval, which serves as a reference point for calculating the deviation of subsequent data points within the interval. Multiple sampling points are extracted uniformly or randomly from the normal time interval. For each sampling point, the degree of deviation between its value and the benchmark value is calculated, and the degree of deviation is defined as the prediction difference, which is used to reflect the natural fluctuation or drift of the value over time under a stable state without anomalies. The average prediction difference of all sampling points is calculated to obtain the average prediction difference of the parameter to be evaluated. This value represents the average drift of the parameter to be evaluated in long-term operation. After obtaining the average prediction difference, the bias interval is determined based on the average prediction difference and the preset deviation threshold. This bias interval represents the reasonable range of movement of the parameter baseline in the long-term scale. The preset deviation threshold is preferably 1.5 times the average prediction difference. Based on the bias interval, the starting and ending points of the current normal value interval are shifted or scaled accordingly to obtain the updated normal value interval.

[0043] By periodically or through event-triggered updates, the definition of normal value ranges is adaptively adjusted, thereby maintaining the accuracy and sensitivity of anomaly identification over long-term operation.

[0044] Example 2 See Figure 2 This embodiment provides an ecological environment monitoring data analysis system based on statistical and computational models: The data acquisition module connects to one or more data sources via a standard data protocol to acquire data to be evaluated at a preset time frequency. The data to be evaluated is a time series of one or more parameters to be evaluated. Preferably, the parameters to be evaluated are PM2.5 concentration in the air, chemical oxygen demand in water, or heavy metal content in soil. The corresponding time series is a sequence of monitoring values ​​at consecutive time points. After acquisition, the module formats the data for subsequent analysis by the module.

[0045] The data analysis module is configured to perform exception handling and risk assessment. To achieve complex logical functions, this module can be further subdivided internally into multiple cooperating sub-functions.

[0046] In response to the acquired data to be evaluated, this module performs anomaly handling. In the specific execution process, it determines the initial normal value range based on statistical models such as the 3-sigma criterion or box plot method; it iterates through each value in the time series and identifies values ​​outside the normal value range as abnormal values.

[0047] For each identified abnormal value, determine its cause type, specifically: Using the time point when the abnormal value appears in the time series as a benchmark, backtrack for a second preset time period to form an inspection interval; within the inspection interval, calculate the rate of change of the value at each time point; When it is found that the rate of change of the value at at least one time point in the range to be checked is greater than the preset rate of change threshold, such as when the value changes by an order of magnitude within a single sampling period, the module classifies this abnormal value as abnormal data acquisition, which represents sensor failure, signal interference or data transmission error. Conversely, if the numerical value changes gradually but continues to exceed the normal range, it is classified as abnormal external data, which usually reflects real environmental pollution events or drastic changes in the natural environment.

[0048] Trend correction is performed based on the classification results. It's important to note that trend correction is only applied to values ​​classified as outlier external data. Specifically: In the time series, identify whether there is an anomalous data cluster consisting of one or more consecutive anomalous external data. If so, determine the correction start point and correction end point of the anomalous data cluster. Within the time period determined by the start point and end point, use time series interpolation, fitting or other smoothing algorithms to correct the anomalous external data, so as to generate a trend-corrected time series that can reflect the true trend of change and eliminate abrupt fluctuations. For data points classified as anomalous data collection, we choose to mark them, delete them, or replace them with nearby normal values, but do not perform trend correction.

[0049] Risk assessment is performed based on trend-corrected time series data. Specifically: Based on the preset forecast target, the first forecast period and the second forecast period are determined; the values ​​of the parameters to be evaluated in these two forecast periods are extracted from the trend-corrected time series as the values ​​of the anomaly-related parameters. At the same time, the module obtains the meteorological condition data of the same period from the external data source and combines it with the data of at least one other preset parameter. All of this data, including values ​​of anomaly-related parameters, meteorological conditions, and other preset parameters, is fed into a pre-trained computational model to output a quantified, comprehensive risk value, which is then mapped to a predefined risk level.

[0050] The average prediction difference determination module is configured to determine the average prediction difference, which is used to achieve adaptive adjustment of the anomaly judgment benchmark. In practical applications, within a time series, a normal time interval is selected where all values ​​fall within the current normal value range; the value of the parameter to be evaluated at the starting point of this normal time interval is extracted as the baseline value. Multiple sampling points are extracted from this normal time interval, and for each sampling point, the degree of deviation between its value and the benchmark value is calculated. This degree of deviation is the prediction difference. The average of the prediction differences for all these sampling points is calculated to obtain the average prediction difference that reflects the natural drift or seasonal trend of the current parameter.

[0051] The normal value range update module is configured to perform the step of updating the normal value range; specifically: Receive the average prediction difference calculated by the average prediction difference determination module, and dynamically update the normal value range used by the data analysis module based on the average prediction difference and the preset deviation threshold. Preferably, the center value of the normal value range is adjusted to the sum of the current benchmark value and the average prediction difference, or the width of the range is adjusted according to the fluctuation range of the prediction difference, so that the system's anomaly identification capability can adapt to long-term changes in environmental parameters and avoid misjudgment or omission due to benchmark rigidity.

