Method and platform for monitoring and early warning of odor in waterworks

Abstract

The application discloses a tap water plant peculiar smell monitoring and early warning method and platform, relates to the related technical field of monitoring and early warning, and the method comprises the following steps: identifying key nodes based on the structure design information of a target tap water plant, deploying a sensor network, and constructing M water plant perception networks; obtaining water plant node perception data flow, building a peculiar smell identification channel, and identifying peculiar smell to obtain M initial node peculiar smell parameters; generating a peculiar smell diffusion influence network and performing influence gain analysis according to the spatial distribution and process sequence of the key nodes of the water plant process, and determining the M water plant node peculiar smell parameters; constructing a hierarchical early warning mechanism, performing early warning grade matching and peculiar smell hierarchical early warning. The technical problems that the present application solves are that the sensor deployment lacks pertinence, peculiar smell identification and diffusion analysis are disconnected, and the real-time performance and accuracy of water plant peculiar smell monitoring and early warning are poor, and the technical effects of improving the real-time performance and coverage of peculiar smell monitoring and enhancing the accuracy of peculiar smell identification are achieved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of monitoring and early warning, and particularly relates to a tap water plant odor monitoring and early warning method and platform. BACKGROUND

[0002] How to realize real-time monitoring, accurate identification and rapid early warning of the odor of a tap water plant has become a key problem to be solved in the water treatment industry. Traditional tap water plant water quality monitoring relies on laboratory analysis, and has long detection cycles, poor real-time performance and limited coverage, and it is difficult to timely detect odor abnormalities. By deploying a multi-parameter sensor network, continuous monitoring of key process nodes in a water plant can be realized. However, odor substances can be caused by various factors such as algal metabolites, industrial pollutants, disinfection by-products or pipeline corrosion, and single sensor data cannot accurately identify odor types and pollution levels. In addition, the dynamic changes of water flow in the water plant, such as mixing, sedimentation, filtration and disinfection, will affect the diffusion path and concentration distribution of the odor, and affect the accuracy and real-time performance of the monitoring and early warning of the water quality of the water plant.

[0003] Therefore, in the related art, the sensor deployment lacks pertinence, the odor identification is disconnected from the diffusion analysis, and the real-time performance and accuracy of the odor monitoring and early warning of the water plant are poor. SUMMARY

[0004] The application provides a tap water plant odor monitoring and early warning method and platform, which solves the technical problems of lack of pertinence in sensor deployment, disconnection between odor identification and diffusion analysis, and poor real-time performance and accuracy of odor monitoring and early warning of the water plant in the prior art, and achieves the technical effects of improving the real-time performance and coverage of odor monitoring and enhancing the accuracy of odor identification.

[0005] The application provides a tap water plant odor monitoring and early warning method, which comprises the following steps: identifying key nodes based on structure design information of a target tap water plant to obtain M water plant process key nodes, and deploying a sensor network on the M water plant process key nodes to construct M water plant perception networks; monitoring and acquiring M water plant node perception data streams by using the M water plant perception networks, building a water plant odor identification channel, identifying odor based on the water plant odor identification channel, and obtaining M initial node odor parameters; generating an odor diffusion influence network according to the spatial distribution and process sequence of the M water plant process key nodes, analyzing the influence gain of the M initial node odor parameters based on the odor diffusion influence network, and determining M water plant node odor parameters; constructing a hierarchical early warning mechanism, and matching the M water plant node odor parameters with early warning levels and performing odor hierarchical early warning by using the hierarchical early warning mechanism.

[0006] In a possible implementation, the waterworks odor monitoring and early warning method further performs the following processing: parameter extraction is performed on structure design information of the target waterworks to obtain a waterworks key parameter set, the waterworks key parameter set including a spatial layout parameter, a treatment process flow parameter, and a process equipment parameter; three-dimensional modeling is performed based on the spatial layout parameter and the process equipment parameter to generate a waterworks spatial model; N key treatment process sections are divided according to the treatment process flow parameter; and M waterworks process key nodes are obtained by performing key point identification on the waterworks spatial model based on the N key treatment process sections.

[0007] In a possible implementation, the waterworks odor monitoring and early warning method further performs the following processing: process node association is performed between the N key treatment process sections and the waterworks spatial model respectively to obtain an N-process-section-associated-node set; waterworks treatment simulation parameters are determined according to an application scenario of the target waterworks; finite element simulation is performed on the waterworks spatial model by applying the waterworks treatment simulation parameters to the waterworks spatial model to obtain odor substance distribution parameters; and key point screening is performed on the N-process-section-associated-node set based on the odor substance distribution parameters to obtain the M waterworks process key nodes, where M is greater than or equal to N.

[0008] In a possible implementation, the waterworks odor monitoring and early warning method further performs the following processing: M process node sensor parameters are obtained by performing monitoring demand analysis and sensor selection on the M waterworks process key nodes respectively; M process node deployment location parameters are determined by performing deployment location analysis on the M waterworks process key nodes based on the odor substance distribution parameters; M node sensor deployment schemes are generated according to the M process node sensor parameters and the M process node deployment location parameters; and a sensor network is deployed on the M waterworks process key nodes according to the M node sensor deployment schemes to construct the M waterworks perception networks.

[0009] In a possible implementation, the waterworks odor monitoring and early warning method further performs the following processing: waterworks node odor monitoring data is obtained by performing historical odor data mining and generating process node identification based on the target waterworks; odor identification key elements are obtained, the odor identification key elements including an odor type, a concentration level, and a diffusion range; a waterworks odor element identification branch channel is obtained by performing classification identification training on the waterworks node odor monitoring data set by using the odor identification key elements; and the waterworks odor element identification branch channel is fused in parallel to build the waterworks odor identification channel.

