An intelligent monitoring method for water pollution damage

Abstract

The present invention belongs to the field of data processing technology, and specifically relates to an intelligent monitoring method for water pollution damage, the method comprising dividing a sewage discharge area into multiple sub-areas, setting a number of intelligent sensors in each sub-area, collecting sewage component data of the sub-areas, integrating data based on the collection timestamp to obtain combined data, and fusing the combined data belonging to different sub-areas to generate fused data; after a remote server receives the fused data, the pollution degree of the fused data is judged based on the score and weight value; the pollution degree of the fused data is judged based on the score and weight value; based on the predicted sewage component data and abnormal conditions, an artificial intelligence model is used to generate corresponding countermeasures, the implementation time of the countermeasures is estimated, and the implementation effect of the countermeasures is evaluated based on the newly collected fused data. The present invention can quickly and accurately judge the pollution degree of water pollution.

Description

Technical Field

[0001] The invention belongs to the technical field of data processing, and in particular relates to an intelligent monitoring method for water pollution damage. Background Art

[0002] Traditional water pollution monitoring methods often rely on manual sampling and laboratory analysis, which is not only time-consuming and labor-intensive, but also difficult to achieve real-time monitoring and rapid response. Therefore, it is of great practical significance to develop a method that can monitor the water pollution status in real time and quickly assess the damage. A similar prior art Chinese patent application with publication number CN118037057A provides a municipal garden water pollution monitoring system based on big data analysis, including a garden information data acquisition module, a risk level classification module, a linear regression model module and an alarm module; the garden information data acquisition module obtains multiple environmental data outside the garden water body and obtains the corresponding garden water pollution data; the risk level classification module divides the risk level according to multiple environmental data and garden water pollution data; the linear regression model module trains a linear regression model based on the data collected multiple times; and uses the linear regression model to predict the risk level of garden water pollution; when the risk level is higher than the threshold, the alarm module issues an alarm. Similar prior art also has a Chinese patent application with publication number CN116562185A, which publicly provides a water quality simulation method, system and water pollution monitoring and analysis device, and the method includes: (a) coupling strategy of one-dimensional and two-dimensional water quality models: a one-dimensional and two-dimensional coupled water quality model is obtained by coupling a one-dimensional water quality model with a two-dimensional water quality model, and the interface of the coupling calculation of the one-dimensional and two-dimensional models is processed by a distribution function based on the microscopic level; the coupling interface is regarded as an internal node in the one-dimensional calculation domain; first, the unknown distribution function of the node is calculated, and then its macroscopic concentration is calculated by the distribution function, and finally the unknown distribution function of the two-dimensional boundary at that location is solved; (b) GPU parallel implementation of one-dimensional and two-dimensional coupled water quality model; (c) parallel model optimization strategy: the calculation process stores the calculation data of the two-dimensional model on a shared memory, and replaces the access to the global memory with the access to the shared memory. However, when monitoring water pollution, the above two application documents do not consider the problem of how to improve the monitoring efficiency and monitoring accuracy. Therefore, the present invention provides an intelligent monitoring method for water pollution damage. Summary of the invention

[0003] The present invention collects sewage composition data through intelligent sensors, integrates and fuses the collected sewage composition data to generate fused data, accurately and quickly judges the pollution degree of the sewage composition data based on the fused data, helps relevant managers to quickly locate the pollution source, and efficiently trains the prediction model based on the historical fused data. The prediction model can accurately predict the sewage composition data and abnormal conditions, and also uses the artificial intelligence model based on the predicted sewage composition data and abnormal conditions to generate corresponding response measures, helping relevant personnel to better monitor and manage water pollution.

[0004] In order to achieve the above-mentioned purpose of the invention, the present invention provides an intelligent monitoring method for water pollution damage as described below, which is implemented by executing the following steps:

[0005] Divide the sewage discharge area into multiple sub-areas, set a number of smart sensors in each sub-area, set different collection intervals for the smart sensors based on the characteristics of the sewage composition data collected by the smart sensors, collect the sewage composition data of the sub-areas, and obtain the corresponding collection timestamp and device ID, integrate the sewage composition data corresponding to any sub-area based on the collection timestamp to generate combined data, also fuse all the combined data to generate fused data, and transmit the fused data to a remote server through a network;

[0006] After receiving the fused data, the remote server saves the collected fused data in a memory module, obtains multiple measurement parameters in the fused data, pre-sets multiple different pollution levels, sets a corresponding score for each measurement parameter for each pollution level, and also sets a corresponding weight value for each measurement parameter, and judges the pollution level of the fused data based on the score and the weight value. If the pollution level is higher than the preset pollution level, it is judged that the sewage discharge does not meet the standard, matches the sewage composition data with the known pollution source information to locate the pollution source, and sends the pollution source information to relevant management personnel at the same time, and also marks the fused data that does not meet the standard as abnormal;

[0007] The fused data collected historically and the corresponding abnormal marks are used as learning data to train a prediction model, and the prediction model is used to predict and generate future sewage composition data based on the fused data. When it is predicted that the future sewage composition data is not within a preset normal range, the corresponding abnormal situation is sent to relevant management personnel;

[0008] Based on the predicted sewage composition data and abnormal conditions, an artificial intelligence model is used to generate corresponding countermeasures, and the implementation time of the countermeasures is estimated. The countermeasures are implemented after the implementation time is reached. After that, the sewage discharge area is continuously monitored and the fused data is continued to be collected. The implementation effect of the countermeasures is evaluated based on the newly collected fused data, and the artificial intelligence model is adjusted according to the actual situation.