[0052] The risk alarm module is configured to perform the step of outputting an alarm signal; specifically: It continuously receives risk levels determined by the data analysis module, which has preset alarm conditions, such as "risk level is level three or above"; When the received risk level meets this condition, the module is immediately triggered and outputs an alarm signal. The alarm signal can take various forms, including: Highlighted alerts can pop up on the large screen in the monitoring center, push notifications or text messages can be sent to the mobile devices of maintenance personnel, detailed risk analysis reports can be automatically generated and sent to a designated email address, or the emergency response plan system can be triggered through API calls.

[0053] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for analyzing ecological and environmental monitoring data based on statistical and computational models, characterized in that, include: Acquire the data to be evaluated, which includes a time series of at least one parameter to be evaluated; In response to acquiring the data to be evaluated, anomaly handling is performed to identify and correct outlier values ​​in the time series, thereby generating a trend-corrected time series; based on the trend-corrected time series, a risk assessment is performed to determine the risk level; When the risk level meets the preset alarm conditions, an alarm signal is output; The risk assessment based on the trend-corrected time series includes: determining the first and second forecast periods based on the trend-corrected time series; and extracting the values ​​of the parameters to be assessed within the first and second forecast periods as anomaly-related parameter values. The values ​​of abnormal parameters, meteorological conditions, and at least one other preset parameter are input into the calculation model to obtain the risk value, and the risk level is determined based on the risk value.

2. The method for analyzing ecological environment monitoring data based on statistical and computational models according to claim 1, characterized in that, In response to acquiring the data to be evaluated, anomaly handling is performed, including: determining the normal value range based on a statistical model; and identifying values ​​in the time series that are outside the normal value range as anomalous values.

3. The method for analyzing ecological environment monitoring data based on statistical and computational models according to claim 2, characterized in that, The method further includes: selecting a normal time interval in the time series where all values ​​are within the normal value range to determine the average prediction difference of the parameter to be evaluated; and updating the normal value range based on the average prediction difference and a preset deviation threshold.

4. The method for analyzing ecological environment monitoring data based on statistical and computational models according to claim 3, characterized in that, In the time series, a normal time interval is selected where all values ​​are within the normal range to determine the average prediction difference of the parameter to be evaluated, including: Extract the value of the parameter to be evaluated at the beginning of the normal time interval as the baseline value; extract multiple sampling points from the normal time interval; for each sampling point, calculate the degree of deviation between its value and the baseline value as the prediction difference; calculate the average of the prediction differences of multiple sampling points to obtain the average prediction difference.

5. The method for analyzing ecological environment monitoring data based on statistical and computational models according to claim 2, characterized in that, In response to acquiring the data to be evaluated, the anomaly handling also includes: determining the cause type of the abnormal values ​​and classifying the abnormal values ​​as abnormal collected data or abnormal external data; among which, trend correction is performed on the values ​​classified as abnormal external data.

6. The method for analyzing ecological environment monitoring data based on statistical and computational models according to claim 5, characterized in that, The causes of outlier values ​​can be identified by: Based on the time point when the abnormal value appears in the time series, the rate of change of the value at each time point in the inspection interval with a second preset duration before that time point is calculated; when there is at least one time point in the inspection interval whose rate of change of value is greater than the preset rate of change threshold, the abnormal value is classified as abnormal collected data, otherwise it is classified as abnormal external data.

7. The method for analyzing ecological environment monitoring data based on statistical and computational models according to claim 5, characterized in that, The implementation of trend corrections includes: Identify whether there are anomalous data clusters containing anomalous external data in the time series; if anomalous data clusters exist, perform trend correction on the anomalous external data within the time period between the correction start point and correction end point determined by the anomalous data clusters.

8. An ecological environment monitoring data analysis system based on statistical and computational models, employing the ecological environment monitoring data analysis method based on statistical and computational models as described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to acquire the data to be evaluated, which includes a time series of at least one parameter to be evaluated. The data analysis module is used to respond to the data to be evaluated acquired by the data acquisition module, perform anomaly processing to identify and correct outlier values ​​in the time series, thereby generating a trend-corrected time series, and perform risk assessment based on the trend-corrected time series to determine the risk level. The risk alarm module is used to output an alarm signal when the risk level determined by the data analysis module meets the preset alarm conditions. The normal value range update module is used to update the normal value range for the data analysis module to perform anomaly handling based on the normal time range in the time series.

9. The ecological environment monitoring data analysis system based on statistical and computational models according to claim 8, characterized in that, Also includes: The average prediction difference determination module is used to select a normal time interval in the time series where all values ​​are within the normal value range, in order to determine the average prediction difference of the parameter to be evaluated; the normal value interval update module is used to update the normal value interval based on the average prediction difference determined by the average prediction difference determination module and a preset deviation threshold.

10. The ecological environment monitoring data analysis system based on statistical and computational models according to claim 8, characterized in that, When performing anomaly handling, the data analysis module is also used to: determine the cause type of the abnormal values ​​and classify the abnormal values ​​into abnormal collected data or abnormal external data; among them, the data analysis module performs trend correction on the values ​​classified as abnormal external data.

Images