[0010] In a possible implementation, the tap water plant odor monitoring and early warning method further performs the following processing: based on the spatial distribution and node process sequence of the M water plant process key nodes, diffusion path analysis is performed to generate an odor diffusion node path network; based on the tap water plant node odor monitoring data set, the diffusion influence of the odor diffusion node path network is quantified to determine a diffusion path influence coefficient set; the diffusion path influence coefficient set is mapped to the odor diffusion node path network for identification to generate the odor diffusion influence network.

[0011] In a possible implementation, the tap water plant odor monitoring and early warning method further performs the following processing: node coordinate extraction is performed on the spatial distribution of the M water plant process key nodes to obtain a process node spatial coordinate database; process topology connection is performed on the M water plant process key nodes according to the process node spatial coordinate database and the node process sequence to construct a process node topology graph; odor diffusion directional identification is performed on each connected node in the process node topology graph to generate the odor diffusion node path network.

[0012] In a possible implementation, the tap water plant odor monitoring and early warning method further performs the following processing: the tap water plant node odor monitoring data set is associated and clustered with each odor diffusion path in the odor diffusion node path network to obtain a plurality of diffusion path associated odor data sets; diffusion concentration decay fitting quantification is respectively performed on the plurality of diffusion path associated odor data sets to determine a plurality of path diffusion decay coefficients; and the diffusion path influence coefficient set is determined according to the plurality of path diffusion decay coefficients.

[0013] In a possible implementation, the tap water plant odor monitoring and early warning method further performs the following processing: based on the odor diffusion influence network, diffusion influence accumulation is sequentially performed on the M initial node odor parameters to obtain M node odor influence cumulative coefficients; dynamic gain analysis is performed on the M initial node odor parameters through the M node odor influence cumulative coefficients to determine the M water plant node odor parameters.

[0014] The application also provides a waterworks odor monitoring and early warning platform, comprising: a waterworks sensing network construction module, configured to identify key nodes based on structure design information of a target waterworks, obtain M waterworks process key nodes, and deploy a sensor network on the M waterworks process key nodes to construct M waterworks sensing networks; a waterworks odor identification module, configured to monitor and acquire M waterworks node sensing data streams by using the M waterworks sensing networks, build a waterworks odor identification channel, identify odor based on the waterworks odor identification channel, and obtain M initial node odor parameters; an influence gain analysis module, configured to generate an odor diffusion influence network according to the spatial distribution and process sequence of the M waterworks process key nodes, analyze the influence gain of the M initial node odor parameters based on the odor diffusion influence network, and determine M waterworks node odor parameters; and an odor grading early warning module, configured to construct a grading early warning mechanism, and match the M waterworks node odor parameters with early warning levels and perform odor grading early warning by using the grading early warning mechanism.

[0015] The waterworks odor monitoring and early warning method and platform provided by the application identify key nodes based on structure design information of a target waterworks, deploy a sensor network, construct M waterworks sensing networks, acquire waterworks node sensing data streams, build a waterworks odor identification channel, identify odor, obtain M initial node odor parameters, generate an odor diffusion influence network according to the spatial distribution and process sequence of waterworks process key nodes, analyze the influence gain, determine M waterworks node odor parameters, construct a grading early warning mechanism, and match the M waterworks node odor parameters with early warning levels and perform odor grading early warning. The technical problems of poor real-time performance and accuracy of waterworks odor monitoring and early warning caused by the lack of pertinence of sensor deployment and the disconnection between odor identification and diffusion analysis in the prior art are solved, and the technical effects of improving the real-time performance and coverage of odor monitoring and enhancing the accuracy of odor identification are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments of the present disclosure will be briefly introduced below. The flowcharts are used in the present application to illustrate the operations performed by the platform according to the embodiments of the present application. It should be understood that the foregoing or the following operations are not necessarily performed in sequence. On the contrary, various steps can be processed in reverse order or simultaneously according to needs. Meanwhile, other operations can be added to these processes, or a step or several steps can be removed from these processes.

[0017] Figure 1 The waterworks odor monitoring and early warning method provided by the embodiments of the present application is shown in the flowchart.

[0018] Figure 2 The structure of the waterworks odor monitoring and early warning platform provided by the embodiments of the present application is shown in the schematic diagram.

[0019] The reference signs are as follows: water plant perception network construction module 10, water plant odor identification module 20, influence gain analysis module 30, and odor grading early warning module 40. DETAILED DESCRIPTION

[0020] The above description is only a summary of the technical scheme of the present application. In order to enable the technical means of the present application to be more clearly understood, and to be implemented according to the content of the description, and in order to enable the above and other purposes, characteristics and advantages of the present application to be more apparent and easy to understand, the following specific embodiments of the present application are described.

[0021] In order to make the purposes, technical schemes and advantages of the present application more clear, the present application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting the present application, and all other embodiments obtained by those skilled in the art without making creative efforts fall within the scope of protection of the present application.

[0022] In the following description, "some embodiments" are referred to, which describe a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict. The term "first\second" referred to is only to distinguish similar objects, and does not represent a specific order of the objects. The terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, platform, product or server including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art in the technical field to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application.

