Water tower wall condensate water quality detection method and system

Abstract

The application discloses a kind of water tower wall condensate water quality detection method and system, it is related to industrial water quality detection technical field, including obtaining target water tower real-time running environment data set, calculating the temperature gradient of each monitoring area in inner wall, combining humidity to predict condensate water active condensation region, in the region arrangement micro water quality sampling array timing collection water sample, after physical pretreatment, through multichannel parallel spectrum analysis device synchronous acquisition ultraviolet absorption, near infrared transmission and Raman scattering spectrum data, to multidimensional spectrum data fusion analysis and extract feature, generate multidimensional water quality feature vector, construct space water quality image.This method can match condensate water actual generation area sampling, through multispectral fusion analysis complete characterization water quality information, realize the accurate detection of water tower wall condensate water quality and space distribution image construction.

Description

Technical Field

[0001] This invention belongs to the field of industrial water quality testing technology, specifically a method and system for testing the water quality of condensate from the walls of a water tower. Background Technology

[0002] Current methods for testing the condensate water quality of water tower walls mostly involve collecting water samples by deploying sampling devices at fixed locations. The testing process relies on a single type of spectral detection equipment to analyze the water samples, which can only obtain detection data corresponding to a single spectrum. Data processing only involves basic extraction and recording of the single detection data, without combining the real-time operating environmental parameters of the water tower to predict the location of condensate generation, and without using a method of parallel detection of multiple types of spectra and multi-dimensional data fusion analysis to complete the water quality testing process.

[0003] Current detection methods do not integrate environmental parameters such as temperature at multiple points on the inner wall of the water tower, internal air humidity, external atmospheric pressure, and internal airflow velocity distribution. This makes it impossible to accurately determine the actual area of ​​condensate formation on the tower wall surface. Sampling points cannot match the condensate formation area, and the collected water samples do not accurately correspond to the core condensate formation location, resulting in insufficient sample representativeness. Single-spectral detection can only obtain local water quality characterization information and cannot comprehensively cover the water quality characteristics of condensate. Multi-dimensional spectral data cannot be acquired and fused simultaneously, making it impossible to extract multi-dimensional water quality features and construct a spatial water quality profile. Therefore, the detection results cannot fully reflect the true water quality state of the condensate on the tower wall.

[0004] This invention aims to solve the problem of being unable to dynamically predict the active condensation area of ​​condensate based on the real-time operating environment parameters of the water tower and to carry out targeted sampling, and to solve the problem of being unable to obtain multi-dimensional spectral data through multi-channel parallel spectral detection and to complete data fusion analysis, feature extraction and spatial water quality profile construction. Summary of the Invention

[0005] This invention proposes a method for detecting the water quality of condensate on the walls of a water tower, comprising: acquiring a real-time operating environment dataset of the target water tower, the real-time operating environment dataset including temperature readings at multiple points on the inner wall of the water tower, ambient air humidity inside the water tower, atmospheric pressure outside the water tower, and airflow velocity distribution inside the water tower; using the temperature readings at multiple points on the inner wall of the water tower, calculating the temperature gradient of each monitoring area on the inner wall of the water tower, and dynamically predicting active condensation areas formed by condensate on the surface of the inner wall of the water tower in conjunction with the ambient air humidity inside the water tower; deploying a micro water quality sampling array in the active condensation area, activating the micro water quality sampling array, and sampling the active condensation area. Condensate is collected through timed adsorption to form a raw condensate water sample set. This raw condensate water sample set undergoes preliminary physical processing to obtain a pre-treated condensate water sample set. The pre-treated condensate water sample set is then input into a multi-channel parallel spectral analyzer, which simultaneously acquires the ultraviolet absorption spectrum, near-infrared transmission spectrum, and Raman scattering spectrum of the pre-treated condensate water sample set, generating multi-dimensional raw spectral data. This multi-dimensional raw spectral data is then fused and analyzed for feature extraction to generate a multi-dimensional water quality feature vector for each condensate water sample. Finally, a spatial water quality profile is constructed based on the multi-dimensional water quality feature vector.

[0006] Furthermore, using the temperature readings at multiple points on the inner wall of the water tower, the temperature gradient of each monitoring area on the inner wall of the water tower is calculated. Combined with the humidity of the ambient air inside the water tower, the active condensation area formed by condensate on the surface of the inner wall of the water tower is dynamically predicted. This includes: constructing a digital model of the temperature field of the inner wall of the water tower based on the spatial distribution of the temperature readings at multiple points on the inner wall of the water tower; performing spatial differentiation on the digital model of the temperature field of the inner wall of the water tower to obtain the rate of change of the normal temperature of the wall surface corresponding to each coordinate point on the inner wall of the water tower, i.e., the temperature gradient; reading the current air dew point temperature from the real-time updated ambient air humidity data inside the water tower; marking all coordinate points in the digital model of the temperature field of the inner wall of the water tower where the actual wall temperature is lower than the current air dew point temperature to form a preliminary set of condensation points; selecting coordinate points in the preliminary set of condensation points where the absolute value of the temperature gradient exceeds the condensation intensity threshold, and defining the physical area corresponding to the selected coordinate points as the active condensation area.

[0007] Further, the micro water quality sampling array is activated to periodically adsorb and collect condensate from the active condensation region, forming a raw condensate water sample set. This includes: the micro water quality sampling array is composed of multiple capillary adsorption micro sampling heads; each capillary adsorption micro sampling head in the micro water quality sampling array is assigned a sub-region within the active condensation region to be collected; all capillary adsorption micro sampling heads are controlled to extend simultaneously during the predicted peak condensation period, so that the tip of the sampling head contacts the condensate film on the tower wall, and continuously adsorbs a specific volume of condensate water through capillary force; after reaching the preset single sampling volume, all capillary adsorption micro sampling heads are controlled to retract into the anti-contamination sealed cavity; a micro metering pump located in the anti-contamination sealed cavity is driven to transfer the specific volume of condensate water adsorbed by each capillary adsorption micro sampling head to the corresponding independently numbered sample storage tube; at the end of the sampling cycle, the condensate water samples in all independently numbered sample storage tubes are collected to form the raw condensate water sample set containing spatiotemporal location information.

[0008] Further, the original condensate water sample set undergoes preliminary physical treatment to obtain a pretreated condensate water sample set, including: the preliminary physical treatment includes removing solid suspended particles through microporous filtration and removing dissolved gases through low-temperature constant-temperature settling; each sample in the original condensate water sample set is sequentially passed through an inorganic ceramic microporous filter membrane with a constant pore size to retain solid suspended particles with a particle size larger than a set threshold, obtaining a filtered water sample; the filtered water sample is placed in a temperature-controlled transparent sample cell, and under light-protected conditions, the temperature of the transparent sample cell is lowered and stabilized at a predetermined temperature. A low-temperature isothermal point is set; at the low-temperature isothermal point, the filtered water sample in the transparent sample cell is allowed to stand for a continuous period of time, allowing the gas dissolved in the water to slowly precipitate due to changes in solubility; a constant micro-negative pressure environment lower than the ambient pressure is maintained in the upper space of the transparent sample cell to accelerate and guide the precipitated gas to detach from the liquid surface; after the standing process is completed, the liquid that has undergone gas removal treatment is extracted from the middle of the transparent sample cell and injected into a new clean sample tube to complete the preliminary physical treatment of a single sample. All samples undergoing the same treatment constitute the pretreated condensate sample set.

[0009] Further, the pretreated condensate sample set is input into a multi-channel parallel spectroscopic analysis device. This device simultaneously acquires the ultraviolet absorption spectrum, near-infrared transmission spectrum, and Raman scattering spectrum of the pretreated condensate sample set, generating multidimensional raw spectral data. This includes: dividing a single sample from the pretreated condensate sample set into three sub-samples; introducing the first sub-sample into the ultraviolet spectral channel of the multi-channel parallel spectroscopic analysis device, and scanning and recording the absorbance curve of the sample in the ultraviolet band as a function of wavelength under continuous-wave ultraviolet light irradiation, i.e., the ultraviolet absorption spectrum; and introducing the second sub-sample into the near-infrared spectral channel of the multi-channel parallel spectroscopic analysis device. Under near-infrared broadband light source illumination, the curve of light intensity transmitted through the sample changing with wavelength is detected and recorded, and the near-infrared transmission spectrum is calculated by comparing it with a blank control. The third sample is introduced into the Raman spectroscopy channel of the multi-channel parallel spectral analysis device, and the sample is excited using a monochromatic laser. The curve of Raman scattering light intensity generated by the sample changing with Raman shift is collected and recorded, which is the Raman scattering spectrum. The ultraviolet absorption spectrum, the near-infrared transmission spectrum, and the Raman scattering spectrum from the same sample are integrated and aligned according to a preset data structure to form a multidimensional original spectral data package that uniquely corresponds to the same sample. The data packages of all samples together constitute the multidimensional original spectral data.

[0010] Further, the multidimensional raw spectral data is fused and analyzed, and features are extracted to generate a multidimensional water quality feature vector for each condensate sample. This includes: extracting the characteristic absorption peak position and peak area of ​​the ultraviolet absorption spectrum, the characteristic transmission valley position and valley depth of the near-infrared transmission spectrum, and the characteristic Raman peak position and peak intensity of the Raman scattering spectrum for each multidimensional raw spectral data package; mapping and associating the characteristic absorption peak position, characteristic transmission valley position, and characteristic Raman peak position from the same sample, searching a pre-set "spectral feature-water quality parameter" association knowledge base, and initially identifying the ion types, organic functional group categories, and colloidal substance types; normalizing the characteristic absorption peak area, characteristic transmission valley depth, and characteristic Raman peak intensity from the same sample, and inputting them into a trained concentration inversion model to calculate preliminary concentration estimates of the ion types, organic functional group categories, and colloidal substance types; and fusing the preliminary identification results and the preliminary concentration estimates to generate a multidimensional water quality feature vector for each condensate sample.

