Online detection method and system for emerging pollutants in water body and storage medium

Abstract

The invention discloses an online detection method and system for emerging pollutants in a water body and a storage medium, and belongs to the technical field of water quality detection.The method comprises the steps that basic data are collected, and an initial knowledge graph is constructed based on the basic data; simplifying the initial knowledge graph to obtain an optimized graph; collecting environment information of a to-be-analyzed area, and determining potential pollutants and potential pollution sources in combination with the environment information and the optimization atlas; selecting N potential pollutants as preliminary screening pollutants, and obtaining a detection method of the preliminary screening pollutants in combination with the optimized atlas; if a sensor detection method exists in the detection methods, a corresponding sensor is used for collecting water quality data, and if only a sampling detection method is included, water quality is sampled to obtain a water quality sample; detecting the water quality sample in a laboratory to obtain water quality data, and analyzing the water quality data to generate the water quality improvement method. According to the method, the knowledge graph is constructed through existing literature data, so that the working efficiency of an emerging pollutant detection process is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of water quality detection, and specifically relates to an online detection method, system and storage medium for emerging pollutants in water. Background Art

[0002] The concentration of emerging pollutants in water bodies is usually low, but the health risks to organisms and humans are very high. For example, endocrine disruptors and persistent organic pollutants have the characteristics of high fat solubility and difficulty in biodegradation. They pose a serious threat to the safety of drinking water through water body transmission and bioaccumulation. Therefore, how to accurately determine the types of emerging pollutants in water quality has become the mainstream research direction.

[0003] In the prior art, for example, a Chinese patent document with publication number CN117310117A discloses a water quality management and assessment method for new pollutants. This method deploys a detection patrol robot equipped with a water quality sensor to the water area that needs to be monitored, obtains multiple parameters in the water area through the movement of the patrol robot, and conducts a comprehensive analysis of these parameters to determine whether the water quality has deteriorated. Another example is a Chinese patent document with publication number CN116908361A, which discloses a method for identifying and tracing new pollutants in the water environment of an industrial park basin. This method uses a high-throughput non-targeted screening method to obtain the types of emerging pollutants, and then combs through domestic and foreign literature to understand the main production and emission source types of antibiotics, locks in key enterprises in the park that need to be investigated, and locates the source of emerging pollutants in the water body by taking actual samples of key enterprises.

[0004] Since emerging pollutants usually need to be detected in laboratories, there are relatively few types that can be directly detected using sensors. Therefore, the first patent document mentioned above can detect relatively few types of emerging pollutants. The non-targeted screening method used in the detection process of the second patent document needs to ensure that the pollutants are not removed during the pre-treatment process, and appropriate mass spectrometry scanning parameters need to be set, which may lead to inaccurate detection problems. Summary of the invention

[0005] To solve the above problems, the present invention provides an online detection method, system and storage medium for emerging pollutants in water, so as to speed up the working efficiency of the existing emerging pollutant detection process.

[0006] In order to achieve the above-mentioned invention object, the present invention proposes an online detection method for emerging pollutants in water, comprising: Collect basic data, which are academic texts on publicly available emerging pollutants, and construct an initial knowledge graph based on the basic data; The initial knowledge graph is simplified to obtain an optimized graph, wherein the optimized graph includes nodes corresponding to the names, causes and detection methods of emerging pollutants respectively; Collect environmental information of the area to be analyzed, and determine potential pollutants and potential pollution sources by combining the environmental information and the causes in the optimization map; Selecting N potential pollutants as primary screening pollutants, and obtaining a detection method for the primary screening pollutants in combination with the optimization map; If the detection method includes a sensor detection method, the corresponding sensor is used to collect water quality data. If only a sampling detection method is included, the water quality is sampled to obtain a water quality sample; The water quality data is obtained after testing the water quality sample in the laboratory, and the water quality improvement method is generated by analyzing the water quality data.