[0023] The embodiments of the present application provide a tap water plant odor monitoring and early warning method, as shown in the method, the method comprises: Figure 1

[0024] Step S100, based on the structure design information of the target tap water plant, key nodes are identified, M water plant process key nodes are obtained, and a sensor network is deployed on the M water plant process key nodes to construct M water plant perception networks.

[0025] ​The step S100 further comprises a step S110 of performing parameter extraction on the structure design information of the target waterworks to obtain a waterworks key parameter set, the waterworks key parameter set comprising a spatial layout parameter, a treatment process flow parameter, and a process equipment parameter; a step S120 of performing three-dimensional modeling based on the spatial layout parameter and the process equipment parameter to generate a waterworks spatial model; a step S130 of dividing N key treatment process sections according to the treatment process flow parameter; and a step S140 of performing key point identification on the waterworks spatial model based on the N key treatment process sections to obtain M waterworks process key nodes.

[0026] Preferably, the structure design information of the target waterworks is obtained by analyzing design drawings, process manuals, equipment lists, and the like, and parameter extraction is performed thereon to obtain the spatial layout parameter, the treatment process flow parameter, and the process equipment parameter, wherein the spatial layout parameter comprises a plant plan, geometric dimensions of structures such as sedimentation tanks, filter tanks, and clear water tanks, relative positions and elevations, and the like; the treatment process flow parameter refers to the process of water treatment, such as coagulation → sedimentation → filtration → disinfection, water flow direction, residence time, treatment capacity, and the like; and the process equipment parameter refers to the model, installation position, and operating parameter of equipment such as water pumps, dosing devices, ozone generators, and activated carbon filter tanks, thereby forming a structured waterworks key parameter set.

[0027] Preferably, modeling is performed using a BIM (Building Information Modeling) or a CAD (Computer Aided Design) tool according to the spatial layout parameter and the process equipment parameter to obtain a visual waterworks spatial model, and the geometric shape of the sedimentation tank, the filter tank, the pipe connection mode, and the equipment position are accurately labeled in the waterworks spatial model; and the entire water treatment process of the waterworks is divided into N key treatment process sections according to the treatment process flow parameter, the functional independence of each section, the water quality variation characteristics, and the potential sources of odor, wherein N is a positive integer greater than 1, representing the total number of water treatment process sections, for example, N = 6, including a raw water inlet process section, a pretreatment process section, a coagulation and sedimentation process section, a filtration process section, a disinfection process section, and a finished water process section, as well as a pipe network stage and a secondary water supply stage, wherein old metal pipes release iron, manganese, and the like due to corrosion, generating an iron rust smell, and poor quality plastic pipes may release organic compounds, causing abnormal odor, and community water storage tanks and water tanks that are not cleaned in a timely manner may breed microorganisms or accumulate impurities, causing mold and fishy odors.

[0028] Preferably, N key process segments identify key points of the waterworks space model, wherein the key nodes are the core positions of odor monitoring, which need to meet the process sensitivity and spatial representativeness, and are prone to odor, such as chlorine residue in the disinfection segment, algae breeding in the sedimentation segment, or reflect the overall water quality of the process segment, such as the inlet / outlet of the filter tank. Specifically, in each water treatment process segment, all possible monitoring points are marked, such as the inlet, middle, and outlet, to form a set of N points. Then, the distribution of odor substances is analyzed and screened through finite element simulation, including simulating the diffusion path of odor substances such as algae toxins and chloroform through fluid dynamics, analyzing the concentration distribution, and selecting nodes with significant concentration gradient changes or long-term high-risk, such as the sedimentation tank outlet prone to earthy smell due to algae aggregation. Finally, M waterworks process key nodes are determined.

[0029] Preferably, according to the risk characteristics of each waterworks process key node, a matching sensor is selected and deployed at important locations such as the inlet of the sedimentation tank, the middle of the filter tank, and the outlet of the disinfection tank to form a sensor network, i.e., a cluster of Internet of Things sensing devices, for real-time collection of water quality parameters such as chlorophyll, total organic carbon, ammonia nitrogen, pH, turbidity, and residual chlorine, iron, manganese, and oxidation-reduction potential. Each sensor at a waterworks process key node is connected through wireless or wired communication to generate M waterworks perception networks, enabling dynamic perception of odors throughout the waterworks, and providing high-reliability data for odor identification and hierarchical early warning.

[0030] Further, step S140 further includes step S141 of associating N key process segments with the waterworks space model to obtain a set of N process segment associated nodes; step S142 of determining waterworks treatment simulation parameters according to the application scenario of the target waterworks; step S143 of applying the waterworks treatment simulation parameters to the waterworks space model for finite element simulation to obtain odor substance distribution parameters; and step S144 of key point screening of the set of N process segment associated nodes based on the odor substance distribution parameters to obtain M waterworks process key nodes, wherein M≥N.