[0009] As a preferred technical solution of the present invention, data integration based on the collection timestamp to obtain combined data includes the following steps:

[0010] A collection cycle is preset, and after the length of the collection cycle has passed since the start time point of the collection, the timestamp range of each collection cycle is calculated based on the collection cycle and the start time point, and all the sewage component data and the corresponding device ID in the same collection cycle are obtained, and the first value is calculated for all the sewage component data with the same device ID. If the first value is greater than the preset first threshold, the maximum value of all the sewage component data is obtained as the representative value of the device ID in the current collection cycle. If the first value is less than or equal to the first threshold, the second value of all the sewage component data is calculated, and the second value is used as the representative value corresponding to the smart sensor corresponding to the device ID in the current collection cycle. The representative values ​​corresponding to all device IDs in the sub-area are obtained, and the representative values, the collection cycle and the sub-area are combined in a preset format to generate the combined data.

[0011] As a preferred technical solution of the present invention, the combined data belonging to different sub-areas are fused to generate fused data, including the following steps:

[0012] The accuracy of the sewage composition data collected in each sub-region is obtained, and based on the accuracy and the type of intelligent sensor in each sub-region, different regional weights are assigned to different sub-regions, and the sewage composition data of each sub-region with the same collection period is obtained, and the sewage composition data of different sub-regions and the corresponding regional weights are multiplied and then added to generate comprehensive sewage composition data, and all the comprehensive sewage composition data and the collection period are combined to generate the fused data.

[0013] As a preferred technical solution of the present invention, setting a corresponding weight value for each of the measurement parameters includes the following steps:

[0014] Obtain the first evaluation value, the second evaluation value and the third evaluation value made by relevant professionals for each of the measurement parameters corresponding to the other measurement parameters, obtain the first sum of all the second evaluation values ​​of each measurement parameter relative to the other measurement parameters, divide the first evaluation value, the second evaluation value and the third evaluation value corresponding to each measurement parameter by the corresponding first sum to obtain the first value, the second value and the third value, calculate the first average value of the first value, the second value and the third value, obtain multiple first average values ​​corresponding to the measurement parameter, calculate the second average value of the multiple first average values, obtain the second average values ​​corresponding to other measurement parameters, obtain the sum of all second average values ​​as the second sum, and divide the second average value corresponding to the measurement parameter by the second sum as the weight value corresponding to the measurement parameter.

[0015] As a preferred technical solution of the present invention, judging the pollution degree of the fused data based on the score and the weight value comprises the following steps:

[0016] The weight value of each measurement parameter is obtained, and the score corresponding to each measurement parameter is set for each pollution degree, and the pollution degree result is calculated using a first formula, wherein the first formula is: , ,in, are the weight values ​​corresponding to each of the measurement parameters, are the scores corresponding to the i-th measurement parameter for n different pollution levels, m is the number of measurement parameters, n is the number of pollution levels, and the first formula is used to obtain the m comprehensive values ​​corresponding to each measurement parameter for n different pollution levels. , obtaining the pollution degree corresponding to the maximum value of the m comprehensive values ​​as the pollution degree of the fused data.

[0017] As a preferred technical solution of the present invention, the fusion data collected historically and the corresponding abnormal labels are used as learning data to train a prediction model, including the following steps:

[0018] An initial prediction model is created, and initial parameter values ​​are set for the initial prediction model, where the initial parameter values ​​are obtained based on historical experience data, historical fusion data collected historically is obtained from a memory module, and several data are extracted from the historical fusion data as a training data set, and the initial prediction model is trained using the training data set to generate a first prediction model, and the initial parameter values ​​are optimized during the training process, and several data are selected from the historical fusion data as test data, and the first prediction model is tested using the test data to obtain the accuracy of the first prediction model, and if the accuracy is less than a preset second threshold, the training data set is updated to generate a new training data set, and the first prediction model is trained using the new training data set, and the initial parameter values ​​are further optimized during the training process to generate a new prediction model to obtain the corresponding new accuracy, and if the accuracy is still less than the second threshold, the training data set is continued to be updated to train the first prediction model and optimize the parameter values ​​until the accuracy is greater than or equal to the second threshold.