[0007] Furthermore, constructing the initial knowledge graph includes the following steps: Extracting text from the basic data, splitting the text into multiple text paragraphs, selecting a large language model, generating a question sentence pattern, combining the question sentence pattern with the text paragraphs into a text to be analyzed, inputting the text to be analyzed into the large language model, and extracting the subject, predicate and object from the text to be analyzed and converting them into multiple standard sentences by the large language model; An extraction method is set, and the subject, predicate and object in the standard sentence are detected based on the extraction method, and the subject and the object are used as the nodes. According to the causal relationship between the subject and the object, directional connecting lines are marked between the nodes, and the corresponding predicates are marked on the connecting lines to complete the construction of the basic knowledge graph corresponding to the text to be analyzed, and the basic knowledge graphs of multiple texts to be analyzed are combined to obtain the initial knowledge graph.

[0008] Furthermore, optimizing the initial knowledge graph includes the following steps: Calculate the evaluation score of each of the connection lines in the initial knowledge graph, the higher the evaluation score, the higher the importance of the connection line, count the number of times each node is pointed to by the connection line, combine the nodes with the same number of times into a first set, set a first threshold, extract the first set with the number of times greater than the first threshold and define it as a second set; Counting the first number of the nodes included in the second set, setting a counter for each of the second sets, the initial value of each of the counters being 1, taking the ratio of the counter to the first number of the corresponding second set as an evaluation value, locating the second set corresponding to the smallest evaluation value, adding 1 to the corresponding counter value, calculating the evaluation value of each of the second sets again, locating the second set corresponding to the smallest evaluation value, adding 1 to the corresponding counter value, repeating this step until the number of increases of all the counters is equal to the preset simplified number, and defining the value of each counter at this time as a final value; Calculate the difference between the final value of each counter and 1, set the difference as the number of deletions of each second set, define the connection line pointing from each node in the second set to itself as the baseline, sort the baselines in the second set from small to large according to the evaluation score, delete the first M baselines, and use the knowledge graph obtained after the deletion as the optimized graph, where the value of M is the number of deletions.

[0009] Furthermore, calculating the evaluation score includes the following steps: Define the predicate marked by the k-th connecting line as the target word, the connecting line points from the first target point to the second target point, count the second number of connecting lines marked with the target word in the initial knowledge graph, and the third number of connecting lines marked with the target word in the connecting lines pointing to the second target point, and calculate the first score Xk of the k-th connecting line based on the first formula, the first formula is: , wherein ak is the third number, and bk is the second number; A fourth number of the connecting lines pointing to the first target point is counted, and a second score Yk of the k-th connecting line is calculated based on a second formula, where the second formula is: , wherein M is the fourth number, Xm is the first score of the mth connecting line, and the second score is used as the evaluation score.

[0010] Furthermore, when using sensors to collect water quality data, the sensors collect the concentration of emerging pollutants of corresponding types and record the collection time points, and generate a time change sequence of the concentration of emerging pollutants based on the collection time points. If the concentration of emerging pollutants exceeds a second threshold, a first warning message is generated; if the concentration of emerging pollutants in the time change sequence continuously increases at multiple collection time points, a second warning message is generated.

[0011] Furthermore, a spectral database is established, which includes characteristic spectral data of multiple emerging pollutants. When the water quality sample is detected using a spectral detection method, actual spectral data of each component in the water quality sample is obtained, and the actual spectral data is compared with the characteristic spectral data in the spectral database to identify the emerging pollutants present in the water quality sample.

[0012] Furthermore, the characteristic spectrum data includes ultraviolet-visible spectrum characteristics, infrared spectrum characteristics and Raman spectrum characteristics.

[0013] Further, obtaining the detection method includes the following steps: A detection list is constructed, wherein the detection list includes multiple detection methods, and a priority is set for each detection method. After the detection method of the emerging pollutant is obtained in the optimization map, the detection method with the highest priority is returned.

[0014] The present invention also provides an online detection system for emerging pollutants in water, which is used to implement the above-mentioned online detection method for emerging pollutants in water. The system comprises: A collection module, used to collect basic data, wherein the basic data is academic texts of disclosed emerging pollutants; A construction module, constructing an initial knowledge graph based on the basic data, simplifying the initial knowledge graph to obtain an optimized graph, wherein the optimized graph includes nodes corresponding to the names, causes and detection methods of emerging pollutants respectively; A primary screening module collects environmental information of the area to be analyzed, determines potential pollutants and potential pollution sources by combining the environmental information and the causes of generation in the optimization map, selects N types of potential pollutants as primary screening pollutants, and obtains a detection method for the primary screening pollutants by combining the optimization map; The detection module uses the corresponding sensor to collect water quality data if the detection method includes a sensor detection method. If it only includes a sampling detection method, the water quality is sampled to obtain a water quality sample, and the water quality data is obtained after the water quality sample is tested in the laboratory. The water quality data is analyzed to generate a water quality improvement method.