[0031] Preferably, based on N key process segments respectively associated with the waterworks spatial model at process node, i.e. for each key process segment in the waterworks spatial model, all potential monitoring points contained therein are labeled to determine the N process segment association node set, for example, the sedimentation segment association node set includes the inlet, the middle baffle, the outlet, and the sludge outlet, and the disinfection segment association node set includes the chlorination contact tank inlet, the middle, the outlet, and the residual chlorine monitor installation point; according to the application scenario of the target waterworks, historical water quality data, process design manuals, and real-time monitoring data are obtained to analyze and determine waterworks treatment simulation parameters, including hydraulic parameters, pollution source parameters, and environmental parameters, wherein the hydraulic parameters include flow, flow rate, and water flow direction, the pollution source parameters include odor substance type, initial concentration, and release location, and the environmental parameters include temperature, pH value, etc. According to the waterworks spatial model, a waterworks fluid dynamics model is established using CFD software, the waterworks treatment simulation parameters are injected thereinto, and finite element simulation calculation is performed to obtain odor substance distribution parameters, including an odor substance concentration field, i.e. the concentration distribution of each point in a three-dimensional space, such as a higher chloroform concentration at the outlet of the disinfection tank, and a diffusion path, i.e. the migration trajectory of the odor substance along with the water flow, such as the diffusion of algal toxins from the sedimentation tank to the filtration tank; finally, based on the odor substance distribution parameters, the N process segment association node set is screened for key points, i.e. in each key process segment, combined with the water quality change characteristics, potential pollution risks, and necessary nodes for odor substance migration, M waterworks process key nodes are identified and determined, wherein M≥N, indicating that each key process segment may contain multiple key nodes, such as the need to monitor the inlet, filter layer, and backwash drainage of the filtration segment.

[0032] Further, step S100 further includes step S150 of respectively performing monitoring demand analysis and sensor selection on the M waterworks process key nodes to obtain M process node sensor parameters; step S160 of performing deployment location analysis on the M waterworks process key nodes based on the odor substance distribution parameters to determine M process node deployment location parameters; step S170 of generating M node sensor deployment schemes according to the M process node sensor parameters and the M process node deployment location parameters; and step S180 of deploying a sensor network on the M waterworks process key nodes according to the M node sensor deployment schemes to construct the M waterworks perception networks.

[0033] Preferably, for the process characteristics and odor risk characteristics of each waterworks process key node, such as algal metabolism in the sedimentation tank and chloroform generation in the disinfection tank, monitoring demand analysis is performed, i.e. the parameters to be monitored are analyzed, such as 2-methylisoborneol, residual chlorine, THM Setc. and determine the technical parameters of each sensor such as range, accuracy and sampling frequency, etc. to finally form M process node sensor parameters. Based on the odor substance distribution parameters obtained from the previous finite element simulation, combined with the physical conditions such as water flow state and equipment layout on site, the optimal installation position of each sensor is determined, for example, in the sedimentation tank, the algae sensor is deployed at the water depth of 1.2-1.5 m (algae enrichment layer); in the pipeline, avoid the vortex area, select the middle of the straight pipe section to ensure data representativeness, and then generate M process node deployment position parameters. Then, the M process node sensor parameters and M process node deployment position parameters are fused to develop M sets of node sensor deployment schemes including equipment list, installation drawing and communication configuration. Finally, according to the M node sensor deployment schemes, the sensor network is deployed at the M process key nodes of the waterworks, and each node sensor is integrated through the Internet of Things, and finally M waterworks sensing networks are constructed to support real-time data acquisition, edge computing and cloud collaborative analysis to ensure high spatio-temporal resolution of odor early warning and accurate matching of risk early warning.

[0034] Step S200, using the M waterworks sensing networks to monitor and obtain M waterworks node sensing data streams, building a waterworks odor identification channel, and identifying the odor of the M waterworks node sensing data streams based on the waterworks odor identification channel to obtain M initial node odor parameters.

[0035] Preferably, the waterworks sensing network deployed through the M key nodes, such as electronic nose, GC-MS, optical sensor, etc. monitors and obtains multi-dimensional water quality parameters at a set sampling frequency to form M waterworks node sensing data streams; the historical process node odor monitoring data of the target waterworks is obtained, which is preprocessed including standardization, missing value filling and feature extraction to determine the training data, and then an identification model is constructed based on machine learning such as SVM, the training data is trained to obtain parallel identification branches which can identify and detect different odor types, and then the attention mechanism is used to weight and integrate each identification branch to determine the waterworks odor identification channel; then the M waterworks node sensing data streams are identified by the waterworks odor identification channel, and the structured M initial node odor parameters are output, which may include odor type, intensity, time decay, etc. to ensure the speed and accuracy of odor identification.

[0036] Further, step S200 further comprises step S210 of performing historical odor data mining and generating process node identification based on the target waterworks to obtain a waterworks node odor monitoring dataset; step S220 of acquiring odor identification key elements, the odor identification key elements including odor type, concentration level, and diffusion range; step S230 of performing classification and identification training on the waterworks node odor monitoring dataset using the odor identification key elements to obtain a waterworks odor element identification branch channel; and step S240 of performing parallel fusion of the waterworks odor element identification branch channel to build the waterworks odor identification channel.

[0037] Preferably, historical odor data mining is performed based on the target waterworks, that is, odor event records, water quality detection reports, and sensor logs accumulated by the waterworks over the years are collected, and process node identification is generated, that is, a unique identification is added to each process node such as the outlet of the sedimentation tank and the inlet of the filter tank, and the process attributes thereof are associated, and then the waterworks node odor monitoring dataset is obtained by alignment, including odor monitoring historical data of each process node such as timestamp, node position, meteorological data, water quality parameters, and odor event records. The odor type, concentration level, and diffusion range are determined as odor identification key elements, wherein the odor type is used as the output label of the classification model, such as “earthy / moldy smell”, “fishy smell”, “chlorine / disinfection byproduct smell”, etc.; the concentration level refers to the mapping of odor intensity or odorant concentration to 0-5 levels, which is used as the output label of the regression model; and the diffusion range refers to the number of downstream nodes affected, which is used as the output label of the regression model.