[0019] As a preferred technical solution of the present invention, extracting a number of data from the historical fusion data as a training data set includes the following steps:

[0020] Acquire the historical fusion data, acquire all the measurement parameters and the corresponding weight values ​​for each piece of the historical fusion data, sort all the measurement parameters by number from large to small according to the corresponding weight values ​​to generate a measurement parameter sequence, select several measurement parameters with forward numbers from the measurement parameter sequence as first measurement parameters, combine all the first measurement parameters and the corresponding collection cycles to generate a first fusion data, select the same first measurement parameters and the corresponding collection cycle combination from the remaining historical fusion data to generate the corresponding first fusion data, and combine all the first fusion data to generate the training data set.

[0021] As a preferred technical solution of the present invention, updating the training data set to generate a new training data set includes the following steps:

[0022] Selecting a number of metric parameters with forward numbers from the remaining metric parameter sequences that have not been selected as second metric parameters, and adding the second metric parameters to the training data set to generate a new training data set.

[0023] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0024] In the present invention, firstly, the sewage discharge area is divided into multiple sub-areas, and multiple intelligent sensors are set for each sub-area to collect sewage composition data, so that the collected sewage composition data is more accurate, and the collected sewage composition data is integrated and fused to generate fused data, so that the subsequent analysis of the sewage composition data is more accurate; a reasonable weight is set for each measurement parameter of the fused data, and the pollution degree of the sewage composition data corresponding to the fused data is obtained based on the weight value and the score set for each pollution degree for the measurement parameter, and a more scientific basis is provided for the final judgment of the pollution degree by quantitatively analyzing complex problems; based on the historically collected fused data and abnormal marks as learning data to train the prediction model, first use part of the data as the training data set to train the initial prediction model to generate a first prediction model, and update the training data set to continue training the first prediction model when the first prediction model does not meet the predetermined conditions until the trained prediction model meets the predetermined conditions, while improving the training efficiency of the prediction model and ensuring the prediction accuracy of the prediction model; finally, based on the predicted sewage composition data and abnormal conditions, an artificial intelligence model is used to generate corresponding countermeasures and evaluate the implementation effect of the countermeasures to ensure the effectiveness and sustainability of water pollution management. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 The present invention is a flowchart of the steps of an intelligent monitoring method for water pollution damage. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0027] It is understood that the terms "first", "second", etc. used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, without departing from the scope of this application, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script.

[0028] The present invention provides Figure 1 The intelligent monitoring method for water pollution damage shown is implemented by executing the following steps:

[0029] Step S1, dividing the sewage discharge area into multiple sub-areas, setting a number of smart sensors in each sub-area, setting different collection intervals for the smart sensors based on the characteristics of the sewage composition data collected by the smart sensors, collecting the sewage composition data of the sub-areas, and obtaining the corresponding collection timestamp and device ID, integrating the sewage composition data corresponding to any sub-area based on the collection timestamp to generate combined data, and also fusing all the combined data to generate fused data, and transmitting the fused data to the remote server through the network.

[0030] Specifically, in order to accurately monitor water pollution damage, the sewage discharge area is divided into multiple sub-areas, and multiple smart sensors are set in the sub-areas to collect different sewage component data, such as water temperature, water turbidity, dissolved oxygen, suspended solids, nitrate and ammonia nitrogen content. Since parameters such as suspended solids or nitrates may not change quickly, while dissolved oxygen, water temperature, etc. may change with changes in weather and other short-term environmental factors, more frequent monitoring may be required to collect sewage component data from each sub-area, and at the same time obtain the corresponding collection timestamp and the device ID used to collect the data. In order to make a more accurate judgment on the sewage component data, the sewage component data collected by the smart sensors belonging to the same sub-area are integrated based on the collection timestamp to generate combined data. The specific integration method will be explained in detail later. After that, the combined data belonging to different sub-areas are fused to generate fused data, and the fused data is sent to a remote server for subsequent processing. By integrating and fusion processing the collected data, it is convenient to make a more accurate judgment on the pollution degree in the future, and it can also provide data support for the training of subsequent prediction models.

[0031] Step S2: After receiving the fused data, the remote server saves the collected fused data in a memory module, obtains multiple measurement parameters in the fused data, pre-sets multiple different pollution levels, sets a corresponding score for each measurement parameter for each pollution level, and sets a corresponding weight value for each measurement parameter. The pollution level of the fused data is judged based on the score and the weight value. If the pollution level is higher than the preset pollution level, it is judged that the sewage discharge does not meet the standard. The sewage composition data and the known pollution source information are matched to locate the pollution source. At the same time, the pollution source information is sent to relevant management personnel, and the fused data that does not meet the standard is marked as abnormal.