[0015] The present invention also discloses a computer storage medium, wherein the computer storage medium stores program instructions, wherein when the program instructions are executed, the device where the computer storage medium is located is controlled to execute the above-mentioned method.

[0016] Compared with the prior art, the beneficial effects of the present invention are at least as follows: In response to the current problem of a wide variety of emerging pollutants, the present invention automatically constructs a knowledge graph about emerging pollutants based on the existing text database. The constructed knowledge graph contains known causes and detection methods of various emerging pollutants. The detection personnel can automatically obtain various potential pollutants that may exist by inputting the acquired environmental information into the knowledge graph system, thereby saving the detection personnel the time of flipping through the materials. On this basis, if there is a method that can use a sensor in the detection method, the knowledge graph will give priority to recommendation. The use of sensor detection has the advantages of real-time and convenience. Therefore, through the knowledge graph, the detection personnel can obtain the most cutting-edge and convenient detection methods without consulting the information, thereby optimizing the detection process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the steps of the online detection method of emerging pollutants in water bodies of the present invention; Figure 2 It is a schematic diagram of the structure of the initial knowledge graph of the present invention; Figure 3 It is a structural schematic diagram of the online detection system for emerging pollutants in water bodies of the present invention. DETAILED DESCRIPTION

[0018] 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.

[0019] 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.

[0020] like Figure 1 As shown, an online detection method for emerging pollutants in water is S1: Collect basic data, which are academic texts on publicly available emerging pollutants, and build an initial knowledge graph based on the basic data.

[0021] S2: Simplify the initial knowledge graph to obtain an optimized graph, which includes nodes corresponding to the names, causes and detection methods of emerging pollutants.

[0022] Academic texts include various industry standards, journal articles, dissertations, blogs, patent texts, etc. published on the Internet. The initial knowledge graph is constructed by analyzing academic texts. There will be a large number of redundant connections in the initial knowledge graph generated by the original academic text. Therefore, after obtaining the initial knowledge graph, the present invention simplifies and optimizes it to obtain an optimized graph. There are multiple types of nodes in the optimized graph, specifically representing the name, nature, cause of generation, detection method, etc. of emerging pollutants. In the present invention, the cause of generation includes the source of the emerging pollutant, such as the production process A from the electroplating plant, or the flame retardant of the building material, and the detection method includes a specific detection method for obtaining emerging pollutants, such as using gas chromatography, liquid chromatography, etc.

[0023] Before or during testing, relevant personnel can quickly retrieve the desired detection method by optimizing the spectrum, or quickly determine the source or type of emerging pollutants, thereby omitting the literature query process and greatly improving the work efficiency of the entire analysis process.

[0024] S3: Collect environmental information of the area to be analyzed, and determine potential pollutants and potential pollution sources by combining the environmental information and the causes in the optimization map.

[0025] S4: Select N potential pollutants as primary screening pollutants, and obtain the detection method of the primary screening pollutants by combining the optimized spectrum.

[0026] First, specify the river or lake to be analyzed, and define the area within 5km around the river or lake as the area to be analyzed. Then collect the environmental information of the area to be analyzed, which includes the distribution of factories, businesses, farmlands, etc. in the area to be analyzed. After obtaining the environmental information, determine which emerging pollutants may exist in the water body based on the optimized map, and use these emerging pollutants as potential pollutants. There may be a large number of potential pollutants screened out, so select N potential pollutants with a higher probability of occurrence as the initial screening pollutants. The probability of occurrence can be determined based on actual conditions, such as the specific production process used in the factory, or the brand and type of pesticides used in farmland.