[0011] Furthermore, a spatial water quality profile is constructed based on the multi-dimensional water quality feature vectors, including: reading the spatiotemporal location information recorded at the time of collection for each condensate sample, the spatiotemporal location information including the three-dimensional coordinates of the sampling point on the inner wall of the water tower and the sampling time point; establishing a three-dimensional spatial grid model of the inner wall of the water tower, and assigning the multi-dimensional water quality feature vectors corresponding to each condensate sample to the grid nodes of the corresponding coordinates in the three-dimensional spatial grid model; for grid nodes not directly covered by the sample, using a distance-weighted spatial interpolation algorithm, calculating their values ​​using the multi-dimensional water quality feature vectors of neighboring nodes to complete the filling of the entire three-dimensional spatial grid model; based on the fully filled three-dimensional spatial grid model, generating a spatial distribution contour map of the concentration of each water quality component on the surface of the inner wall of the water tower; integrating the spatial distribution contour maps of the concentration of all water quality components and overlaying them with the structural map of the inner wall of the water tower to construct the spatial water quality profile reflecting the spatial heterogeneity of water quality parameters.

[0012] Furthermore, the method also includes a dynamic correlation analysis step based on the spatial water quality profile and the real-time operating environment dataset: extracting the airflow velocity distribution data inside the water tower corresponding to the collection time point of the condensate sample from the real-time operating environment dataset; spatially overlaying the outline of the high-concentration area of ​​specific pollutants in the spatial water quality profile with the airflow velocity distribution map inside the water tower at the same time; analyzing the spatial fit between the outline of the high-concentration area of ​​specific pollutants and the low-speed airflow area, vortex area, or airflow impact wall area, and calculating the spatial overlap coefficient; performing time series correlation analysis on the historical change curves of temperature readings at multiple points on the inner wall of the water tower and the specific ion concentration change curves at the corresponding points, and calculating the time-delay correlation coefficient; recording the spatial overlap coefficient and the time-delay correlation coefficient to form a dynamic correlation dataset for explaining the causes of spatial distribution of water quality.

[0013] Furthermore, the method also includes an abnormal condensate water quality early warning step based on the dynamic correlation dataset: setting a concentration safety threshold and a spatial distribution uniformity threshold for key water quality parameters; comparing the actual concentration values ​​and actual spatial distributions of each water quality parameter in the spatial water quality profile generated in the current period with the concentration safety threshold and the spatial distribution uniformity threshold; triggering a water quality anomaly marker when the actual concentration value exceeds the concentration safety threshold or the non-uniformity of the actual spatial distribution exceeds the spatial distribution uniformity threshold; retrieving the dynamic correlation dataset related to the spatial region and time point corresponding to the water quality anomaly marker, analyzing the main environmental factors causing the anomaly; and combining the main environmental factors with the water quality anomaly marker to generate an abnormal condensate water quality early warning report containing the anomaly location, anomaly parameters, anomaly degree, and associated environmental conditions.

[0014] Furthermore, based on the spatial distribution of the multi-point temperature readings on the inner wall of the water tower, a digital model of the temperature field of the inner wall of the water tower is constructed, including: based on the multi-point temperature readings obtained by the sensor network deployed on the surface of the inner wall of the water tower, calculating the temperature estimate of the area on the surface of the inner wall of the water tower where no sensors are directly deployed, using a spatial interpolation algorithm; and fusing the measured temperature values ​​of the multi-point temperature readings with the temperature estimate to generate a continuous digital model of the temperature field of the inner wall of the water tower.

[0015] Furthermore, after generating the abnormal condensate water quality early warning report, the system also includes a response control step based on the early warning: sending the abnormal condensate water quality early warning report to the water tower operation control system; the water tower operation control system generates and executes targeted cleaning or control commands based on the abnormal location and associated environmental conditions in the abnormal condensate water quality early warning report.

[0016] Furthermore, the micro water quality sampling array is an online array that can move along the guide rails on the tower wall; the timed adsorption and collection of condensate from the active condensation zone, the preliminary physical treatment of the original condensate water sample set, and the input of the pretreated condensate water sample set into a multi-channel parallel spectral analysis device are all performed automatically and continuously on an integrated online detection platform.

[0017] Furthermore, the present invention also includes a water tower wall condensate water quality detection system, the system including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein when the processor executes the computer program, it implements the steps of the water tower wall condensate water quality detection method described above.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] The temperature gradient of each monitoring area is calculated based on the temperature readings at multiple points on the inner wall of the water tower. Combined with the dynamic prediction of the active condensation area of ​​condensate on the inner wall surface of the water tower by the ambient air humidity inside the water tower, a micro water quality sampling array is set up in the predicted active condensation area and timed adsorption collection is started. This can ensure that the sampling points are highly correlated with the actual condensate generation area. The water sampling process closely matches the actual state of dynamic condensate generation. The original water sample set can completely cover the core area of ​​condensate generation, avoiding the situation where fixed sampling points do not match the actual condensation area. This makes the collected water samples more closely match the real distribution of condensate on the tower wall, and the targeting and authenticity of the water samples are enhanced. The original water sample set can completely reflect the condensate state in the active condensation area.

[0020] By simultaneously acquiring the ultraviolet absorption spectrum, near-infrared transmission spectrum, and Raman scattering spectrum of pretreated condensate samples using a multi-channel parallel spectral analysis device, multi-dimensional raw spectral data is formed. Fusion analysis and feature extraction are then performed on this multi-dimensional raw spectral data to integrate water quality characterization information corresponding to different spectra, generating a multi-dimensional water quality feature vector for each condensate sample. Based on this multi-dimensional water quality feature vector, a spatial water quality profile is constructed, enabling a complete characterization of the condensate's water quality attributes from multiple dimensions. The detection information from different types of spectra can complement each other, and multi-dimensional data fusion processing eliminates the information limitations of single-spectrum detection. The multi-dimensional water quality feature vector can fully carry the water quality parameter information of the water sample, and the spatial water quality profile can intuitively present the water quality distribution of condensate in different areas of the tower wall, making the water quality detection results more comprehensive and complete. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the steps of a method for detecting the quality of condensate water on the wall of a water tower, as described in this invention.

[0022] Figure 2 This is a flowchart for the preliminary physical treatment of the original condensate water sample collection.

[0023] Figure 3 This is the ultraviolet absorption spectrum of the condensate sample.

[0024] Figure 4 This is a time-series analysis diagram of multiple parameters in the real-time operating environment of the water tower.

[0025] Figure 5 This is a graph showing the correlation between temperature changes and the time series of ion concentrations. Detailed Implementation

[0026] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0027] See Figure 1This invention provides a method for detecting the water quality of condensate on the walls of a water tower. The method includes: acquiring a real-time operating environment dataset of the target water tower, which includes temperature readings at multiple points on the inner wall of the water tower, ambient air humidity inside the water tower, atmospheric pressure outside the water tower, and airflow velocity distribution inside the water tower. This data is continuously collected and transmitted to a central processing unit via a sensor network installed inside and outside the water tower. Using the temperature readings at multiple points on the inner wall of the water tower, the temperature gradient of each monitored area on the inner wall is calculated. This calculation is based on spatial interpolation and differential algorithms of the temperature readings, combined with ambient air humidity data inside the water tower, to dynamically predict active condensation areas formed by condensate on the inner wall surface of the water tower. The prediction process is completed by comparing the wall temperature with the air dew point temperature. A miniature water quality sampling array is deployed in the active condensation area. This array consists of multiple miniature sampling heads. After the miniature water quality sampling array is activated, the condensate in the active condensation area is periodically adsorbed and collected. The collection process is automatically triggered according to a preset time interval, forming a collection of raw condensate water samples. Each sample is accompanied by collection time and location information. The raw condensate water samples underwent preliminary physical treatment, including filtration and degassing, to obtain a pretreated condensate water sample set, which was stored in a low-temperature environment to maintain stability. The pretreated condensate water sample set was then input into a multi-channel parallel spectroscopic analysis device equipped with UV, NIR, and Raman spectral channels. This device simultaneously acquired the UV absorption spectrum, NIR transmission spectrum, and Raman scattering spectrum of the pretreated condensate water sample set, generating multi-dimensional raw spectral data, which was stored in digital format. The multi-dimensional raw spectral data underwent fusion analysis and feature extraction. The fusion analysis employed data alignment and feature matching algorithms to generate a multi-dimensional water quality feature vector for each condensate water sample, containing ion concentration and organic matter information. A spatial water quality profile was constructed based on the multi-dimensional water quality feature vectors, using spatial interpolation and visualization techniques to generate a water quality distribution map of the water tower's inner wall.