[0032] Specifically, in order to facilitate the subsequent use of the fused data, after receiving the fused data, the remote server saves the collected fused data in the memory module, and then judges the pollution degree of the obtained fused data. First, multiple measurement parameters in the fused data are obtained. For example, there are 4 measurement parameters, namely x1, x2, x3 and x4. For example, 4 pollution degrees are pre-set, namely excellent, good, medium and poor. For each pollution degree, a corresponding score is set for each measurement parameter. In other words, the possibility of judging the current water pollution degree is based on only one measurement parameter. For example, the score of the measurement parameter x1 belonging to the excellent pollution degree is 0.4, the correlation score belonging to the good pollution degree is 0.5, the correlation score belonging to the medium pollution degree is 0.6, and the correlation score belonging to the poor pollution degree is 0. 7. A weight value is also set for each measurement parameter. The weight value refers to the importance of the measurement parameter to the evaluation of the pollution level. The specific weight value calculation and setting method will be explained in detail later. Then, the pollution level of the fused data is judged based on the weight value and the score. If the pollution level is higher than the preset pollution level, it is judged that the sewage discharge does not meet the standard. The measurement parameters in the fused data are matched with the known pollution source information to locate the pollution source. The known pollution source information includes industrial wastewater pollution, domestic wastewater pollution, etc. The pollution source information is then sent to relevant management personnel. After judging that the sewage discharge does not meet the standard, the non-compliant fused data is also marked as an abnormality to facilitate the subsequent training of the prediction model. The above method can quickly discover the substandard water quality and provide technical support for water environment management and rapid response to water pollution incidents.

[0033] Step S3: Use the historically collected fusion data and corresponding abnormal labels as learning data to train the prediction model, and use the prediction model to predict and generate future sewage composition data based on the fusion data. When it is predicted that the future sewage composition data is not within the preset normal range, the corresponding abnormal situation will be sent to the relevant management personnel.

[0034] Specifically, in order to provide early warning for water pollution incidents, the historically collected fusion data and the corresponding abnormal labels are used as learning data to train the prediction model. In the process of training the prediction model, in order to speed up the generation of the prediction model and to obtain a more accurate prediction model, part of the data is first extracted from the historical fusion data for training to generate a first prediction model. The first prediction model is then tested to obtain the accuracy of the prediction model. If the accuracy does not meet the preset standard, more data is extracted from the historical fusion data to train the first prediction model until the generated prediction model meets the preset standard. The prediction model is then used to predict future fusion data and determine whether the fusion data will have abnormal conditions. If so, the corresponding abnormal conditions are sent to the application terminal of the relevant management personnel. The relevant management personnel can take countermeasures in advance based on the predicted abnormal conditions to prevent the occurrence of abnormal events.

[0035] Step S4: Based on the predicted sewage composition data and abnormal conditions, use the artificial intelligence model to generate corresponding response measures, estimate the implementation time of the response measures, implement the response measures after the implementation time is reached, and then continue to monitor the sewage discharge area and continue to collect fused data. Evaluate the implementation effect of the response measures based on the newly collected fused data, and adjust the artificial intelligence model according to the actual situation.

[0036] Specifically, when an abnormal situation is predicted in the sewage composition data, a pre-trained artificial intelligence model is used to generate corresponding response measures. The response measures include starting filtration facilities, increasing the amount of chlorine injection, etc. The implementation time of the response measures is also estimated, and the response measures are implemented after the implementation time is reached. The sewage discharge area is then continuously monitored to continue to collect fused data. The effectiveness of the response measures is also evaluated based on the newly collected fused data to ensure that the sewage composition data meets safety standards. After the implementation of the response measures, the above method continuously monitors and evaluates the sewage composition data to ensure the effectiveness and sustainability of water pollution management.

[0037] Further, data integration is performed based on the collection timestamp to obtain combined data, including the following steps:

[0038] A collection cycle is preset, and after the duration of the collection cycle has passed from the start time point of the collection, the timestamp range of each collection cycle is calculated based on the collection cycle and the start time point, and all sewage composition data and the corresponding device ID in the same collection cycle are obtained. The first value is obtained for all sewage composition data with the same device ID. If the first value is greater than the preset first threshold, the maximum value of all sewage composition data is obtained as the representative value of the device ID in the current collection cycle. If the first value is less than or equal to the first threshold, the second value of all sewage composition data is calculated, and the second value is used as the representative value corresponding to the smart sensor corresponding to the device ID in the current collection cycle. The representative values ​​corresponding to all device IDs in the sub-area are obtained, and the representative values, collection cycles and sub-areas are combined in a preset format to generate combined data.