[0027] S5: If the detection method includes a sensor detection method, the corresponding sensor is used to collect water quality data; if only a sampling detection method is included, the water quality is sampled to obtain a water quality sample; S6: Obtain water quality data after testing water quality samples in the laboratory, and analyze the water quality data to generate water quality improvement methods.

[0028] On the basis of establishing a knowledge graph, the present invention uses the corresponding detection method for each initially screened pollutant to perform detection. If the academic text records that sensor A can be used to detect pollutant B, then the method will be extracted into the optimization graph. The detection personnel can know the existence of sensor A through the optimization graph, and can purchase sensor A and deploy it in the river area for remote detection. Or, if the academic text records the production of sensor A for detecting pollutant B, then the detection personnel can obtain sensor A according to its production method, thereby realizing remote monitoring of water quality data in the river area. The types of sensors include chemical sensors, optical sensors or biological sensors.

[0029] If there is no detection method with a sensor, water sampling is actually carried out on site to obtain water samples, and the water samples are analyzed by optimizing the detection method recorded in the atlas to determine whether the water contains the primary screening pollutants. For example, in this embodiment, the spectral features that appear for each primary screening pollutant are obtained, and the spectral features are compared with the spectral features already in the database to determine whether the pollutant exists in the water body.

[0030] In response to the current problem of a wide variety of emerging pollutants, the present invention automatically constructs a knowledge graph about emerging pollutants based on the existing text database. The constructed knowledge graph contains known causes and detection methods of various emerging pollutants. The detection personnel can automatically obtain various potential pollutants that may exist by inputting the acquired environmental information into the knowledge graph system, thereby saving the detection personnel the time of flipping through the materials. On this basis, if there is a method that can use a sensor in the detection method, the knowledge graph will give priority to recommendation. The use of sensor detection has the advantages of real-time and convenience. Therefore, through the knowledge graph, the detection personnel can obtain the most cutting-edge and convenient detection methods without consulting the information, thereby optimizing the detection process.

[0031] It is particularly noteworthy that the present invention constructs a knowledge graph through existing literature data, thereby speeding up the work efficiency of the emerging pollutant detection process.

[0032] This embodiment constructs the initial knowledge graph including the following steps: Extract text from basic data, split the text into multiple text paragraphs, select a large language model, generate question sentences, combine the question sentences with the text paragraphs into the text to be analyzed, input the text to be analyzed into the large language model, and the large language model extracts the subject, predicate and object in the text to be analyzed and converts them into multiple standard sentences.

[0033] Set the extraction method, detect the subject, predicate and object in the standard sentence based on the extraction method, take the subject and object as nodes, mark directional connecting lines between the nodes according to the causal relationship between the subject and the object, mark the corresponding predicate on the connecting line to complete the construction of the basic knowledge graph of the text to be analyzed, and combine the basic knowledge graphs of multiple texts to be analyzed to obtain an initial knowledge graph.

[0034] Specifically, the generated question sentence is, for example, "Extract the triples in the following paragraph, which are clauses containing a subject, a predicate, and an object." The question sentence and the text paragraph are input into the large language model together, and the large model will generate multiple standard sentences. For example, the standard sentences are "TBBPA, properties, persistent chemical substances," "TBBPA, lipophilicity, high," "TBBPA, extraction method from soil, sediment, sewage sludge, Soxhlet extraction method," etc. The extraction method is to extract words at predetermined positions. As above, the standard sentence includes three parts. Therefore, the name of the emerging pollutant is extracted from the first part, and the extraction method of the emerging pollutant is extracted from the third part. The predicate includes connection relationships such as "extraction method," "detection method," "pretreatment," and "cause of generation." For example, "TBBPA, extraction method from soil, sediment, sewage sludge, Soxhlet extraction method," and the node of TBBPA and the node of Soxhlet extraction method can be generated through this standard sentence, and the connection line between the two nodes is marked with the sewage detection method, and the node representing TBBPA points to the node representing the Soxhlet extraction method.

[0035] Based on the above steps, multiple basic knowledge graphs can be generated according to a paragraph. After processing all the basic data, a large number of basic knowledge graphs will be obtained. Therefore, these basic knowledge graphs need to be integrated. During the integration, the same nodes are merged while retaining the connection relationship in the original basic knowledge graph. After the integration is completed, the initial knowledge graph is obtained.