[0028] In one embodiment of the present invention, a digital model of the water tower's inner wall temperature field is constructed based on the spatial distribution of temperature readings at multiple points on the inner wall. This model represents temperature values ​​using a three-dimensional coordinate grid. Spatial differentiation is performed on the digital model of the water tower's inner wall temperature field using the finite difference method to obtain the rate of change of the wall's normal temperature at each coordinate point, i.e., the temperature gradient. The gradient values ​​are stored in vector form. The current air dew point temperature is read from the real-time updated ambient air humidity data inside the water tower, calculated using humidity sensor data. In the digital model of the water tower's inner wall temperature field, all coordinate points whose actual wall temperature is lower than the current air dew point temperature are marked, forming a preliminary set of condensation points. The marking process is achieved by comparing temperature values ​​with a dew point temperature threshold. From the preliminary set of condensation points, coordinate points whose absolute temperature gradient value exceeds a condensation intensity threshold are selected. The condensation intensity threshold is set based on historical data. The physical area corresponding to the selected coordinate points is defined as an active condensation region, which is used to guide the arrangement of the sampling array.

[0029] In specific implementation, a digital model of the water tower's inner wall temperature field is constructed based on the spatial distribution of temperature readings from multiple points on the inner wall. These readings are obtained from an array of temperature sensors installed on the surface of the inner wall. The temperature sensor array is arranged in a regular grid, with each sensor recording its three-dimensional coordinates and real-time temperature value. The digital model of the water tower's inner wall temperature field represents the temperature values ​​through a three-dimensional coordinate grid. The nodes of the three-dimensional coordinate grid correspond to the positions of the temperature sensors. For grid nodes where no temperature sensors are directly installed, the temperature value is calculated using a spatial interpolation algorithm. This algorithm employs an inverse distance weighting method, which is a weighted average based on readings from neighboring temperature sensors. The weight of the weighted average is inversely proportional to the distance. In some embodiments, the digital model of the water tower's inner wall temperature field is stored in the memory of a digital processing system. The digital processing system periodically reads data from the temperature sensor array and updates the digital model of the water tower's inner wall temperature field. The update frequency is synchronized with the data acquisition frequency. The digital model of the water tower's inner wall temperature field is organized in matrix form, with the matrix's row and column indices corresponding to the two-dimensional surface coordinates of the grid nodes, and the matrix elements storing the temperature values. Spatial differentiation is performed on the digital model of the temperature field inside the water tower using the finite difference method. The finite difference method calculates partial derivatives based on the temperature values ​​at the grid nodes, obtaining the rate of change of the wall's normal temperature at each coordinate point, i.e., the temperature gradient. The temperature gradient is stored as a vector, with each component representing the rate of change of temperature along the coordinate axis. In practice, the temperature gradient is calculated using the following formula:

[0030]

[0031] in: Represents grid nodes The temperature gradient vector at that point, Represents grid nodes Temperature value at that location, and These represent the grid in and Spacing in the direction, and The unit vector represents the direction of the coordinate axes, and the magnitude of the temperature gradient vector represents the rate of temperature change. The current air dew point temperature is read from the real-time updated ambient air humidity data inside the water tower. The ambient air humidity data comes from a humidity sensor installed inside the water tower. The humidity sensor continuously measures relative humidity and air temperature. The current air dew point temperature is calculated from the humidity sensor data using a dew point temperature formula based on relative humidity and air temperature. The dew point temperature formula is a known physical relationship.

[0032] In practice, the digital processing system acquires real-time relative humidity and air temperature from humidity sensors, calculates the current air dew point temperature using the dew point temperature formula, and stores the current air dew point temperature in numerical form, aligned with the timestamp of the digital model of the water tower's inner wall temperature field. All coordinate points in the digital model of the water tower's inner wall temperature field that have an actual wall temperature lower than the current air dew point temperature are marked. This marking process is achieved by comparing the temperature value of each grid node in the digital model with the current air dew point temperature. The comparison operation is performed within the digital processing system. For each grid node, if the stored temperature value is lower than the current air dew point temperature, the node is marked as a potential condensation point. All marked nodes form a preliminary condensation point set, which is stored in list form, containing the coordinates and temperature value of each node. In some embodiments, the marking operation is completed by traversing all grid nodes of the digital model of the temperature field on the inner wall of the water tower. During the traversal, the digital processing system compares the temperature value of each node with the current air dew point temperature threshold, which is the calculated current air dew point temperature value. When the temperature value is lower than the threshold, the node coordinates are added to the preliminary condensation point set. Coordinates with an absolute temperature gradient exceeding the condensation intensity threshold are selected from the preliminary condensation point set. The condensation intensity threshold is set based on historical data, which comes from temperature gradients and condensation observation records accumulated during the operation of the water tower. The condensation intensity threshold is a preset value representing the minimum rate of temperature change required to trigger active condensation. The selection process involves reading the temperature gradient value corresponding to each coordinate point from the preliminary condensation point set. The temperature gradient value is retrieved from the stored gradient vector, and the magnitude of the temperature gradient vector is compared with the condensation intensity threshold. If the magnitude of the temperature gradient vector is greater than the condensation intensity threshold, the coordinate point is retained. It can be understood that the screening operation in the digital processing system is implemented through iterative comparison. Each coordinate point in the initial condensation point set is iterated through, the magnitude of the corresponding temperature gradient vector is calculated, and the value is compared with the condensation intensity threshold. Coordinate points that meet the conditions are collected into a new list. Optionally, the condensation intensity threshold can be dynamically adjusted according to the water tower material or environmental conditions. This dynamic adjustment is based on real-time operating data; for example, when the airflow velocity distribution inside the water tower changes, the condensation intensity threshold is corrected accordingly. The physical region corresponding to the screened coordinate points is defined as the active condensation region. The active condensation region is represented by the set of screened coordinate points in physical space. The digital processing system maps these coordinate points to their actual positions on the inner wall of the water tower, forming continuous or discrete region contours. These region contours are used to guide the layout of the micro water quality sampling array. In specific implementation, the active condensation region is defined by connecting the screened coordinate points. The connection algorithm is based on the distance between adjacent coordinate points. When the distance between coordinate points is less than a set value, these coordinate points are considered to belong to the same region, thus generating the boundary polygon of the active condensation region. The boundary polygon is stored in the form of a coordinate sequence.It is understandable that the output format of the active condensation region includes the region center coordinates and a spatial extent description. The spatial extent description is provided through a minimum bounding box or a list of polygon vertices. This information is transmitted to the control system of the micro water quality sampling array, which adjusts the deployment position of the sampling head according to the coordinates of the active condensation region.

[0033] In one embodiment of the present invention, the micro water quality sampling array comprises multiple capillary adsorption micro-sampling heads, each connected to an independent drive mechanism. Each capillary adsorption micro-sampling head in the micro water quality sampling array is assigned a sub-region within the active condensation area to be sampled, an assignment based on a spatial partitioning algorithm. All capillary adsorption micro-sampling heads are controlled to extend simultaneously during the predicted peak condensation period. The extension is driven by a motor, bringing the tip of the sampling head into contact with the condensate film on the tower wall. Through capillary force, a specific volume of condensate is continuously adsorbed, with the adsorption time adjusted according to the flow rate. After reaching the preset single sampling volume, all capillary adsorption micro-sampling heads are controlled to retract into a contamination-proof sealed cavity, preventing sample contamination during the retraction process. A micro-quantitative pump located within the contamination-proof sealed cavity is driven to transfer the specific volume of condensate adsorbed by each capillary adsorption micro-sampling head to a corresponding independently numbered sample storage tube, maintaining a sealed transfer process. At the end of the sampling period, condensate samples from all individually numbered sample storage tubes are collected to form a raw condensate water sample set containing spatiotemporal location information derived from the sampling head number. Preliminary physical treatment includes removing suspended solid particles through microfiltration and removing dissolved gases through low-temperature isothermal settling; this treatment is performed in specialized equipment. (See also...) Figure 2 Each sample from the original condensate water sample set is sequentially passed through an inorganic ceramic microporous filter membrane with a constant pore size to trap solid suspended particles larger than a set threshold, resulting in a filtered water sample. The filter membrane is replaced periodically. The filtered water sample is placed in a temperature-controlled transparent sample cell. Under light-protected conditions, the temperature of the transparent sample cell is lowered and stabilized at a preset low-temperature isothermal point. Temperature control is achieved through a thermoelectric module. At the low-temperature isothermal point, the filtered water sample in the transparent sample cell is allowed to stand for an extended period, allowing dissolved gases to slowly precipitate due to changes in solubility. The standing time is set according to the type of gas. A constant micro-negative pressure environment, lower than ambient pressure, is maintained in the upper space of the transparent sample cell to accelerate and guide the precipitated gases to detach from the liquid surface. This micro-negative pressure is generated by a vacuum pump. After the standing process, the liquid that has undergone gas removal treatment is extracted from the middle of the transparent sample cell and injected into a new clean sample tube, completing the initial physical treatment of the individual sample. All samples undergo the same treatment to form a pre-treated condensate water sample set.