[0039] Specifically, in order to be able to analyze the sewage composition data more accurately, a collection cycle is preset, and the collection cycle is used as the analysis unit of the sewage composition data, such as one hour. After the collection period has passed from the starting time point of the collection, the timestamp range of each collection cycle is calculated based on the collection cycle and the starting time point, and it is determined whether each collection timestamp is within the timestamp range. If so, it is determined that the sewage composition data corresponding to the collection timestamp is within the current collection cycle. If not, it is determined that the sewage composition data corresponding to the collection timestamp is not within the current collection cycle. All sewage composition data and the corresponding device ID in the same collection cycle are obtained, and a first value is calculated for all sewage composition data with the same device ID. The first value refers to the standard deviation or variance of all sewage composition data of the same device in the current collection cycle. If the first value is greater than the preset first threshold, it means that the sewage composition data of the same device ID is not very stable in the current collection cycle, and the maximum value of all sewage composition data is obtained as the device ID in the current collection. The representative value of the collection cycle is calculated. If the first value is less than or equal to the first threshold, it means that the sewage composition data of the same device ID in the current collection cycle is relatively stable. The second value of all sewage composition data is calculated. The second value refers to the average value of all sewage composition data of the same device in the current collection cycle. The second value is used as the representative value of the smart sensor corresponding to the device ID in the current collection cycle. The representative values ​​corresponding to all device IDs in the sub-area are obtained, and the representative value, collection cycle and sub-area are combined in a preset format to generate combined data. The sub-area refers to the identifier of the sub-area. Taking three sewage composition data as an example, the following object format combined data is generated: {value1: 0.008, value2: 0.001, value3: 0.002, value4: 0.002, cycle: cycle1, zone: zone1}. The integrated combined data is obtained through the above method, which is convenient for further processing and analysis.

[0040] Furthermore, the combined data belonging to different sub-regions are fused to generate fused data, including the following steps:

[0041] Obtain the accuracy of sewage composition data collected in each sub-area, assign different regional weights to different sub-areas based on the accuracy and the type of intelligent sensors in each sub-area, obtain sewage composition data of each sub-area with the same collection period, multiply the sewage composition data of different sub-areas by the corresponding regional weights and then add them together to generate comprehensive sewage composition data, and combine all comprehensive sewage composition data and collection periods to generate fused data.

[0042] Specifically, the accuracy of sewage composition data collected in each sub-area of ​​historical data is obtained. Different regional weights are assigned to different sub-areas based on the accuracy and the type of smart sensors in each sub-area. Assume that there are three sub-areas A, B and C. The accuracy of historical data collected in sub-area A is 0.9, the accuracy of historical data collected in sub-area B is 0.6, and the accuracy of historical data collected in sub-area C is 0.9. The types of smart sensors in sub-areas A and C are high-precision monitoring equipment, and the type of smart sensors in sub-area B is aging equipment with a long service life. Based on the above situation, since the accuracy of historical data collected in sub-areas A and C is high and the equipment is relatively advanced, the collected data The accuracy may be relatively high, so the regional weight of sub-area A and sub-area C is 0.4. The accuracy of historical collection data in sub-area B is relatively low and the equipment is relatively old. The accuracy of collected data may be reduced, so the regional weight of sub-area B is 0.2, and the sewage composition data of each sub-area with the same collection period is obtained. The sewage composition data of different sub-areas are multiplied by the corresponding regional weights and then added to obtain the comprehensive sewage composition data. All comprehensive sewage composition data and collection periods are combined to generate fused data. The data of each sub-area are fused through the above method to obtain the fused data of the entire sewage discharge area, which is convenient for further precise analysis based on the fused data.

[0043] Furthermore, setting a corresponding weight value for each measurement parameter includes the following steps:

[0044] Obtain the first evaluation value, second evaluation value and third evaluation value made by relevant professionals for each measurement parameter corresponding to other measurement parameters, obtain the first sum of all second evaluation values ​​of each measurement parameter relative to other measurement parameters, divide the first evaluation value, second evaluation value and third evaluation value corresponding to each measurement parameter by the corresponding first sum to obtain the first value, second value and third value, calculate the first average value of the first value, second value and third value, obtain multiple first average values ​​corresponding to the measurement parameter, calculate the second average value of the multiple first average values, obtain the second average values ​​corresponding to other measurement parameters, obtain the sum of all second average values ​​as the second sum, and divide the second average value corresponding to the measurement parameter by the second sum as the weight value corresponding to the measurement parameter.

[0045] Specifically, the other measurement parameters include the measurement parameters themselves, that is, the first evaluation value, the second evaluation value and the third evaluation value of the measurement parameter relative to its own measurement parameters are also obtained. Since they are relative to themselves, the first evaluation value, the second evaluation value and the third evaluation value of the measurement parameter relative to its own measurement parameters are all 1. The first evaluation value represents the minimum value of the importance of the measurement parameter to the evaluation of the degree of water pollution relative to other measurement parameters, the second evaluation value represents the most likely value of the importance of the measurement parameter to the evaluation of the degree of water pollution relative to other measurement parameters, and the third evaluation value represents the maximum value of the importance of the measurement parameter to the evaluation of the degree of water pollution relative to other measurement parameters. The following example illustrates the calculation process of the weight value. For example, there are three measurement parameters X, Y and Z, and the three evaluation values ​​of X to X are (1,1,1). Assuming that the three evaluation values ​​of X to X are (1,2,3), and the three evaluation values ​​of X to Z are (0.5,1,1.5), the first sum corresponding to X is 1+2+1=5. Divide the above three groups of numbers by 5 to obtain (0.2,0.2,0.2), (0.2,0.4, 0.6), (0.1, 0.2, 0.3), the three values ​​in each of these three groups correspond to the first value, the second value and the third value respectively. The average values ​​of each group of numbers are calculated to get the three first average values, which are 0.2, 0.4 and 0.2 respectively. The average value of the three first average values ​​is calculated to get the second average value, which is 0.2667. 0.2667 is the comprehensive importance of the measurement parameter X. The same method is used to obtain the comprehensive importance of the measurement parameter Y and the measurement parameter Y, assuming that they are 0.1377 and 0.5642 respectively. Then The weight value of measurement parameter X is 0.2667 / (0.2667+0.1377+0.5642), the weight value of measurement parameter Y is 0.1377 / (0.2667+0.1377+0.5642), and the weight value of measurement parameter Z is 0.5642 / (0.2667+0.1377+0.5642). The above method helps us calculate effective weight values ​​in the presence of uncertainty and obtain more reasonable and accurate weight values, providing scientific and reasonable data support for subsequent judgment of the degree of pollution.