[0036] Compared with traditional construction methods, the large language model has been trained with a large amount of corpus and can accurately locate the subject, predicate and object in natural language. It can also organize them into standard sentences and quickly create triples. Therefore, with the help of the large speech model, the initial knowledge graph can be constructed quickly and accurately.

[0037] Optimizing the initial knowledge graph includes the following steps: Calculate the evaluation score of each connection line in the initial knowledge graph. The higher the evaluation score, the more important the connection line is. Count the number of times each node is pointed to by the connection line, combine nodes with the same number of times into a first set, set a first threshold, extract the first set with a number greater than the first threshold and define it as a second set. Since the initial knowledge graph is composed of a large number of basic knowledge graphs, there will inevitably be redundant connections. In academic text 1, it is recorded that emerging pollutant A can be detected by method E. In academic text 2, it is recorded that emerging pollutant A can be detected by method B first and then by method E. In fact, method B is the pre-processing of method E. For example, Figure 2 As shown in the figure, node A is connected to node B, node B is connected to node E, and node A is connected to node E. Then = the connection between node A and node E may be redundant; therefore, in this embodiment, the evaluation score of each connection line in the initial knowledge graph is first calculated, and the specific calculation method of the evaluation score will be introduced later. Then count the number of times each node is pointed to by a connection line, such as Figure 2 In the example, node A is pointed to once and node E is pointed to twice. Then nodes with the same number of points are grouped into the first set. For example, nodes A and C are in the same first set because their points are both 1. Here, the first threshold is set to 3, so the first set containing node E will be retained as the second set, and the first set containing node A will be filtered out.

[0038] Count the first number of nodes included in the second set, set a counter for each second set, the initial value of each counter is 1, use the ratio of the counter to the first number of the corresponding second set as the evaluation value, locate the second set corresponding to the numerical minimum evaluation value, add 1 to the corresponding counter value, calculate the evaluation value of each second set again, locate the second set corresponding to the numerical minimum evaluation value, add 1 to the corresponding counter value, repeat this step until the number of increases in all counters is equal to the preset streamlined number, and define the value of each counter at this time as the final value.

[0039] Specifically, before optimizing the knowledge graph, the number to be streamlined is first set, which can be determined based on the number of nodes in the initial knowledge graph, such as setting the streamlined number to one-fifth of the total number of nodes. The following example explains this step. Assuming that there is a second set 1-3, the second set 1 includes 4 nodes, the second set 2 includes 6 nodes, and the third set 3 includes 10 nodes. The streamlined number is set to 3, then the evaluation values ​​corresponding to the second sets 1-3 are 1 / 4, 1 / 6, and 1 / 10, respectively. The evaluation value of the minimum value is 1 / 10, which corresponds to the second set 3. Therefore, the counter of the second set 3 is increased by 1, and the evaluation value of the second set 3 is calculated again. The corresponding evaluation value is 2 / 10. At this time, the minimum evaluation value is 1 / 6, which corresponds to the second set 2. Then the counter of the second set 2 is increased by 1, and the process is repeated until the total value of all counters increases equals the streamlined number. It can be inferred that the final value of each counter in this embodiment is 2. Calculate the difference between the final value of each counter and 1, set the difference as the number of deletions in each second set, define the connection line from each node in the second set to itself as the baseline, sort the baselines in the second set from small to large according to the evaluation points, delete the first M baselines, and use the knowledge graph obtained after deletion as the optimized graph, where the value of M is the number of deletions.

[0040] Based on the above embodiment, the corresponding counter of each second set 1-3 is increased by 1, so the number of deletions in the second set 1-3 is 1, that is, the value of M is 1. Among the nodes included in the second set 1-3, the connection line pointing to the node itself is first located as the base line, that is, the arrow of the connection line points to itself. For example, there are two nodes in the second set 1, and both nodes have 4 base lines. Then there are 8 base lines in the second set 1. The evaluation scores of these 8 base lines are sorted from small to large, and the base line with the smallest value, that is, the base line with the frontmost position after sorting, is deleted. The deletion process of the second sets 2 and 3 is the same. Through this method, one of the connection lines in the node that has been pointed to too many times in the knowledge graph can be deleted, thereby streamlining the knowledge graph.