[0034] In practical implementation, the micro water quality sampling array consists of multiple capillary adsorption micro-sampling heads. Each capillary adsorption micro-sampling head includes a hollow microtube structure with a hydrophilic coating at the tip to enhance capillary force. Each capillary adsorption micro-sampling head is connected to an independent drive mechanism, which consists of a micro stepper motor and a linear guide. The micro stepper motor controls the extension and retraction of the capillary adsorption micro-sampling head. For each capillary adsorption micro-sampling head in the micro water quality sampling array, a sub-region within the active condensation area is assigned to it. This assignment is based on a spatial partitioning algorithm. The spatial partitioning algorithm takes the digital contour of the active condensation area as input. The digital contour is defined by a series of boundary coordinate points. The algorithm divides the digital contour into multiple sub-regions of approximately equal area. Each sub-region establishes a one-to-one correspondence with a capillary adsorption micro-sampling head, and the assignment results are stored as a mapping table. In some embodiments, the sub-regions are divided using the Thiessen polygon method, which is based on a set of seed points distributed within the active condensation region. Each seed point corresponds to a preset installation position of a capillary micro-sampling head. The Thiessen polygon method ensures that the distance from any point within each sub-region to its corresponding seed point is less than the distance to other seed points. The boundary coordinates of the sub-regions are sent to the corresponding drive mechanism. All capillary micro-sampling heads are controlled to extend simultaneously during the predicted condensation peak period, which is output by the prediction model. The drive mechanism receives a synchronization command, and a micro-stepping motor drives a linear guide to move the capillary micro-sampling head from its storage position toward the inner wall of the water tower, so that the tip of the sampling head contacts the condensate film on the tower wall. Through capillary force, a specific volume of condensate is continuously adsorbed, and the adsorption process continues until a preset single sampling volume is reached. In specific implementations, the preset single sampling volume is determined by the microtube size of the capillary micro-sampling head and the adsorption time. The adsorption time is estimated using a formula:

[0035]

[0036] in: Indicates the adsorption volume. Indicates the microtubule radius, Indicates capillary pressure difference. Indicates the dynamic viscosity of water. Indicates the length of the microtubule. The adsorption time is used to control the single sampling volume. Once the preset single sampling volume is reached, all capillary adsorption micro-sampling heads retract into the contamination-proof sealed cavity. This retraction is driven in reverse by a micro-stepping motor. The contamination-proof sealed cavity is a sealed container filled with inert gas. After retraction, the sampling tips of the capillary adsorption micro-sampling heads are located inside the cavity, preventing contact with the external environment. A micro-quantitative pump located within the cavity is activated. This pump is connected to the base of each capillary adsorption micro-sampling head via a flexible tube. Upon startup, the pump generates negative pressure, transferring a specific volume of condensate adsorbed by each head to a corresponding individually numbered sample storage tube. These tubes, each with a unique RFID tag, are placed on a sample rack within the cavity. At the end of the sampling cycle, condensate samples from all individually numbered sample storage tubes are collected. This collection operation is performed by a robotic arm, which picks up all the individually numbered sample storage tubes from the sample rack and places them onto a centralized conveyor belt, forming a raw condensate water sample set containing spatiotemporal location information derived from the RFID tag of each individually numbered sample storage tube and its corresponding sampling head position mapping. In some embodiments, preliminary physical processing includes removing solid suspended particles through microporous filtration and removing dissolved gases through low-temperature isothermal settling. This processing is carried out in dedicated equipment, which includes a filtration unit and a degassing unit. Each sample from the initial collection of condensate water samples is sequentially passed through an inorganic ceramic microporous filter membrane with a constant pore size. This membrane is installed within a filtration unit, whose inlet is connected to an individually numbered sample storage tube via a pipe. Driven by a micropump, the sample liquid flows through the membrane, trapping solid suspended particles larger than a set threshold. This threshold is determined by the pore size of the membrane, resulting in a filtered water sample. This filtered water sample is collected in a temporary storage cup below the filtration unit. Optionally, the inorganic ceramic microporous filter membrane has a pore size of 0.45 micrometers to trap most bacteria and particulate matter. The filtered water sample is then placed in a temperature-controlled transparent sample cell made of transparent quartz glass with a jacket. Coolant from a thermostatic bath circulates within the jacket. Under light-protected conditions, the temperature of the transparent sample cell is lowered and stabilized at a preset low-temperature isothermal point, set at 4 degrees Celsius. Temperature control is achieved through a thermoelectric module and feedback circuit. At a low temperature constant temperature point, the filtered water sample in the transparent sample cell was left to stand for a continuous period of 30 minutes, so that the gas dissolved in the water would slowly precipitate out due to changes in solubility. The precipitated gas formed tiny bubbles in the liquid and floated to the surface.In the upper space of the transparent sample cell, a constant micro-negative pressure environment, lower than ambient pressure, is maintained. This micro-negative pressure is generated and maintained by a miniature vacuum pump connected to the top of the transparent sample cell, with the pressure value maintained at -20 kPa. This accelerates and guides the released gas to detach from the liquid surface, and the detached gas is then removed by the miniature vacuum pump. After the settling process, the liquid that has undergone gas removal is extracted from the center of the transparent sample cell. The extraction operation is performed using a slender injection needle inserted into the center of the transparent sample cell. The needle is connected to an injection pump, which extracts the liquid at a constant rate and injects it into a new clean sample tube, completing the initial physical treatment of a single sample. All samples undergo the same treatment to form a pre-treated condensate sample set. Optionally, the new clean sample tube is a pre-cleaned glass ampoule filled with high-purity nitrogen to prevent the sample from being re-contaminated or oxidized after treatment.

[0037] In one embodiment of the present invention, a single sample from a pretreated condensate sample set is divided into three equal subsamples using a precision pipette. The first subsample is introduced into the ultraviolet (UV) spectral channel of a multi-channel parallel spectroscopic analyzer. Under continuous-wave UV irradiation, the absorbance of the sample in the UV band is scanned and recorded as a function of wavelength, i.e., the UV absorption spectrum, with a scanning range covering 200 nm to 400 nm. The second subsample is introduced into the near-infrared (NIIR) spectral channel of the multi-channel parallel spectroscopic analyzer. Under near-infrared broadband irradiation, the intensity of light transmitted through the sample is detected and recorded as a function of wavelength. The NIIR transmission spectrum is calculated by comparing it with a blank control, with a detection range covering 700 nm to 2500 nm. The third subsample is introduced into the Raman spectral channel of the multi-channel parallel spectroscopic analyzer. The sample is excited using a monochromatic laser, and the intensity of Raman scattered light generated by the sample is acquired and recorded as a function of Raman shift, i.e., the Raman scattering spectrum, with the Raman shift range set to 100 to 2000 wavenumbers. The ultraviolet absorption spectrum, near-infrared transmission spectrum, and Raman scattering spectrum from the same sample are integrated and aligned according to a preset data structure. The data structure is based on timestamps and sample numbers to form a multidimensional original spectral data package that uniquely corresponds to the same sample. The data packages of all samples together constitute the multidimensional original spectral data.

[0038] In practice, a single sample from the pretreated condensate sample set is divided into three subsamples. The division is performed using a precision pipette with three independent output channels, each connected to a sample container. The precision pipette draws a fixed volume of liquid (1.5 mL) from a single sample vial of the pretreated condensate sample set and then distributes it equally into the three subsample vials, with each subsample vial receiving 0.5 mL of liquid. The three subsample vials are labeled as UV subsample, near-infrared subsample, and Raman subsample, respectively. The first sample is introduced into the ultraviolet (UV) spectral channel of the multichannel parallel spectral analyzer. The UV spectral channel includes a continuous-wave UV light source, a sample cell, and an array detector. The continuous-wave UV light source emits a continuous spectrum with wavelengths from 200 nm to 400 nm. The first sample is injected into the sample cell of the UV spectral channel, which is made of quartz. Under continuous-wave UV light irradiation, the array detector scans and records the absorbance of the sample in the UV band as a function of wavelength, i.e., the UV absorption spectrum. The scan step size is 1 nm, and the absorbance is calculated using the formula:

[0039]

[0040] in: Indicates at wavelength The absorbance value at that point Indicates the light intensity transmitted through the sample. The ultraviolet absorption spectrum is stored in array form, representing the light intensity transmitted through the blank reference. The array index corresponds to the wavelength value, and the array element stores the absorbance value. In some embodiments, the optical path of the ultraviolet spectral channel is guided by an optical fiber, which transmits the light from the continuous-wave ultraviolet light source to the sample cell, and then from the sample cell to the array detector. The array detector is a charge-coupled device (CCD) detector, which converts the optical signal into an electrical signal. The electrical signal is digitized by an analog-to-digital converter and transmitted to the data processing unit. A second sample is introduced into the near-infrared spectral channel of the multi-channel parallel spectral analysis device. The near-infrared spectral channel includes a near-infrared broadband light source, a sample chamber, and a grating spectrometer. The near-infrared broadband light source covers a wavelength range of 700 nm to 2500 nm. The second sample is injected into the sample chamber of the near-infrared spectral channel, which has a potassium bromide window. Under the illumination of the near-infrared broadband light source, the grating spectrometer detects and records the curve of the light intensity transmitted through the sample as a function of wavelength. The near-infrared transmission spectrum is calculated by the data processing unit using the transmittance formula:

[0041]