[0046] Furthermore, judging the degree of contamination of the fused data based on the score and the weight value includes the following steps:

[0047] The weight value of each measurement parameter is obtained, and the corresponding score set for each measurement parameter for each pollution degree is obtained, and the pollution degree result is calculated using the first formula, which is: , ,in, are the weight values ​​corresponding to each measurement parameter, are the scores corresponding to the i-th measurement parameter for n different pollution levels, m is the number of measurement parameters, n is the number of pollution levels, and the first formula is used to obtain the m comprehensive values ​​corresponding to each measurement parameter for n different pollution levels. , obtain the pollution degree corresponding to the maximum value of the m comprehensive values ​​as the pollution degree of the fused data.

[0048] Specifically, in order to quickly obtain the pollution degree of the fused data and make the obtained results more scientific and reasonable, based on the weight value of the measurement parameter and the score set for the measurement parameter for each pollution degree, the above-mentioned first formula is used to calculate the comprehensive numerical value corresponding to each pollution degree, and the pollution degree corresponding to the maximum value is obtained as the pollution degree of the fused data. The data values ​​of multiple measurement parameters and weights are combined through the above method, which reflects the degree or possibility of each measurement parameter at each pollution degree in the form of possibility, quantifies a complex problem, and provides a more scientific basis for the final judgment of the pollution degree.

[0049] Furthermore, the historically collected fusion data and the corresponding anomaly labels are used as learning data to train the prediction model, including the following steps:

[0050] Create an initial prediction model, set initial parameter values ​​for the initial prediction model, the initial parameter values ​​are obtained based on historical experience data, obtain historical fusion data collected historically from a memory module, extract several data from the historical fusion data as a training data set, use the training data set to train the initial prediction model to generate a first prediction model, optimize the initial parameter values ​​during the training process, select several data from the historical fusion data as test data, use the test data to test the first prediction model to obtain the accuracy of the first prediction model, if the accuracy is less than a preset second threshold, update the training data set to generate a new training data set, use the new training data set to train the first prediction model, further optimize the initial parameter values ​​during the training process, generate a new prediction model to obtain the corresponding new accuracy, if the accuracy is still less than the second threshold, continue to update the training data set to train the first prediction model and optimize the parameter values ​​until the accuracy is greater than or equal to the second threshold.

[0051] Specifically, in order to improve the prediction accuracy of the prediction model and improve the training efficiency of the prediction model, an initial prediction model is created, and initial parameter values ​​are set based on historical experience data. Since a large amount of historical fusion data is stored in the memory module, and each fusion data includes multiple measurement parameters, it may be possible to train a prediction model that can accurately predict using only a small number of measurement parameters. Therefore, first, a part of the measurement parameters are extracted from each fusion data in the historical fusion data as training data. For example, if there are 20 measurement parameters in a fusion data, 5 measurement parameters are selected from the 20 measurement parameters as the training data set, and the initial prediction model is trained to generate the first prediction model. During the training process, the initial parameter values ​​are optimized, and part of the data is selected from the historical fusion data as test data. The accuracy of the first prediction model is obtained by test data. If the accuracy is greater than the preset second threshold, it means that the selected measurement parameters can train a prediction model for accurate prediction, and there is no need to use other measurement parameters to participate in the training. The prediction accuracy of the prediction model is ensured while the training efficiency of the prediction model is improved. If the accuracy is less than the second threshold, it means that the selected measurement parameters are not sufficient to train a prediction model for accurate prediction. The training data set is updated. The specific updating method will be explained later. The updated training data set is used to continue training the first prediction model until the accuracy of the first prediction model is greater than or equal to the second threshold. The above method trains the prediction model by gradually adding training data sets, which improves the training efficiency of the prediction model while ensuring the prediction accuracy of the prediction model.