[0041] The calculation of the evaluation score in this embodiment includes the following steps: Define the predicate marked by the k-th connecting line as the target word, and the connecting line points from the first target point to the second target point. Count the second number of connecting lines marking the target word in the initial knowledge graph, and the third number of connecting lines marking the target word in the connecting lines pointing to the second target point. Calculate the first score Xk of the k-th connecting line based on the first formula. The first formula is: , where ak is the third quantity and bk is the second quantity.

[0042] The fourth number of connection lines pointing to the first target point is counted, and the second score Yk of the kth connection line is calculated based on the second formula, and the second formula is: , where M is the fourth number, Xm is the first score of the mth connecting line, and the second score is used as the evaluation score.

[0043] Continue to refer to Figure 2To explain this step, to calculate the evaluation score of the connection line L1, first obtain the predicate marked by the connection line L1. Here, the marked predicate is "detection method". The connection line L1 points from the first target point A to the second target point E. After statistics, there are 20 connection lines marked with "detection method" in the initial knowledge graph, so the second value is 20. Among the connection lines pointing to node E, connection lines L1-L3 are all marked with "detection method", so the third number is 3. Substituting the above parameters into the first formula, the first score of the connection line L1 can be obtained. In the first formula, the larger the second number, the higher the calculated first score, and the smaller the third number, the higher the calculated first score. In other words, among the connection lines pointing to node E, the more connection lines of the same type there are, the smaller the corresponding final evaluation score will be.

[0044] The first target point is A, and there is only one connection line L3 pointing to it, then the fourth number is 1, and the first score of L3 and the first score of L1 calculated previously are substituted into the second formula to obtain the second score, and the second score is the evaluation score. In the second formula, the higher the first score of the connection line L1, the higher the calculated evaluation score, and the lower the sum of the first scores of the connection lines pointing to node A, the higher the calculated evaluation score. In other words, as the previous node A of node E, the more times it is pointed to, the lower the evaluation score calculated for the connection line L1, that is, the more paths there are to reach node A, the lower the evaluation score of the connection line L1. Through the above calculation, the connection line with less replaceability will obtain a higher evaluation score. For example, from node D to node A, there is only one path from node A to node E, then the evaluation score corresponding to the connection line will be higher, and this connection line will not be deleted. Figure 2 If the connection line L1 is deleted, node A can still reach node E through node B, so the connection line L1 can be deleted. Through this step, the loss of the relationship between nodes in the process of deleting the connection line can be avoided.

[0045] In this embodiment, when using sensors to collect water quality data, the sensors collect the concentrations of emerging pollutants of corresponding types and record the collection time points, and generate a time change sequence of the concentrations of emerging pollutants based on the collection time points. If the concentration of emerging pollutants exceeds a second threshold, a first warning message is generated. If the concentration of emerging pollutants in the time change sequence increases continuously at multiple collection time points, a second warning message is generated.

[0046] Specifically, when using sensors to collect water quality data, the sensors use cellular networks, satellite communications, etc. to remotely send the collected emerging pollutant concentrations to a data server. The data server generates a time change sequence of the corresponding emerging pollutant concentration based on the collection time and stores it. A second threshold is set in the data server. When it is detected that the concentration data sent back by the sensor is greater than the second threshold, a first warning information of excessive pollutant concentration is generated; when it is detected that the pollutant concentration in the time change sequence continues to rise at multiple time points, a second warning information of continuously increasing river pollution is generated.

[0047] A spectral database is established, which includes characteristic spectral data of various emerging pollutants. When using spectral detection methods to detect water quality samples, the actual spectral data of each component in the water quality samples is obtained, and the actual spectral data is compared with the characteristic spectral data in the spectral database to identify emerging pollutants present in the water quality samples.

[0048] In this embodiment, the characteristic spectrum data includes ultraviolet-visible spectrum characteristics, infrared spectrum characteristics and Raman spectrum characteristics.