[0042] in: Indicates at wavelength Transmittance at that location Indicates the light intensity transmitted through the sample. The near-infrared transmission spectrum, representing the light intensity transmitted through the blank control, is stored as a function of transmittance relative to wavelength. Optionally, the sample chamber of the near-infrared spectral channel is temperature-controlled at 25 degrees Celsius to reduce the impact of temperature fluctuations on the measurement results. A third sample is introduced into the Raman spectral channel of the multi-channel parallel spectral analysis device. The Raman spectral channel includes a monochromatic laser, a focusing lens system, and a spectrometer. The monochromatic laser emits a 785 nm wavelength laser with a power of 100 mW. The third sample is placed in the sample stage of the Raman spectral channel. The sample is excited using the monochromatic laser, which is focused onto the sample surface through the focusing lens system. The excitation point diameter is approximately 10 μm. The Raman scattering spectrum, representing the change in intensity of the Raman scattered light produced by the sample as a function of Raman shift, is collected and recorded. The collected Raman scattered light is then sent to the spectrometer, which is equipped with a back-illuminated detector. The spectrometer has a resolution of 4 wavenumbers and a Raman shift range from 100 to 2000 wavenumbers. The Raman scattering spectrum is stored as an array of intensity relative to the Raman shift. In some embodiments, the Raman spectroscopy channel is equipped with an autofocus system. This system adjusts the focusing lens system by detecting the position of the reflected laser, ensuring the laser focus remains on the sample surface. It is understood that the three spectral channels of the multi-channel parallel spectral analysis device are spatially separated but synchronously triggered by the same control system. The control system ensures that the data acquisition timestamps of the ultraviolet (UV), near-infrared (NIIR), and Raman spectroscopy channels are consistent, with timestamp accuracy down to the millisecond level. The UV absorption spectrum, NIIR transmission spectrum, and Raman scattering spectrum from the same sample are integrated and aligned according to a preset data structure. This preset data structure defines a multidimensional array containing fields for spectral type identification, wavelength or Raman shift, intensity, timestamp, and sample number. The integration process maps the wavelength and absorbance arrays of the UV absorption spectrum, the wavelength and transmittance arrays of the NIIR transmission spectrum, and the Raman shift and intensity arrays of the Raman scattering spectrum to the corresponding fields of the multidimensional array. The alignment operation associates the three spectral data with the same sample based on the timestamp, forming a multidimensional raw spectral data package uniquely corresponding to the same sample. This multidimensional raw spectral data package is saved in binary file format. Optional data structure for the multidimensional raw spectral data package, see Table 1.

[0043] Table 1: Data Structure Table of Multidimensional Raw Spectral Data Package

[0044]

[0045] The data packets of all samples together constitute multidimensional raw spectral data, which is stored in a database system. The database system creates an index for each data packet, based on the sample number and timestamp. It can be understood that the generation process of multidimensional raw spectral data is fully automated; from sample aliquoting to data integration, no manual intervention is required. This automation is controlled by the built-in program of the multichannel parallel spectral analysis device, which coordinates the workflow of the pipette, light source, detector, and data storage unit.

[0046] See Figure 3 This is a UV absorption spectrum of a condensate sample. The curve shows two distinct characteristic absorption peaks, consistent with the typical UV absorption characteristics of organic matter, ions, and other pollutants in condensate. The first absorption peak corresponds to the characteristic absorption of aromatic organic matter and humic acids in the water, a typical indicator of natural organic matter (NOM). The second absorption peak corresponds to the absorption of inorganic ions such as nitrates and nitrites, as well as some conjugated double-bonded organic compounds, an important indicator of water pollution. The extremely low noise level (absorbance <0.05) indicates good stability of the spectral detection system and compliant sample pretreatment (impurity removal and degassing). This visually presents the core principle of UV spectroscopy for detecting condensate water quality, identifying pollutant types through characteristic peaks, clearly demonstrating the validity of the detection data, and providing reliable raw data support for subsequent spectral fusion and water quality profiling.

[0047] In one embodiment of the present invention, for each multidimensional raw spectral data packet in the multidimensional raw spectral data, the characteristic absorption peak positions and peak areas of its ultraviolet absorption spectrum are extracted using a peak detection algorithm; the characteristic transmission valley positions and valley depths of its near-infrared transmission spectrum are extracted using a valley detection algorithm; and the characteristic Raman peak positions and peak intensities of its Raman scattering spectrum are extracted using a spectral deconvolution method. The characteristic absorption peak positions, characteristic transmission valley positions, and characteristic Raman peak positions from the same sample are mapped and associated, based on a wavelength and Raman shift correspondence table. A pre-set "spectral features - water quality parameters" association knowledge base is consulted to preliminarily identify ion types, organic functional group categories, and colloidal substance types. The knowledge base stores standard spectral features. The characteristic absorption peak areas, characteristic transmission valley depths, and characteristic Raman peak intensities from the same sample are normalized using a minimum-maximum scaling method and input into a trained concentration inversion model to calculate preliminary concentration estimates of ion types, organic functional group categories, and colloidal substances. The concentration inversion model is based on a machine learning algorithm. By integrating preliminary identification results and preliminary concentration estimates, a multi-dimensional water quality feature vector is generated for each condensate sample, stored as an array. The spatiotemporal location information of each condensate sample, recorded during collection and including the 3D coordinates of the sampling point on the inner wall of the water tower and the sampling time, is extracted from the sampling log. A 3D spatial mesh model of the water tower's inner wall is established, and the multi-dimensional water quality feature vector corresponding to each condensate sample is assigned to the corresponding grid nodes in the 3D spatial mesh model based on coordinate matching. For grid nodes not directly covered by the sample, a distance-weighted spatial interpolation algorithm is used to calculate their values ​​using the multi-dimensional water quality feature vectors of neighboring nodes, thus filling the entire 3D spatial mesh model. The interpolation algorithm uses the inverse distance weighting method. Based on the fully filled 3D spatial mesh model, a contour map of the spatial distribution of concentration on the inner wall surface of the water tower is generated for each water quality component, and the contour maps are rendered using graphics software. By integrating the spatial distribution contour maps of the concentrations of all water quality components and overlaying them with the structural map of the water tower's inner wall, a spatial water quality profile reflecting the spatial heterogeneity of water quality parameters is constructed, and the profile is output in image format.

[0048] In specific implementation, for each multidimensional raw spectral data package in the multidimensional raw spectral data, the characteristic absorption peak positions and peak areas of its ultraviolet absorption spectrum are extracted. The extraction process uses a peak detection algorithm, which traverses the ultraviolet spectral absorbance array to identify the wavelength position corresponding to the local absorbance maximum value as the characteristic absorption peak position. The area under the curve centered on the characteristic absorption peak position within a preset wavelength range is calculated as the characteristic absorption peak area. The characteristic transmission valley positions and valley depths of its near-infrared transmission spectrum are then extracted using a valley value detection algorithm. The infrared spectral transmittance array is used to identify the wavelength position corresponding to the local transmittance minimum as the characteristic transmission valley position. The difference between the transmittance at this point and the average of the transmittance of the adjacent local maximum values ​​on both sides is calculated as the characteristic transmission valley depth. The characteristic Raman peak position and peak intensity of the Raman scattering spectrum are extracted. The extraction adopts the spectral deconvolution method. The spectral deconvolution method separates overlapping Raman peaks by performing a deconvolution operation on the Raman spectral intensity array with a known instrument function, thereby accurately determining the position of each characteristic Raman peak and its corresponding peak intensity. The peak intensity is the peak intensity value of a single Raman peak after deconvolution. Characteristic absorption peaks, characteristic transmission valleys, and characteristic Raman peaks from the same sample are mapped and correlated. This mapping is based on a pre-defined wavelength-Raman shift correspondence table stored in the digital system's memory. A pre-defined "spectral features-water quality parameters" association knowledge base is then consulted. This knowledge base is a relational database storing standard spectral characteristic data of standard ions, standard organic functional groups, and standard colloidal substances in the ultraviolet, near-infrared, and Raman bands. This process initially identifies the ion types, organic functional group categories, and colloidal substance types. The identification process involves matching the measured characteristic peaks with the standard peaks in the "spectral features-water quality parameters" association knowledge base. The substance category corresponding to a successfully matched standard peak is the preliminary identification result. In some embodiments, the results of feature extraction and mapping can be organized into a data structure, as shown in Table 2.

[0049] Table 2: Feature Extraction and Preliminary Recognition Results

[0050]

[0051] The characteristic absorption peak area, characteristic transmission valley depth, and characteristic Raman peak intensity from the same sample were normalized using a minimum-maximum scaling method. The characteristic values ​​extracted from different spectral channels were scaled to between 0 and 1 using the following scaling formula:

[0052]