[0052] Furthermore, some data are extracted from the historical fusion data as a training data set, including the following steps:

[0053] Obtain historical fusion data, obtain all measurement parameters and corresponding weight values ​​for each piece of historical fusion data, sort all measurement parameters by number from large to small according to the corresponding weight values ​​to generate a measurement parameter sequence, select several measurement parameters with forward numbers from the measurement parameter sequence as first measurement parameters, combine all first measurement parameters with corresponding collection cycles to generate a first fusion data, select the same first measurement parameters and corresponding collection cycles from the remaining historical fusion data to generate corresponding first fusion data, and combine all first fusion data to generate a training data set.

[0054] Specifically, the weight values ​​of the measurement parameters calculated in the previous step are obtained, and several first measurement parameters with the largest weight values ​​are selected from all the measurement parameters according to the size of the weight values ​​and combined with the corresponding collection cycles to generate the first fused data. For example, there are 20 measurement parameters, and the weight values ​​corresponding to the respective measurement parameters are 6, 6.5, 5, 5.6, 5.4, 5, 4.8, 4.7, 4.6, 4.5, 4.3, 4, 3.9, 3.7, 3.5, 3.4, 3.2, 3, 2.8, and 2.6. Assuming that 5 measurement parameters are selected, 5 measurement parameters corresponding to the weight values ​​of 6, 6.5, 5, 5.6, and 5.4 are selected. The same first measurement parameter is selected for each historical fused data, and combined with the corresponding collection cycle to generate the first fused data. All the first fused data are combined to generate a training data set. Through the above method, the measurement parameter with the largest weight value is selected based on the weight value as the combination to generate the training data set, which can improve the accuracy and reliability of the prediction model and also improve the training efficiency of the prediction model.

[0055] Further, updating the training data set to generate a new training data set includes the following steps:

[0056] From the remaining measurement parameter sequences that have not been selected, select a number of measurement parameters with forward numbers as second measurement parameters, and add the second measurement parameters to the training data set to generate a new training data set.

[0057] Specifically, if the training data set needs to be updated, several measurement parameters with the largest weight values ​​are selected from the remaining measurement parameter sequences that have not been selected according to the size of the weight values ​​and added to the training data set to generate a new training data set. Assuming that 5 measurement parameters are still selected, five measurement parameters corresponding to weight values ​​of 5, 4.8, 4.7, 4.6, and 4.5 are selected. Updating the training data set by the above method can ensure that measurement parameters with large weight values ​​are selected first, which is more conducive to the training of the prediction model.

[0058] In summary, the present invention provides an intelligent monitoring method for water pollution damage, the method includes dividing the sewage discharge area into multiple sub-areas, setting a number of intelligent sensors in each sub-area, collecting sewage composition data of the sub-areas, integrating data based on the collection timestamp to obtain combined data, and fusing the combined data belonging to different sub-areas to generate fused data; after the remote server receives the fused data, the pollution degree of the fused data is judged based on the score and weight value; the pollution degree of the fused data is judged based on the score and weight value; based on the predicted sewage composition data and abnormal conditions, the corresponding countermeasures are generated using an artificial intelligence model, the implementation time of the countermeasures is estimated, and the implementation effect of the countermeasures is evaluated based on the newly collected fused data. The present invention can quickly and accurately judge the pollution degree of water pollution.

[0059] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0060] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the above-mentioned program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0061] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0062] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

[0063] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent monitoring method for water pollution damage, characterized in that: The steps include: Divide the sewage discharge area into multiple sub-areas, set a number of smart sensors in each sub-area, set different collection intervals for the smart sensors based on the characteristics of the sewage composition data collected by the smart sensors, collect the sewage composition data of the sub-areas, and obtain the corresponding collection timestamp and device ID, integrate the sewage composition data corresponding to any sub-area based on the collection timestamp to generate combined data, also fuse all the combined data to generate fused data, and transmit the fused data to a remote server through a network; After receiving the fused data, the remote server saves the collected fused data in a memory module, obtains multiple measurement parameters in the fused data, pre-sets multiple different pollution levels, sets a corresponding score for each measurement parameter for each pollution level, and also sets a corresponding weight value for each measurement parameter, and judges the pollution level of the fused data based on the score and the weight value. If the pollution level is higher than the preset pollution level, it is judged that the sewage discharge does not meet the standard, matches the sewage composition data with the known pollution source information to locate the pollution source, and sends the pollution source information to relevant management personnel at the same time, and also marks the fused data that does not meet the standard as abnormal; The fused data collected historically and the corresponding abnormal marks are used as learning data to train a prediction model, and the prediction model is used to predict and generate future sewage composition data based on the fused data. When it is predicted that the future sewage composition data is not within a preset normal range, the corresponding abnormal situation is sent to relevant management personnel; Based on the predicted sewage composition data and abnormal conditions, corresponding countermeasures are generated using an artificial intelligence model, the implementation time of the countermeasures is estimated, and the countermeasures are implemented after the implementation time is reached, and then the sewage discharge area is continuously monitored and the fused data is continuously collected, and the implementation effect of the countermeasures is evaluated based on the newly collected fused data, and the artificial intelligence model is adjusted according to the actual situation; The data integration based on the collection timestamp to obtain the combined data includes the following steps: A collection cycle is preset, and after the length of the collection cycle has passed since the start time point of the collection, the timestamp range of each collection cycle is calculated based on the collection cycle and the start time point, and all the sewage component data and the corresponding device ID within the same collection cycle are obtained, and the first value is obtained for all the sewage component data with the same device ID. If the first value is greater than a preset first threshold, the maximum value of all the sewage component data is obtained as the representative value of the device ID in the current collection cycle. If the first value is less than or equal to the first threshold, the second value of all the sewage component data is calculated, and the second value is used as the representative value corresponding to the smart sensor corresponding to the device ID in the current collection cycle. The representative values ​​corresponding to all device IDs in the sub-area are obtained, and the representative value, the collection cycle and the sub-area are combined according to a preset format to generate the combined data; Wherein, judging the pollution degree of the fused data based on the score and the weight value comprises the following steps: The weight value of each measurement parameter is obtained, and the score corresponding to each measurement parameter is set for each pollution degree, and the pollution degree result is calculated using a first formula, wherein the first formula is: ,in, are the weight values ​​corresponding to each of the measurement parameters, are the scores corresponding to the i-th measurement parameter for n different pollution levels, m is the number of measurement parameters, n is the number of pollution levels, and the first formula is used to obtain the m comprehensive values ​​corresponding to each measurement parameter for n different pollution levels. , obtaining the pollution degree corresponding to the maximum value of the m comprehensive values ​​as the pollution degree of the fused data.