[0049] When using the spectral detection method, the present invention establishes a spectral database of various emerging pollutants in advance. Since the detection method is derived from relevant academic texts, when establishing the database, the corresponding drawings can be extracted from the academic texts as the characteristic spectral data of the corresponding emerging pollutants. In this way, the construction of the spectral database can be carried out simultaneously in the process of establishing the knowledge graph. By establishing the spectral database, the types of emerging pollutants present in the water quality sample can be quickly obtained.

[0050] Obtaining a detection method includes the following steps: Build a detection list, which includes multiple detection methods. Set a priority for each detection method. After obtaining the detection method of emerging pollutants in the optimization map, return the detection method with the highest priority.

[0051] This embodiment builds a detection list based on all existing detection methods, and marks the priority of each detection method in the detection list. For example, the detection method of the sensor in this embodiment has the highest priority. When the detection method of a certain emerging pollutant is obtained in the optimization map, the detection method with the highest priority is recommended. If the detection personnel are not satisfied, the detection method with the second highest priority is selected for recommendation.

[0052] like Figure 3 As shown, the present invention also provides an online detection system for emerging pollutants in water, which is used to implement the above-mentioned online detection method for emerging pollutants in water, and the system includes: The collection module is used to collect basic data, which are academic texts on publicly available emerging pollutants; A construction module builds an initial knowledge graph based on basic data, simplifies the initial knowledge graph, and obtains an optimized graph, which includes nodes corresponding to the names, causes, and detection methods of emerging pollutants; The initial screening module collects environmental information of the area to be analyzed, determines potential pollutants and potential pollution sources by combining the environmental information and the causes of generation in the optimization map, selects N potential pollutants as initial screening pollutants, and obtains the detection method of the initial screening pollutants by combining the optimization map; Detection module: If there is a sensor detection method in the detection method, the corresponding sensor is used to collect water quality data. If only a sampling detection method is included, the water quality is sampled to obtain water quality samples, and the water quality data is obtained after the water quality samples are tested in the laboratory. The water quality data is analyzed to generate a water quality improvement method.

[0053] The present invention also discloses a computer storage medium, wherein the computer storage medium stores program instructions, wherein when the program instructions are executed, the device where the computer storage medium is located is controlled to execute the above method.

[0054] It should be understood that the various technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various 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.

[0055] 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. A method for online detection of emerging pollutants in water, characterized in that: include: Collect basic data, which are academic texts on publicly available emerging pollutants, and construct an initial knowledge graph based on the basic data; The initial knowledge graph is simplified to obtain an optimized graph, wherein the optimized graph includes nodes corresponding to the names, causes and detection methods of emerging pollutants respectively; Collect environmental information of the area to be analyzed, and determine potential pollutants and potential pollution sources by combining the environmental information and the causes in the optimization map; Selecting N potential pollutants as primary screening pollutants, and obtaining a detection method for the primary screening pollutants in combination with the optimization map; If the detection method includes a sensor detection method, the corresponding sensor is used to collect water quality data. If only a sampling detection method is included, the water quality is sampled to obtain a water quality sample; The water quality data is obtained after testing the water quality sample in the laboratory, and the water quality improvement method is generated by analyzing the water quality data.

2. The method according to claim 1, characterized in that Constructing the initial knowledge graph includes the following steps: Extracting text from the basic data, splitting the text into multiple text paragraphs, selecting a large language model, generating a question sentence pattern, combining the question sentence pattern with the text paragraphs into a text to be analyzed, inputting the text to be analyzed into the large language model, and extracting the subject, predicate and object from the text to be analyzed and converting them into multiple standard sentences by the large language model; An extraction method is set, and the subject, predicate and object in the standard sentence are detected based on the extraction method, and the subject and the object are used as the nodes. According to the causal relationship between the subject and the object, directional connecting lines are marked between the nodes, and the corresponding predicates are marked on the connecting lines to complete the construction of the basic knowledge graph corresponding to the text to be analyzed, and the basic knowledge graphs of multiple texts to be analyzed are combined to obtain the initial knowledge graph.