[0038] Preferably, classification and identification training is performed on the waterworks node odor monitoring dataset according to the odor identification key elements, that is, an independent waterworks odor element identification branch channel is trained for each odor identification key element. Specifically, the waterworks node odor monitoring dataset is used as training data, support vector machines or random forests are used to build a training odor type identification channel, water quality, operating parameters, and meteorological data corresponding to each key node are used as input features, and the odor type is output; support vector regression is used to build a training concentration level identification channel, water quality, operating parameters, and meteorological data corresponding to each key node are used as input features, and the odor concentration level is output; and support vector regression is used to build a training diffusion range identification channel, water quality, operating parameters, and meteorological data corresponding to each key node are used as input features, and the odor diffusion range is output. Finally, the waterworks odor element identification branch channels are fused in parallel, that is, attention networks are used to automatically calculate the weights of the branches and to adaptively adjust according to real-time operating conditions, and then the waterworks odor identification channel is obtained by weighted integration, odor identification is performed according to the generated process node type and the sensing data corresponding to the node, and the odor type, concentration level, and diffusion range are determined.

[0039] Step S300, according to the spatial distribution and process sequence of the M water plant process key nodes, generate an odor diffusion influence network, based on the odor diffusion influence network, analyze the influence gain of the M initial node odor parameters, and determine the M water plant node odor parameters.

[0040] Step S300 further comprises step S310, based on the spatial distribution and node process sequence of the M water plant process key nodes, performing diffusion path analysis to generate an odor diffusion node path network; step S320, based on the tap water plant node odor monitoring data set, quantifying the diffusion influence of the odor diffusion node path network, and determining a diffusion path influence coefficient set; step S330, mapping the diffusion path influence coefficient set to the odor diffusion node path network for identification, and generating the odor diffusion influence network.

[0041] Preferably, the spatial distribution and process sequence of the M water plant process key nodes are obtained, wherein the spatial distribution refers to the three-dimensional coordinates of the M key nodes, such as the coordinates X = 102.3, Y = 45.6, Z = 0 of the sedimentation tank inlet, and the process sequence refers to the ordered process section number of the water flow direction, such as 1 sedimentation → 2 filtration → 3 disinfection. According to the spatial distribution and node process sequence of the M water plant process key nodes, diffusion path analysis is performed, specifically, according to the process sequence, directed edges are established, such as sedimentation tank outlet → filtration tank inlet, and the topological connection of each node is generated, wherein the vertex represents the key node, the edge represents the water flow path, and a diffusion attribute is attached to each edge, i.e. the diffusion path is labeled, and then the odor diffusion node path network is generated, which is a directed weighted network containing all possible diffusion paths.

[0042] Preferably, based on the tap water plant node odor monitoring data set, the odor diffusion node path network is quantified for diffusion influence, i.e. for each diffusion path, the concentration time series of its upstream and downstream nodes are extracted, and the transfer function is fitted by least squares method to calculate the softmax normalized value of all odor diffusion paths, determine the influence coefficient of each diffusion path, and then compose the diffusion path influence coefficient set; then map the diffusion path influence coefficient set to the odor diffusion node path network for identification, i.e. label the corresponding diffusion path influence coefficient to the corresponding diffusion path edge, obtain the key diffusion path, and finally form the odor diffusion influence network, which has real-time risk visualization capability, such as red path prompting priority control area.

[0043] Further, step S310 further comprises step S311 of extracting node coordinates of the spatial distribution of the M water plant process key nodes to obtain a process node spatial coordinate database; step S312 of connecting the M water plant process key nodes according to the process node spatial coordinate database and the node process sequence to construct a process node topology graph; and step S313 of performing odor diffusion directed identification on each connected node in the process node topology graph to generate the odor diffusion node path network.

[0044] Preferably, based on the spatial distribution of the M water plant process key nodes, three-dimensional coordinates of each water plant process key node are extracted from the tap water plant spatial model to form a process node spatial coordinate database, and then the M water plant process key nodes are connected according to the process node spatial coordinate database and the node process sequence. Specifically, the nodes are connected according to the node process sequence to form directed edges, such as water inlet→grid flocculation→inclined tube sedimentation→V-type filtration→disinfection→water outlet, to construct a process node topology graph and verify the connection rationality. Finally, each connected node in the process node topology graph is identified by odor diffusion directed identification, i.e., a diffusion direction arrow is added to all edges of the process node topology graph, and a diffusion attribute parameter is added to each edge, which can include diffusion time, diffusion attenuation coefficient, risk weight, etc., to identify the odor diffusion path and generate an odor diffusion node path network, i.e., a directed weighted graph with multi-dimensional attributes, to ensure enhanced odor identification accuracy.

[0045] Further, step S300 further comprises step S340 of associating and clustering the tap water plant node odor monitoring dataset with each odor diffusion path in the odor diffusion node path network to obtain a plurality of diffusion path associated odor datasets; step S350 of respectively performing diffusion concentration attenuation fitting quantification based on the plurality of diffusion path associated odor datasets to determine a plurality of path diffusion attenuation coefficients; and step S360 of determining the diffusion path influence coefficient set according to the plurality of path diffusion attenuation coefficients.