[0053] in: Represents the normalized eigenvalues. This represents the original peak area, valley depth, or peak intensity value. This represents the minimum value of the feature across all training samples. The maximum value of this feature across all training samples is input into the trained concentration inversion model. This model is a machine learning model based on support vector machine regression, trained using spectral feature data of historically known concentrations of condensate samples. It calculates preliminary concentration estimates for ion types, organic functional group categories, and colloidal substance types, outputting these estimates as floating-point numbers in milligrams per liter. The preliminary identification results and preliminary concentration estimates are then fused. The preliminary identification results are text labels for substance categories, while the preliminary concentration estimates are numerical values. This generates a multi-dimensional water quality feature vector for each condensate sample. This vector is a one-dimensional array, with elements arranged sequentially, containing all identified substance categories and their corresponding preliminary concentration estimates. The array length equals the total number of identified substance categories. The spatiotemporal location information recorded during the collection of each condensate sample is read. This information includes the three-dimensional coordinates of the sampling point on the inner wall of the water tower and the sampling time. This information is extracted from the sampling log database, which records the number of each capillary micro-sampling head, its corresponding three-dimensional coordinates, and the timestamp of the sampling action completion. A three-dimensional spatial mesh model of the water tower's inner wall was established, based on computer-aided design drawings. The model discretizes the inner wall surface into a dense triangular mesh, with each mesh node possessing unique three-dimensional coordinates. The multi-dimensional water quality feature vector corresponding to each condensate sample was assigned to the corresponding mesh node in the three-dimensional spatial mesh model. This assignment was achieved using a coordinate matching algorithm, which calculated the Euclidean distance between the sampling point coordinates and the coordinates of all mesh nodes, assigning the multi-dimensional water quality feature vector to the nearest mesh node. For mesh nodes not directly covered by the sample, a distance-weighted spatial interpolation algorithm was used. This algorithm employed the inverse distance weighting method, calculating the value of the multi-dimensional water quality feature vector using neighboring nodes. The inverse distance weighting method assigned a weight to the feature vector of each neighboring node, with the weight inversely proportional to the p-th power of the distance from that node to the mesh node to be interpolated. The weighted average of the feature vectors of all neighboring nodes was calculated to complete the filling of the entire three-dimensional spatial mesh model. It can be understood that, based on a fully filled 3D spatial mesh model, a spatial distribution contour map of the concentration of each water quality component on the inner wall surface of the water tower is generated. Each water quality component corresponds to a concentration dimension in a multi-dimensional water quality feature vector. The generation process extracts the concentration values ​​of all nodes in the 3D spatial mesh model along that dimension for each concentration dimension, and uses a contour line drawing algorithm to connect points of equal concentration on a 2D projection plane to form a spatial distribution contour map of the concentration. In some embodiments, the contour line drawing algorithm uses the moving cube algorithm, which processes the 3D scalar field data and generates 2D contour lines. The spatial distribution contour map of the concentration is saved in an image file format, and the image file includes coordinate scales.This process integrates isopleth maps of the spatial distribution of all water quality components' concentrations, overlaid with a 3D wireframe model of the water tower's inner wall structure. This creates a spatial water quality profile reflecting the spatial heterogeneity of water quality parameters. The spatial water quality profile is a composite layer: the bottom layer is the water tower's inner wall structure map, and the top layer, semi-transparently overlaid, displays isopleth maps of the spatial distribution of multiple water quality components' concentrations. Different water quality components are coded with different colors. Optionally, the spatial water quality profile can be viewed interactively, allowing users to choose to show or hide the water quality component layers. The construction of the spatial water quality profile is entirely automated by the software. The software reads the filled 3D spatial mesh model data and the water tower's inner wall structure map data, calls graphics library functions for rendering and overlaying, and finally outputs the spatial water quality profile file.

[0054] See Figure 4 This is a time-series analysis chart of multiple parameters in the real-time operating environment of the water tower, showing the changing trends of core environmental parameters of the water tower wall condensate detection system during a 25-hour monitoring period. The air pressure remained stable with a slight increase, fluctuating slightly around 100 kPa throughout the period; humidity continuously decreased to a low point from 0 to 12 hours, and gradually rose to a high point from 12 to 25 hours; temperature rose slightly to a peak from 0 to 8 hours, slowly decreased to a low point from 8 to 20 hours, and slightly rebounded after 20 hours; airflow velocity remained in a low-speed range of 0~2 m / s throughout the period, without drastic fluctuations. The stable low-speed airflow velocity throughout indicates that the "capillary adsorption sampling head" of the micro-sampling array will not cause sampling deviation due to strong airflow interference, verifying the feasibility of the sampling process. Stable air pressure eliminates the interference of sudden changes in the external environment on spectral detection, providing environmental assurance for the accuracy of ultraviolet, near-infrared, and Raman spectral data. The parameter changes exhibit a clear periodicity (around 24 hours), consistent with the diurnal environmental variation pattern of the water tower operation.

[0055] In one embodiment of the invention, the airflow velocity distribution data inside the water tower corresponding to the collection time point of the condensate sample is extracted from the real-time operating environment dataset. The data comes from an array of wind speed sensors. The contours of high-concentration areas of specific pollutants in the spatial water quality profile are spatially superimposed with the airflow velocity distribution map inside the water tower at the same time. The superposition is achieved through coordinate alignment. The spatial conformity between the contours of the high-concentration areas of specific pollutants and the low-speed airflow areas, vortex areas, or airflow impact wall areas is analyzed, and the spatial overlap coefficient is calculated based on the area overlap ratio. The historical change curves of temperature readings at multiple points on the inner wall of the water tower are compared with the corresponding ion concentration change curves at the points using time-series correlation analysis. The time-delay correlation coefficient is calculated using the Pearson correlation coefficient method. The spatial overlap coefficient and the time-delay correlation coefficient are recorded to form a dynamic correlation dataset for explaining the causes of spatial distribution of water quality. The dataset is stored in a database. Concentration safety thresholds and spatial distribution uniformity thresholds for key water quality parameters are set, and the thresholds are set based on industry standards. The actual concentration values ​​and spatial distribution of each water quality parameter in the spatial water quality profile generated in the current period are compared with the concentration safety threshold and the spatial distribution uniformity threshold using an automatic scanning algorithm. When the actual concentration value exceeds the concentration safety threshold, or the spatial distribution non-uniformity exceeds the spatial distribution uniformity threshold, a water quality anomaly marker is triggered, and the anomaly type is recorded. Dynamic association datasets related to the spatial region and time point corresponding to the water quality anomaly marker are retrieved to analyze the main environmental factors causing the anomaly, using association rule mining. Combining the main environmental factors and the water quality anomaly marker, an abnormal condensation water quality early warning report is generated, including the anomaly location, anomaly parameters, anomaly degree, and associated environmental conditions. The report is output in text and chart formats.

[0056] In the specific implementation, the airflow velocity distribution data inside the water tower corresponding to the condensate sample collection time point is extracted from the real-time operating environment dataset. The real-time operating environment dataset is stored in a time-series database. During extraction, based on the timestamp of the condensate sample collection time point, the airflow velocity data recorded by the wind speed sensor array installed inside the water tower at the same time is queried from the time-series database. The wind speed sensor array is arranged in a three-dimensional grid, and each sensor records its three-dimensional coordinates and airflow velocity value. The airflow velocity distribution data inside the water tower is represented as a set of velocity vectors at each grid point. The outline of the high-concentration area of ​​specific pollutants in the spatial water quality profile is spatially superimposed with the airflow velocity distribution map inside the water tower at the same time. The spatial water quality profile is stored in digital image format. The outline of the high-concentration area of ​​specific pollutants is extracted from the spatial water quality profile through an image processing algorithm. The image processing algorithm performs binarization processing on the spatial distribution contour map of the concentration of specific pollutants, and marks the area with the concentration value exceeding the preset threshold as the foreground, thereby obtaining the outline polygon. The airflow velocity distribution map inside the water tower is generated from the airflow velocity vector data through streamline visualization technology. Spatial superposition is performed in the digital processing system, and the outline polygon layer and the airflow velocity distribution map layer are registered and superimposed in a unified spatial coordinate system. The spatial overlap coefficient is analyzed by comparing the contour of a high-concentration region of a specific pollutant with the spatial fit of low-velocity airflow regions, vortex regions, or airflow impact wall regions. Low-velocity airflow regions are identified from the airflow velocity distribution map inside the water tower by setting a velocity threshold of 0.5 m / s. Vortex regions are obtained by calculating the curl field of the airflow velocity distribution and identifying regions with high curl values. Airflow impact wall regions are identified by analyzing the magnitude of the normal component of the velocity vector near the wall. The spatial overlap coefficient is calculated, defined as the ratio of the intersection area of ​​the contour area of ​​the high-concentration region of the specific pollutant to the contour area of ​​the characteristic airflow region, divided by the contour area of ​​the high-concentration region of the specific pollutant. The calculation is performed using the following formula:

[0057]

[0058] in: Indicates the spatial overlap coefficient. This represents the area of ​​the overlapping region between the outline of a high-concentration area of ​​a specific pollutant and the outline of a region characterized by airflow. This represents the outline area of ​​a region with high concentrations of a specific pollutant. In some embodiments, the spatial overlap coefficient between the outline of the high-concentration region and the outline of the low-speed airflow region is calculated by first obtaining the overlapping region polygon through polygon Boolean operations in a geographic information system, and then calculating the area of ​​the overlapping region polygon. and the area of ​​the polygonal outline of the high-concentration region Finally, the spatial overlap coefficient is calculated by substituting the values ​​into the formula. A time-series correlation analysis is performed between the historical temperature readings at multiple points on the inner wall of the water tower and the corresponding specific ion concentration change curves. The historical temperature readings at multiple points on the inner wall of the water tower are obtained from a historical database, which stores the temperature value sequence recorded by each temperature sensor in chronological order. The specific ion concentration change curves at the corresponding points are extracted from the multi-dimensional water quality feature vector generated from multiple consecutive sampling periods. The specific ion concentration values ​​at specific coordinate points are extracted and sorted by time to form a concentration time series. The time-delay correlation coefficient is calculated using the Pearson correlation coefficient formula, taking time delay into account. The formula is:

[0059]