3. The method according to claim 2, characterized in that Optimizing the initial knowledge graph includes the following steps: Calculate the evaluation score of each of the connection lines in the initial knowledge graph, the higher the evaluation score, the higher the importance of the connection line, count the number of times each node is pointed to by the connection line, combine the nodes with the same number of times into a first set, set a first threshold, extract the first set with the number of times greater than the first threshold and define it as a second set; Counting the first number of the nodes included in the second set, setting a counter for each of the second sets, the initial value of each of the counters being 1, taking the ratio of the counter to the first number of the corresponding second set as an evaluation value, locating the second set corresponding to the smallest evaluation value, adding 1 to the corresponding counter value, calculating the evaluation value of each of the second sets again, locating the second set corresponding to the smallest evaluation value, adding 1 to the corresponding counter value, repeating this step until the number of increases of all the counters is equal to the preset simplified number, and defining the value of each counter at this time as a final value; Calculate the difference between the final value of each counter and 1, set the difference as the number of deletions of each second set, define the connection line pointing from each node in the second set to itself as the baseline, sort the baselines in the second set from small to large according to the evaluation score, delete the first M baselines, and use the knowledge graph obtained after the deletion as the optimized graph, where the value of M is the number of deletions.

4. The method according to claim 3, characterized in that: Calculating the evaluation score includes the following steps: Define the predicate marked by the k-th connecting line as the target word, the connecting line points from the first target point to the second target point, count the second number of connecting lines marked with the target word in the initial knowledge graph, and the third number of connecting lines marked with the target word in the connecting lines pointing to the second target point, and calculate the first score Xk of the k-th connecting line based on the first formula, the first formula is: , wherein ak is the third number, and bk is the second number; A fourth number of the connecting lines pointing to the first target point is counted, and a second score Yk of the k-th connecting line is calculated based on a second formula, where the second formula is: , wherein M is the fourth number, Xm is the first score of the mth connecting line, and the second score is used as the evaluation score.

5. The method according to claim 1, characterized in that: When using sensors to collect water quality data, the sensors collect the concentration of emerging pollutants of corresponding types and record the collection time points, and generate a time change sequence of the concentration of emerging pollutants based on the collection time points. If the concentration of emerging pollutants exceeds a second threshold, a first warning message is generated. If the concentration of emerging pollutants in the time change sequence continuously increases at multiple collection time points, a second warning message is generated.

6. The method according to claim 1, characterized in that A spectral database is established, which includes characteristic spectral data of multiple emerging pollutants. When the water quality sample is detected using a spectral detection method, actual spectral data of each component in the water quality sample is obtained, and the actual spectral data is compared with the characteristic spectral data in the spectral database to identify the emerging pollutants present in the water quality sample.

7. The method according to claim 6, characterized in that The characteristic spectrum data includes ultraviolet-visible spectrum characteristics, infrared spectrum characteristics and Raman spectrum characteristics.

8. The method according to claim 1, characterized in that Obtaining a detection method includes the following steps: A detection list is constructed, wherein the detection list includes multiple detection methods, and a priority is set for each detection method. After the detection method of the emerging pollutant is obtained in the optimization map, the detection method with the highest priority is returned.

9. An online detection system for emerging pollutants in water, used to implement the method according to any one of claims 1 to 8, characterized in that: include: A collection module, used to collect basic data, wherein the basic data is academic texts of disclosed emerging pollutants; A construction module, constructing an initial knowledge graph based on the basic data, simplifying the initial knowledge graph to obtain an optimized graph, wherein the optimized graph includes nodes corresponding to the names, causes and detection methods of emerging pollutants respectively; A primary screening module collects environmental information of the area to be analyzed, determines potential pollutants and potential pollution sources by combining the environmental information and the causes of generation in the optimization map, selects N types of potential pollutants as primary screening pollutants, and obtains a detection method for the primary screening pollutants by combining the optimization map; The detection module uses the corresponding sensor to collect water quality data if the detection method includes a sensor detection method. If it only includes a sampling detection method, the water quality is sampled to obtain a water quality sample, and the water quality data is obtained after the water quality sample is tested in the laboratory. The water quality data is analyzed to generate a water quality improvement method.

10. A computer storage medium, characterized in that: The computer storage medium stores program instructions, wherein when the program instructions are executed, the device where the computer storage medium is located is controlled to execute the method according to any one of claims 1 to 8.

Images