[0046] Preferably, the tap water plant node odor monitoring data set is associated and clustered with each odor diffusion path in the odor diffusion node path network, that is, the tap water plant node odor monitoring data is associated with the odor diffusion path, and then the odor diffusion paths are clustered and grouped according to the odor substance types such as algal metabolites and disinfection by-products, for example, a first type of algal odor diffusion path, sedimentation tank→filter tank→effluent, and a second type of chemical odor diffusion path, disinfection tank→clear water tank, and finally a plurality of diffusion path associated odor data sets are output, wherein each odor diffusion path corresponds to a set of spatiotemporally matched concentration sequences. Then, based on the plurality of diffusion path associated odor data sets, diffusion concentration decay fitting quantification is performed respectively, specifically, for each set of diffusion path data, a physical-statistical hybrid model is used to fit the concentration decay law, and the diffusion decay coefficient of each path is calculated by least squares method to represent the decay rate of the odor substance in the path, and then the diffusion decay coefficients of the plurality of paths are determined to comprehensively reflect the combined effects of water flow dynamics, pipeline characteristics, and chemical properties of the substance. Finally, the decay coefficients of all odor diffusion paths are normalized to convert into relative influence weights, and the weights are adjusted according to real-time monitoring data, for example, if high-concentration odor frequently occurs in a certain path recently, the weight of the path is increased, and finally a diffusion path influence coefficient set is generated, including the decay coefficient and the influence weight of each path.

[0047] Further, step S300 further includes step S370 of sequentially performing diffusion influence accumulation on the M initial node odor parameters based on the odor diffusion influence network to obtain M node odor influence accumulation coefficients; and step S380 of performing dynamic gain analysis on the M initial node odor parameters by using the M node odor influence accumulation coefficients to determine the M water plant node odor parameters.

[0048] Preferably, the M initial node odor parameters are analyzed for influence gain based on the odor diffusion influence network, specifically, the M initial node odor parameters are sequentially subjected to diffusion influence accumulation according to the odor diffusion influence network, that is, for each target node, all upstream nodes pointing to it are retrieved, and the cumulative influence of the upstream nodes on the current node is calculated and accumulated according to the diffusion path influence coefficients in the odor diffusion influence network to output M node odor influence accumulation coefficients, that is, to quantify the comprehensive values of the influence of the surrounding diffusion on each node. Then, the M initial node odor parameters are subjected to dynamic gain analysis by using the M node odor influence accumulation coefficients, that is, based on the node odor influence accumulation coefficients, a dynamic gain factor is calculated by using (1+node odor influence accumulation coefficient / threshold value), and then the initial node odor parameters are corrected by using the dynamic gain factor to obtain the M water plant node odor parameters, thereby improving the dynamic adaptability, early warning sensitivity and accuracy of the monitoring and early warning.

[0049] Step S400, constructing a hierarchical early warning mechanism, and using the hierarchical early warning mechanism to match the early warning levels of the M water plant nodes and to perform hierarchical early warning on the odor.

[0050] Preferably, the hierarchical early warning mechanism is constructed to divide the odor risk into different early warning levels and trigger differentiated response measures. Specifically, when the odor parameter of a single node exceeds the baseline value but is lower than 120% of the threshold value, a mild early warning is triggered, such as slight algal growth in a local process section; when multiple nodes are jointly out of the standard, i.e., more than 3 nodes exceed the threshold value by 150% or a key node is continuously out of the standard for 30 minutes, a moderate early warning is triggered, such as accumulation of organic matter due to incomplete backwashing of a filter tank; when a core process node is out of the standard by 200% or a highly toxic substance such as microcystin is detected, a severe early warning is triggered, such as sudden pollution of a water source or failure of a disinfection system; the hierarchical early warning mechanism is used to match the early warning levels of the M water plant nodes, including automatically labeling abnormal parameters and matching the early warning levels through a decision tree model, and outputting hierarchical early warning on the odor, wherein the early warning information includes the out-of-standard substance, the concentration, the influence range, and accompanying treatment measures, such as strengthening the inspection of the process section for mild early warning, starting a standby filter tank and adjusting the dosage of disinfectant for moderate early warning, and switching the water source in an emergency and notifying the health supervision department for severe early warning; thereby significantly improving the accuracy of odor identification and the risk control capability of the water plant.

[0051] In the foregoing, reference is made to Figure 1 The tap water plant odor monitoring and early warning method according to the embodiments of the present application is described in detail. Next, the tap water plant odor monitoring and early warning platform according to the embodiments of the present application will be described with reference to Figure 2 The tap water plant odor monitoring and early warning platform according to the embodiments of the present application is described.

[0052] The tap water plant odor monitoring and early warning platform according to the embodiments of the present application solves the technical problem of poor real-time performance and accuracy of water plant odor monitoring and early warning due to the lack of targetedness of sensor deployment and the disconnection between odor identification and diffusion analysis in the prior art, and achieves the technical effects of improving the real-time performance and coverage of odor monitoring and enhancing the accuracy of odor identification. As shown in Figure 2 The tap water plant odor monitoring and early warning platform includes a water plant perception network construction module 10, a water plant odor identification module 20, an influence gain analysis module 30, and an odor hierarchical early warning module 40.

[0053] The water plant perception network construction module 10 is used for identifying key nodes based on the structure design information of the target water plant, obtaining M water plant process key nodes, and deploying a sensor network on the M water plant process key nodes to construct M water plant perception networks; the water plant odor identification module 20 is used for monitoring and obtaining M water plant node perception data streams by using the M water plant perception networks, building a water plant odor identification channel, identifying odor of the M water plant node perception data streams based on the water plant odor identification channel, and obtaining M initial node odor parameters; the influence gain analysis module 30 is used for generating an odor diffusion influence network according to the spatial distribution and process sequence of the M water plant process key nodes, performing influence gain analysis on the M initial node odor parameters based on the odor diffusion influence network, and determining M water plant node odor parameters; and the odor grading early warning module 40 is used for constructing a grading early warning mechanism, and performing early warning grade matching and odor grading early warning on the M water plant node odor parameters by using the grading early warning mechanism.