[0060] in: Indicates time delay The time-delay correlation coefficient is below. Indicates time Temperature value, Indicates time Specific ion concentration values, This represents the average value of the temperature time series within the calculation window. This represents the average value of the concentration time series within the calculation window. This represents the total length of the time series. This represents the time delay value of the attempt. The time delay correlation coefficient is calculated by traversing a preset time delay range, taking the time delay correlation coefficient with the largest absolute value as the final result, and recording the corresponding time delay. The spatial overlap coefficient and time-lag correlation coefficient are recorded to form a dynamic correlation dataset for explaining the causes of spatial distribution of water quality. This dynamic correlation dataset is stored in a structured record format, with each record containing fields such as timestamp, sampling area identifier, specific pollutant identifier, airflow characteristic area type, spatial overlap coefficient, specific ion identifier, time-lag correlation coefficient, and time-lag value. Concentration safety thresholds and spatial distribution uniformity thresholds for key water quality parameters are set. The concentration safety threshold is determined according to industrial circulating water quality standards, while the spatial distribution uniformity threshold is determined by calculating the upper limit of the coefficient of variation (COP) of a certain water quality parameter concentration on the entire inner wall surface of the water tower. The COP is defined as the ratio of the concentration standard deviation to the concentration mean. These thresholds are stored numerically in the configuration file. The actual concentration values ​​of each water quality parameter in the spatial water quality profile generated in the current period are compared with the actual spatial distribution, concentration safety threshold, and spatial distribution uniformity threshold. The comparison process iterates through the concentration values ​​of each water quality parameter at each grid node in the spatial water quality profile, checking if any node's concentration value exceeds the concentration safety threshold for that water quality parameter. Simultaneously, the coefficient of variation (COP) of each water quality parameter's concentration value across all grid nodes is calculated, and its COP is checked to see if it exceeds the spatial distribution uniformity threshold. When the actual concentration value exceeds the concentration safety threshold or the actual spatial distribution non-uniformity exceeds the spatial distribution uniformity threshold, a water quality anomaly marker is triggered. The water quality anomaly marker is a data structure containing information such as the anomaly parameter name, anomaly location coordinates, anomaly type, anomaly value, and occurrence time. The triggered water quality anomaly marker is recorded in the anomaly event log. The system retrieves dynamic correlation datasets related to the spatial regions and time points corresponding to water quality anomaly markers. It analyzes the main environmental factors causing the anomalies. The analysis process involves filtering records from the dynamic correlation dataset that match the spatial regions and time points of the water quality anomaly markers. Matching is based on the spatial coordinate range and time window. Statistical analysis is performed on the filtered records, for example, identifying the environmental factors with the highest spatial overlap coefficient or time-lag correlation coefficient related to the current anomalous pollutant and identifying them as the main environmental factors. Combining the main environmental factors with the water quality anomaly markers, an abnormal condensate water quality early warning report is generated, including the anomaly location, anomaly parameters, anomaly degree, and associated environmental conditions. The abnormal condensate water quality early warning report is generated in a structured text and chart format. The main body of the report details the various information of the water quality anomaly markers, while the appendix displays the relevant dynamic correlation dataset analysis results, such as the specific values ​​of the spatial overlap coefficient and time-lag correlation coefficient. Optionally, the abnormal condensate water quality early warning report can be automatically sent to the water tower operation monitoring center via a message queue. It is understandable that the dynamic correlation analysis step and the abnormal condensation water quality early warning step are automatically executed after each new sampling cycle and the generation of spatial water quality profile. The execution is triggered by the background analysis service, which reads the latest spatial water quality profile and real-time operating environment dataset, and performs calculations, comparisons and report generation according to the above process.In some embodiments, the spatial distribution uniformity threshold can be set by considering the historical operating data of the water tower. The 95th percentile of the coefficient of variation of water quality parameters in the historical spatial water quality profile can be set as the spatial distribution uniformity threshold. Optionally, the triggering conditions for water quality anomaly marking can be configured as a logical combination. For example, marking is only triggered when both concentration exceedance and spatial distribution non-uniformity are simultaneously met. The logical combination is defined by the rule engine in the configuration file.

[0061] See Figure 5 This is a time-series correlation analysis chart of temperature change and ion concentration, showing the temporal relationship between the water tower inner wall temperature and the concentration of core inorganic ions over a 72-hour monitoring period. Temperature exhibits obvious periodic fluctuations, which is the core driving factor for condensate formation. The overall nitrate concentration is higher than sulfate, with the largest fluctuation range, and shows a strong negative correlation with temperature. Sulfate concentration changes relatively smoothly, following the temperature trend but with a slightly weaker lag. Temperature peaks correspond to nitrate / sulfate concentration troughs; temperature troughs correspond to ion concentration peaks. Low temperatures are more conducive to the enrichment of water-soluble inorganic salts in the condensate film, while high temperatures accelerate water evaporation or ion desorption. This explains the spatial heterogeneity of condensate water quality parameters. Nitrate has the fastest response and the most dramatic fluctuations, indicating its rapid migration / adsorption rate in the condensate film and its susceptibility to transient temperature changes. Sulfate's relatively lagging response indicates its more stable binding form, making it more susceptible to the influence of airflow velocity inside the water tower than instantaneous temperature.

[0062] In one embodiment of the present invention, the micro water quality sampling array comprises multiple capillary adsorption micro-sampling heads arranged in a preset two-dimensional grid and connected to a drive unit via a shared support structure. The entire micro water quality sampling array is mounted on a carrier that can move along a guide rail fixed to the inner wall of a water tower. In some embodiments, each capillary adsorption micro-sampling head of the micro water quality sampling array is assigned a sub-region within the active condensation area it is responsible for sampling. The assignment process is based on a digital division of the spatial range of the active condensation area, and each capillary adsorption micro-sampling head is bound to a specific coordinate interval. Optionally, the drive unit controls the micro water quality sampling array to move along the guide rail above the target active condensation area, so that the capillary adsorption micro-sampling head array spatially covers the entire area to be sampled.

[0063] It is understood that controlling all capillary adsorption micro-sampling heads to extend simultaneously during the predicted peak condensation period is achieved. The central controller of the micro water quality sampling array receives the peak condensation period signal from the dynamic prediction module and drives a linear motor to extend the capillary adsorption micro-sampling heads, bringing their tips into contact with the condensate film on the tower wall. The capillary force generated by the micropores inside the capillary adsorption micro-sampling heads continuously adsorbs a specific volume of condensate. The trigger time for this extension action... Rate of change of temperature normal to the wall and relative humidity of air Functions jointly determined The calculation yields the following formula:

[0064]

[0065] function It describes the comprehensive relationship between the rate of temperature change and the influence of humidity on the condensation rate. When the calculation result reaches the preset threshold, the extension command is triggered.

[0066] After reaching the preset single sampling volume, all capillary adsorption micro-sampling heads retract into the contamination-proof sealed cavity. The contamination-proof sealed cavity moves synchronously with the capillary adsorption micro-sampling head array, completely enclosing the capillary adsorption micro-sampling heads during non-sampling periods. A micro-quantitative pump located within the contamination-proof sealed cavity is activated. This pump is connected to each capillary adsorption micro-sampling head via an independent microfluidic channel. The pump transfers a specific volume of condensate adsorbed by each capillary adsorption micro-sampling head to its corresponding independently numbered sample storage tube. At the end of the sampling cycle, the condensate samples from all independently numbered sample storage tubes are collected. Each independently numbered sample storage tube carries an electronic tag recording its sampling location and time, ultimately forming a raw condensate water sample set containing spatiotemporal location information. After completing the sampling task for one area, the micro water quality sampling array can move along the guide rail to the next active condensation area and repeat the above adsorption sampling process.

[0067] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for detecting the water quality of condensate from the wall of a water tower, characterized in that, include: Obtain the real-time operating environment dataset of the target water tower, which includes temperature readings at multiple points on the inner wall of the water tower, air humidity inside the water tower, atmospheric pressure outside the water tower, and airflow velocity distribution inside the water tower. By using the temperature readings at multiple points on the inner wall of the water tower, the temperature gradient of each monitoring area on the inner wall of the water tower is calculated. Combined with the humidity of the air inside the water tower, the active condensation area of ​​condensate on the surface of the inner wall of the water tower is dynamically predicted. A micro water quality sampling array is arranged in the active condensation area. The micro water quality sampling array is activated to perform timed adsorption and collection of condensate in the active condensation area to form a collection of original condensate water samples. The original condensate water sample set was subjected to preliminary physical treatment to obtain a pretreated condensate water sample set; The pretreated condensate sample set is input into a multi-channel parallel spectral analysis device, which simultaneously acquires the ultraviolet absorption spectrum, near-infrared transmission spectrum, and Raman scattering spectrum of the pretreated condensate sample set to generate multidimensional raw spectral data. The multidimensional raw spectral data are fused and analyzed and feature extracted to generate a multidimensional water quality feature vector for each condensate sample; A spatial water quality profile is constructed based on the multi-dimensional water quality feature vectors.

2. The method for detecting the water quality of condensate from a water tower wall according to claim 1, characterized in that, Using multi-point temperature readings on the inner wall of the water tower, the temperature gradient of each monitored area on the inner wall is calculated. Combined with the ambient air humidity inside the water tower, the active condensation zones on the inner wall surface of the water tower are dynamically predicted, including: Based on the spatial distribution of temperature readings at multiple points on the inner wall of the water tower, a digital model of the temperature field on the inner wall of the water tower is constructed. Spatial differentiation is performed on the digital model of the temperature field inside the water tower to obtain the rate of change of the normal temperature of the wall at each coordinate point of the water tower, i.e., the temperature gradient. Read the current air dew point temperature from the real-time updated air humidity data inside the water tower; In the digital model of the temperature field inside the water tower, all coordinate points where the actual temperature of the wall surface is lower than the current air dew point temperature are marked, forming a preliminary set of condensation points; In the preliminary set of condensation points, coordinate points whose absolute temperature gradient exceeds the condensation intensity threshold are selected, and the physical regions corresponding to the selected coordinate points are defined as the active condensation regions.