2. The method according to claim 1, characterized in that The combined data belonging to different sub-areas are fused to generate fused data, comprising the following steps: The accuracy of the sewage composition data collected in each sub-region is obtained, and based on the accuracy and the type of intelligent sensor in each sub-region, different regional weights are assigned to different sub-regions, and the sewage composition data of each sub-region with the same collection period is obtained. The sewage composition data of different sub-regions and the corresponding regional weights are multiplied and then added to generate comprehensive sewage composition data, and all the comprehensive sewage composition data and collection periods are combined to generate the fused data.

3. The method according to claim 1, characterized in that Setting a corresponding weight value for each of the measurement parameters includes the following steps: Obtain the first evaluation value, the second evaluation value and the third evaluation value made by relevant professionals for each of the measurement parameters corresponding to the other measurement parameters, obtain the first sum of all the second evaluation values ​​of each measurement parameter relative to the other measurement parameters, divide the first evaluation value, the second evaluation value and the third evaluation value corresponding to each measurement parameter by the corresponding first sum to obtain the first value, the second value and the third value, calculate the first average value of the first value, the second value and the third value, obtain multiple first average values ​​corresponding to the measurement parameter, calculate the second average value of the multiple first average values, obtain the second average values ​​corresponding to other measurement parameters, obtain the sum of all second average values ​​as the second sum, and divide the second average value corresponding to the measurement parameter by the second sum as the weight value corresponding to the measurement parameter.

4. The method according to claim 1, characterized in that: The fusion data collected historically and the corresponding abnormal labels are used as learning data to train a prediction model, including the following steps: An initial prediction model is created, and initial parameter values ​​are set for the initial prediction model, where the initial parameter values ​​are obtained based on historical experience data, historical fusion data collected historically is obtained from a memory module, and several data are extracted from the historical fusion data as a training data set, and the initial prediction model is trained using the training data set to generate a first prediction model, and the initial parameter values ​​are optimized during the training process, and several data are selected from the historical fusion data as test data, and the first prediction model is tested using the test data to obtain the accuracy of the first prediction model, and if the accuracy is less than a preset second threshold, the training data set is updated to generate a new training data set, and the first prediction model is trained using the new training data set, and the initial parameter values ​​are further optimized during the training process to generate a new prediction model to obtain the corresponding new accuracy, and if the accuracy is still less than the second threshold, the training data set is continued to be updated to train the first prediction model and optimize the parameter values ​​until the accuracy is greater than or equal to the second threshold.

5. The method according to claim 4, characterized in that Extracting a number of data from the historical fusion data as a training data set includes the following steps: Acquire the historical fusion data, acquire all the measurement parameters and the corresponding weight values ​​for each piece of the historical fusion data, sort all the measurement parameters by number from large to small according to the corresponding weight values ​​to generate a measurement parameter sequence, select several measurement parameters with forward numbers from the measurement parameter sequence as first measurement parameters, combine all the first measurement parameters and corresponding collection cycles to generate a first fusion data, select the same first measurement parameters and corresponding collection cycle combination from the remaining historical fusion data to generate the corresponding first fusion data, and combine all the first fusion data to generate the training data set.

6. The method according to claim 4, characterized in that Updating the training data set to generate a new training data set includes the following steps: Selecting a number of metric parameters with forward numbers from the remaining metric parameter sequences that have not been selected as second metric parameters, and adding the second metric parameters to the training data set to generate a new training data set.

Images