[0054] Next, the specific configuration of the water plant perception network construction module 10 will be described in detail. The water plant perception network construction module 10 further includes: parameter extraction on the structure design information of the target water plant to obtain a set of key parameters of the water plant, the set of key parameters of the water plant including spatial layout parameters, process flow parameters, and process equipment parameters; three-dimensional modeling based on the spatial layout parameters and the process equipment parameters to generate a spatial model of the water plant; division into N key process segments according to the process flow parameters; and key point identification on the spatial model of the water plant based on the N key process segments to obtain M water plant process key nodes.

[0055] Next, the specific configuration of the water plant perception network construction module 10 will be described in detail. The water plant perception network construction module 10 further includes: process node association based on the N key process segments and the spatial model of the water plant to obtain a set of N process segment associated nodes; determination of water plant process simulation parameters according to the application scenario of the target water plant; finite element simulation of the spatial model of the water plant by applying the water plant process simulation parameters to obtain odor substance distribution parameters; and key point screening on the set of N process segment associated nodes based on the odor substance distribution parameters to obtain the M water plant process key nodes, where M≥N.

[0056] Next, the specific configuration of the water plant perception network construction module 10 will be described in detail. The water plant perception network construction module 10 further comprises: monitoring demand analysis and sensor selection for the M water plant process key nodes respectively, obtaining M process node sensor parameters; based on the odor substance distribution parameters, deploying the location of the M water plant process key nodes to determine the M process node deployment location parameters; generating M node sensor deployment schemes according to the M process node sensor parameters and the M process node deployment location parameters; deploying sensor networks on the M water plant process key nodes according to the M node sensor deployment schemes to construct the M water plant perception networks.

[0057] Next, the specific configuration of the water plant odor identification module 20 will be described in detail. The water plant odor identification module 20 further comprises: based on the target self-supply water plant, performing historical odor data mining and generating process node identification to obtain a self-supply water plant node odor monitoring dataset; obtaining odor identification key elements, including odor type, concentration level, and diffusion range; using the odor identification key elements to perform classification and identification training on the self-supply water plant node odor monitoring dataset to obtain a water plant odor element identification branch channel; parallel fusion of the water plant odor element identification branch channel to build the water plant odor identification channel.

[0058] Next, the specific configuration of the influence gain analysis module 30 will be described in detail. The influence gain analysis module 30 further comprises: based on the spatial distribution and node process sequence of the M water plant process key nodes, performing diffusion path analysis to generate an odor diffusion node path network; based on the self-supply water plant node odor monitoring dataset, performing diffusion influence quantification on the odor diffusion node path network to determine a diffusion path influence coefficient set; mapping the diffusion path influence coefficient set to the odor diffusion node path network for identification to generate the odor diffusion influence network.

[0059] Next, the specific configuration of the influence gain analysis module 30 will be described in detail. The influence gain analysis module 30 further comprises: node coordinate extraction of the spatial distribution of the M water plant process key nodes to obtain a process node spatial coordinate database; process topology connection of the M water plant process key nodes according to the process node spatial coordinate database and the node process sequence to construct a process node topology graph; odor diffusion directional identification of each connected node in the process node topology graph to generate the odor diffusion node path network.

[0060] Next, the specific configuration of the influence gain analysis module 30 will be described in detail. The influence gain analysis module 30 further comprises: associating and clustering the tap water plant node odor monitoring data set with each odor diffusion path in the odor diffusion node path network, obtaining a plurality of diffusion path associated odor data sets; based on the plurality of diffusion path associated odor data sets, respectively performing diffusion concentration decay fitting quantification to determine a plurality of path diffusion decay coefficients; and determining the diffusion path influence coefficient set according to the plurality of path diffusion decay coefficients.

[0061] Next, the specific configuration of the influence gain analysis module 30 will be described in detail. The influence gain analysis module 30 further comprises: based on the odor diffusion influence network, sequentially performing diffusion influence accumulation on the M initial node odor parameters to obtain M node odor influence cumulative coefficients; and performing dynamic gain analysis on the M initial node odor parameters through the M node odor influence cumulative coefficients to determine the M tap water plant node odor parameters.

[0062] The tap water plant odor monitoring and early warning platform provided in the embodiments of the present application can execute the tap water plant odor monitoring and early warning method provided in any of the embodiments of the present application, and has the corresponding function modules and beneficial effects of the execution method.

[0063] Although the present application makes various references to certain modules in the platform according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server, and each unit and module included is only divided according to the functional logic, but is not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy mutual differentiation, and do not limit the protection scope of the present application.