3. The method for detecting the water quality of condensate from a water tower wall according to claim 2, characterized in that, The micro water quality sampling array is activated to periodically adsorb and collect condensate from the active condensation zone, forming a raw condensate water sample collection, including: The micro water quality sampling array is composed of multiple capillary adsorption micro sampling heads; For each capillary adsorption micro-sampling head of the micro water quality sampling array, it is assigned to collect a sub-region within the active coagulation region. All capillary adsorption micro-sampling heads are controlled to extend simultaneously during the predicted peak condensation period, so that the tip of the sampling head contacts the condensate film on the tower wall, and continuously adsorbs a specific volume of condensate through capillary force. Once the preset single sampling volume is reached, all capillary adsorption micro-sampling heads are controlled to retract into the pollution-proof sealed cavity. A miniature metering pump located in a pollution-proof sealed cavity is driven to transfer a specific volume of condensate adsorbed by each capillary adsorption miniature sampling head to the corresponding independently numbered sample storage tube. At the end of the sampling period, the condensate samples from all the individually numbered sample storage tubes are collected to form the original condensate water sample set containing spatiotemporal location information.

4. The method for detecting the water quality of condensate from a water tower wall according to claim 3, characterized in that, The original condensate water sample set was subjected to preliminary physical treatment to obtain a pretreated condensate water sample set, including: The preliminary physical treatment includes removing solid suspended particles by microfiltration and removing dissolved gases by low-temperature constant-temperature settling. Each sample in the original condensate water sample set is sequentially passed through an inorganic ceramic microporous filter membrane with a constant pore size to retain solid suspended particles with a particle size larger than a set threshold, thus obtaining a filtered water sample. The filtered water sample was placed in a temperature-controlled transparent sample cell, and under light-protected conditions, the temperature of the transparent sample cell was lowered and stabilized at a preset low-temperature constant temperature point. At the aforementioned low-temperature constant temperature point, the filtered water sample in the transparent sample cell is allowed to stand for a continuous period of time, allowing the gas dissolved in the water to slowly precipitate out due to changes in solubility. In the upper space of the transparent sample cell, a constant micro-negative pressure environment lower than the ambient pressure is maintained to accelerate and guide the precipitated gas to escape from the liquid surface; After the settling process is completed, the liquid that has undergone gas removal treatment is extracted from the middle of the transparent sample cell and injected into a new clean sample tube to complete the preliminary physical treatment of a single sample. All samples are treated in the same way to form the pretreated condensate sample set.

5. The method for detecting the water quality of condensate from a water tower wall according to claim 4, characterized in that, The pretreated condensate sample set is input into a multi-channel parallel spectroscopic analysis device, which simultaneously acquires the ultraviolet absorption spectrum, near-infrared transmission spectrum, and Raman scattering spectrum of the pretreated condensate sample set to generate multidimensional raw spectral data, including: A single sample from the pretreated condensate sample set is divided into three equal subsamples. The first sample is introduced into the ultraviolet spectral channel of the multi-channel parallel spectral analysis device. Under the illumination of a continuous wave ultraviolet light source, the absorbance of the sample in the ultraviolet band is scanned and recorded as a function of wavelength, which is the ultraviolet absorption spectrum. The second sample is introduced into the near-infrared spectral channel of the multi-channel parallel spectral analysis device. Under the illumination of the near-infrared broadband light source, the curve of light intensity transmitted through the sample as a function of wavelength is detected and recorded. The near-infrared transmission spectrum is calculated by comparing it with a blank control. The third sample is introduced into the Raman spectroscopy channel of the multi-channel parallel spectral analysis device, the sample is excited by a monochromatic laser, and the curve of the intensity of the Raman scattered light generated by the sample changing with the Raman shift is collected and recorded, which is the Raman scattering spectrum. The ultraviolet absorption spectrum, near-infrared transmission spectrum, and Raman scattering spectrum from the same sample are integrated and aligned according to a preset data structure to form a multidimensional original spectral data package that uniquely corresponds to the same sample. The data packages of all samples together constitute the multidimensional original spectral data.

6. The method for detecting the water quality of condensate from a water tower wall according to claim 5, characterized in that, The multidimensional raw spectral data are fused, analyzed, and feature extracted to generate a multidimensional water quality feature vector for each condensate sample, including: For each of the multidimensional raw spectral data packages in the multidimensional raw spectral data, the characteristic absorption peak position and peak area of ​​its ultraviolet absorption spectrum, the characteristic transmission valley position and valley depth of its near-infrared transmission spectrum, and the characteristic Raman peak position and peak intensity of its Raman scattering spectrum are extracted respectively. The characteristic absorption peaks, characteristic transmission valleys, and characteristic Raman peaks from the same sample are mapped and correlated. A pre-set "spectral characteristics-water quality parameters" correlation knowledge base is searched to preliminarily identify the types of ions, the categories of organic functional groups, and the types of colloidal substances. The characteristic absorption peak area, characteristic transmission valley depth, and characteristic Raman peak intensity from the same sample are normalized and input into a trained concentration inversion model to calculate preliminary concentration estimates of the ion types, organic functional group categories, and colloidal substance types. By combining the preliminary identification results with the preliminary concentration estimates, a multi-dimensional water quality feature vector is generated for each condensate sample.

7. The method for detecting the water quality of condensate from a water tower wall according to claim 6, characterized in that, Constructing a spatial water quality profile from the multi-dimensional water quality feature vectors includes: Read the spatiotemporal location information of each condensate sample recorded at the time of collection. The spatiotemporal location information includes the three-dimensional coordinates of the sampling point on the inner wall of the water tower and the sampling time point. A three-dimensional spatial mesh model of the inner wall of the water tower is established, and the multi-dimensional water quality feature vector corresponding to each condensate sample is assigned to the mesh node of the corresponding coordinate in the three-dimensional spatial mesh model. For grid nodes not directly covered by the sample, a distance-weighted spatial interpolation algorithm is used to calculate their values ​​using the multi-dimensional water quality feature vectors of neighboring nodes, thus completing the filling of the entire three-dimensional spatial grid model. Based on the fully filled three-dimensional spatial mesh model, a spatial distribution contour map of the concentration of each water component on the inner wall surface of the water tower is generated. By integrating the spatial distribution contour maps of the concentrations of all water quality components and overlaying them with the structural map of the water tower's inner wall, a spatial water quality profile reflecting the spatial heterogeneity of water quality parameters is constructed.

8. The method for detecting the water quality of condensate from a water tower wall according to claim 7, characterized in that, It also includes a dynamic correlation analysis step based on the spatial water quality profile and the real-time operating environment dataset: Extract the airflow velocity distribution data inside the water tower corresponding to the time point of the condensate sample collection from the real-time operating environment dataset; The outline of a high-concentration area of ​​a specific pollutant in the spatial water quality image is spatially superimposed with the airflow velocity distribution map inside the water tower at the same time. Analyze the spatial overlap coefficient by comparing the contour of the high-concentration region of the specific pollutant with the spatial fit of the low-speed airflow region, vortex region, or airflow impact wall region. The historical variation curves of temperature readings at multiple points on the inner wall of the water tower are compared with the time series correlation analysis of specific ion concentration variation curves at the corresponding points, and the time lag correlation coefficient is calculated. The spatial overlap coefficient and the time lag correlation coefficient are recorded to form a dynamic correlation dataset for explaining the causes of spatial distribution of water quality.

9. A method for detecting the quality of condensate water from a water tower wall according to claim 8, characterized in that, It also includes an abnormal condensate water quality early warning step based on the aforementioned dynamic correlation dataset: Set safe concentration thresholds and spatial distribution uniformity thresholds for key water quality parameters; The actual concentration values ​​and actual spatial distribution of each water quality parameter in the spatial water quality profile generated in the current period are compared with the concentration safety threshold and the spatial distribution uniformity threshold. When the actual concentration value exceeds the concentration safety threshold, or the non-uniformity of the actual spatial distribution exceeds the spatial distribution uniformity threshold, a water quality anomaly marker is triggered. Retrieve the dynamic correlation dataset related to the spatial region and time point corresponding to the water quality anomaly marker, and analyze the main environmental factors leading to the anomaly. By combining the main environmental factors and the water quality anomaly markers, an abnormal condensate water quality early warning report is generated, which includes the location of the anomaly, the parameters of the anomaly, the degree of the anomaly, and the associated environmental conditions.

10. A method for detecting the quality of condensate water from a water tower wall according to claim 2, characterized in that, Based on the spatial distribution of temperature readings at multiple points on the inner wall of the water tower, a digital model of the temperature field on the inner wall of the water tower is constructed, including: Based on the multi-point temperature readings obtained by the sensor network deployed on the inner wall surface of the water tower, the temperature estimate of the area on the inner wall surface of the water tower where no sensors are directly deployed is calculated based on a spatial interpolation algorithm. The measured temperature values ​​from the multi-point temperature readings are fused with the estimated temperature values ​​to generate a continuous digital model of the temperature field inside the water tower wall.

11. The method for detecting the water quality of condensate from a water tower wall according to claim 9, characterized in that, After generating the abnormal condensate water quality early warning report, the system also includes a response control step based on the early warning: The abnormal condensate quality early warning report is sent to the water tower operation control system. The water tower operation control system generates and executes targeted cleaning or control commands based on the abnormal location and associated environmental conditions in the abnormal condensate water quality early warning report.

12. The method for detecting the water quality of condensate from a water tower wall according to claim 1, characterized in that, The micro water quality sampling array is an online array that can move along the guide rails on the tower wall; The timed adsorption and collection of condensate from the active condensation region, the preliminary physical treatment of the original condensate sample set, and the input of the pretreated condensate sample set into a multi-channel parallel spectral analysis device are all performed automatically and continuously on an integrated online detection platform.

13. A water tower wall condensate water quality detection system, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for detecting the condensate quality of a water tower wall as described in any one of claims 1 to 12.

Images