[0064] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for monitoring and early warning of odor in a waterworks, characterized in that, The method comprises: Based on the structure design information of the target waterworks, key nodes are identified to obtain M waterworks process key nodes, and a sensor network is deployed on the M waterworks process key nodes to construct M waterworks perception networks; M waterworks node perception data streams are monitored and obtained by using the M waterworks perception networks, a waterworks odor identification channel is built, odor identification is performed on the M waterworks node perception data streams based on the waterworks odor identification channel, and M initial node odor parameters are obtained; Based on the spatial distribution and process sequence of the M waterworks process key nodes, an odor diffusion influence network is generated, influence gain analysis is performed on the M initial node odor parameters based on the odor diffusion influence network, and M waterworks node odor parameters are determined; A hierarchical early warning mechanism is constructed, and the M waterworks node odor parameters are matched with early warning levels and subjected to odor hierarchical early warning by using the hierarchical early warning mechanism; The generation of the odor diffusion influence network comprises: Based on the spatial distribution and node process sequence of the M waterworks process key nodes, diffusion path analysis is performed to generate an odor diffusion node path network; Based on a waterworks node odor monitoring data set, diffusion influence quantification is performed on the odor diffusion node path network to determine a diffusion path influence coefficient set; The diffusion path influence coefficient set is mapped to the odor diffusion node path network for identification to generate the odor diffusion influence network; The generation of the odor diffusion node path network comprises: Node coordinates of the spatial distribution of the M waterworks process key nodes are extracted to obtain a process node spatial coordinate database; The M waterworks process key nodes are connected in a process topology according to the process node spatial coordinate database and the node process sequence to construct a process node topology graph; Each connected node in the process node topology graph is identified in a direction of odor diffusion to generate the odor diffusion node path network; The determination of the diffusion path influence coefficient set comprises: The waterworks node odor monitoring data set is associated and clustered with each odor diffusion path in the odor diffusion node path network to obtain a plurality of diffusion path associated odor data sets; Based on the plurality of diffusion path associated odor data sets, diffusion concentration decay fitting quantification is respectively performed to determine a plurality of path diffusion decay coefficients; Based on the plurality of path diffusion decay coefficients, the diffusion path influence coefficient set is determined; The determination of the M waterworks node odor parameters comprises: Based on the odor diffusion influence network, diffusion influence accumulation is sequentially performed on the M initial node odor parameters to obtain M node odor influence cumulative coefficients; Dynamic gain analysis is performed on the M initial node odor parameters by using the M node odor influence cumulative coefficients to determine the M waterworks node odor parameters, wherein the M waterworks node odor parameters are calculated by using a dynamic gain factor of 1+node odor influence cumulative coefficient / threshold value, and the initial node odor parameters are corrected by using the dynamic gain factor.

2. The method for monitoring and early warning of odor in waterworks according to claim 1, characterized in that, The generation of the M waterworks process key nodes comprises: Parameter extraction is performed on the structural design information of the target waterworks to obtain a waterworks key parameter set, the waterworks key parameter set including a spatial layout parameter, a treatment process flow parameter, and a process equipment parameter; Three-dimensional modeling is performed based on the spatial layout parameter and the process equipment parameter to generate a waterworks spatial model; According to the treatment process flow parameter, N key treatment process sections are divided; Based on the N key treatment process sections, key point identification is performed on the waterworks spatial model to obtain M waterworks process key nodes.

3. The method for monitoring and early warning of odor in waterworks according to claim 2, characterized in that, The key point identification based on the N key treatment process sections on the waterworks spatial model to obtain M waterworks process key nodes includes: Based on the N key treatment process sections, process node association is performed with the waterworks spatial model to obtain an N process section associated node set; According to the application scenario of the target waterworks, waterworks treatment simulation parameters are determined; The waterworks treatment simulation parameters are applied to the waterworks spatial model for finite element simulation to obtain odor substance distribution parameters; Based on the odor substance distribution parameters, key point screening is performed on the N process section associated node set to obtain the M waterworks process key nodes, where M≥N.

4. The method for monitoring and early warning of odor in waterworks according to claim 3, characterized in that, The construction of M waterworks perception networks includes: Monitoring demand analysis and sensor selection are respectively performed on the M waterworks process key nodes to obtain M process node sensor parameters; Based on the odor substance distribution parameters, deployment location analysis is performed on the M waterworks process key nodes to determine M process node deployment location parameters; According to the M process node sensor parameters and the M process node deployment location parameters, M node sensor deployment schemes are generated; According to the M node sensor deployment schemes, a sensor network is deployed on the M waterworks process key nodes to construct the M waterworks perception networks.

5. The method for monitoring and early warning of odor in waterworks according to claim 1, wherein The construction of a waterworks odor identification channel includes: Based on the target waterworks, historical odor data mining and process node identification are performed to obtain a waterworks node odor monitoring data set; Odor identification key elements are obtained, including odor type, concentration level, and diffusion range; The waterworks node odor monitoring data set is classified and identified using the odor identification key elements to obtain a waterworks odor element identification branch channel; The waterworks odor element identification branch channel is fused in parallel to construct the waterworks odor identification channel.

6. A waterworks odor monitoring and early warning platform, characterized in that, The platform is used to implement the waterworks odor monitoring and early warning method of any one of claims 1 to 5, and the platform includes: A waterworks perception network construction module is used to identify key nodes based on the structural design information of a target waterworks to obtain M waterworks process key nodes, and a sensor network is deployed on the M waterworks process key nodes to construct M waterworks perception networks; The water plant odor identification module is used to monitor and acquire the sensing data streams of M water plant nodes using the M water plant sensing networks, build a water plant odor identification channel, and perform odor identification on the sensing data streams of the M water plant nodes based on the water plant odor identification channel to obtain M initial node odor parameters. The influence-gain analysis module is used to generate an odor diffusion influence network based on the spatial distribution and process sequence of the M key nodes of the water plant process, and to perform influence-gain analysis on the odor parameters of the M initial nodes based on the odor diffusion influence network to determine the odor parameters of the M water plant nodes. The odor classification and early warning module is used to construct a classification and early warning mechanism, and to perform early warning level matching and odor classification and early warning for the odor parameters of the M water plant nodes using the classification and early warning mechanism.

Images