Intelligent monitoring method for salt deposition of seawater cooling tower

Abstract

The application discloses a seawater cooling tower salt deposition intelligent monitoring method, and particularly relates to the technical field of seawater cooling tower operation state monitoring, and comprises a modular monitoring unit, a data preprocessing module, a salt deposition index construction module, a state identification and prediction module, a risk assessment module and a hierarchical early warning output module; wherein the modular monitoring unit is used for collecting multi-source monitoring data in multiple salt deposition monitoring areas of the seawater cooling tower and transmitting the multi-source monitoring data to the data preprocessing module. The salt deposition degree of each area is quantified. The coupling term of salt mist concentration and humidity quantifies the promoting effect of a high-humidity environment on salt mist condensation and adhesion, so that the surface conductivity can directly reflect the actual salt deposition amount through the corrosion-resistant monitoring sheet. The application prevents problems caused by state judgment through a single parameter by adopting a multi-source data fusion monitoring mode.

Description

Technical Field

[0001] This application relates to the field of monitoring the operating status of seawater cooling towers, and more specifically, to an intelligent monitoring method for salt deposition in seawater cooling towers. Background Technology

[0002] In coastal industrial circulating water systems, seawater cooling towers are core heat exchange equipment, cooling the circulating water through seawater spray evaporation. During operation, salt in the seawater evaporates with the spray water and diffuses as salt mist, gradually depositing on the packing, heat exchange components, and metal structure surfaces. As salt deposits increase, they can cause surface corrosion, decreased heat exchange performance, and blockage of local channels, thus weakening the cooling performance of the circulating water. The packing, as a core heat exchange component, is particularly vulnerable; if it ages and corrodes due to salt deposits, the overall cooling effect will be severely compromised.

[0003] A search revealed that CN104697908A discloses a method for monitoring drip salt deposition in seawater cooling towers. This method includes the following steps: (1) simulating meteorological conditions in the area to be measured to numerically simulate the distribution of drip salt deposition in seawater cooling towers; (2) combining the numerical simulation results with the geographical environment to formulate a site selection plan, and determining monitoring locations through on-site investigation; (3) placing sedimentation collection devices at the monitoring locations and using a wet sedimentation method to collect samples of drip salt deposition; (4) detecting the chloride ion concentration in the samples and calculating the monthly salt deposition amount at different monitoring locations. This method is scientifically effective, highly operable, stable, and the monitoring results accurately reflect the distribution of drip salt deposition in seawater cooling towers. The inventors discovered the following problems with the existing technology during the development of this application:

[0004] Existing methods for monitoring salt deposition in seawater cooling towers generally rely on periodic manual inspections and sampling, laboratory analysis, or the deployment of single sensors to monitor parameters such as salt spray concentration and humidity. However, manual inspections are discrete, leading to discontinuous monitoring that only reflects a local state at a specific moment, making 24-hour online monitoring impossible. Furthermore, the monitoring dimensions are limited, relying on a single parameter like salt spray concentration or humidity, which cannot simultaneously cover the three dimensions of salt input, deposition environment, and actual deposition state, resulting in insufficient accuracy in judging the degree of salt deposition. Additionally, there is the problem of delayed warnings, as it is impossible to predict the development trend of salt deposition; often, equipment malfunctions before any anomalies are detected.

[0005] Therefore, a smart monitoring method for salt deposition in seawater cooling towers is proposed to address the above problems. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, this application provides an intelligent monitoring method for salt deposition in seawater cooling towers to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, this application provides the following technical solution: an intelligent monitoring system for salt deposition in seawater cooling towers, comprising a modular monitoring unit, a data preprocessing module, a salt deposition index construction module, a status identification and prediction module, a risk assessment module, and a graded early warning output module;

[0008] The modular monitoring unit is used to collect multi-source monitoring data in multiple salt deposition monitoring areas of the seawater cooling tower and transmit it to the data preprocessing module.

[0009] Modular monitoring units are deployed in multiple salt deposition-sensitive areas of the seawater cooling tower, including the air inlet area, the periphery of the packing material, high-humidity and high-salt areas, and areas of metal structures prone to corrosion. Each monitoring unit integrates a temperature sensor, humidity sensor, wind speed sensor, salt spray concentration sensor, and conductivity detection module via an RS485 bus to achieve synchronous acquisition of multiple parameters.

[0010] The conductivity detection module includes a corrosion-resistant monitoring plate and a pair of detection electrodes disposed on the surface of the monitoring plate. The corrosion-resistant monitoring plate can be made of corrosion-resistant alloy material, uses bolt mounting, and can be periodically disassembled and replaced; after replacement, the reference value needs to be recalibrated. The detection electrodes are fixedly mounted on the monitoring bracket and are in close contact with the surface of the monitoring plate. During operation, the resistance signal between the detection electrodes is collected and converted into surface conductivity using the following formula: ;

[0011] Where, in the formula, For surface conductivity, To detect resistance, This is an electrode structure constant, determined by the electrode spacing and electrode area;

[0012] Among the five types of parameters collected, salt spray concentration is used to reflect the intensity of salt input in the area, temperature, humidity and wind speed are used to reflect the environmental conditions for salt deposition, and surface conductivity is used to directly reflect the salt adhesion state on the monitored surface. A complete monitoring system is constructed from three dimensions: input, environment and results.

[0013] The data preprocessing module is used to first remove outliers and compensate for missing values ​​in the multi-source monitoring data, then extract the cumulative salt input and salt deposition change rate based on the processed time-series data, and finally perform normalization preprocessing on the original monitoring data and the extracted feature quantities. The normalized multi-source monitoring data and feature quantities are then transmitted to the salt deposition index construction module and the state recognition and prediction module.

[0014] The data preprocessing module receives raw monitoring data uploaded by each modular monitoring unit and sequentially performs outlier removal, missing value compensation, and minimum-maximum normalization. Outlier removal uses the 3σ criterion to remove abnormal data exceeding the reasonable fluctuation range; missing value compensation uses linear interpolation to complete short-term missing data. Normalization uses the minimum-maximum normalization method to uniformly map all physical quantities to... The general formula for intervals is: ;

[0015] in, , For the historical minimum and maximum values ​​of the corresponding physical quantities, after normalization ;

[0016] After preprocessing, two core features are extracted: cumulative salt input and salt deposition rate.

[0017] Cumulative salt input: Calculated based on the time-series data of salt spray concentration within a set time window, using the following formula: ;

[0018] in, The number of samples within the time window, with a total window duration of [value missing]. , The sampling time interval, The first time before the current moment Salt spray concentration at each sampling point;

[0019] Salt deposition rate: calculated based on the temporal change in surface conductivity, using the following formula: : ;

[0020] in, The surface conductivity at the previous sampling time is denoted as .

[0021] The preprocessed raw monitoring data and the extracted feature quantities are transmitted together to the salt deposition index construction module and the state identification and prediction module;

[0022] The salt deposition index construction module is used to weight and fuse the cumulative salt input, the coupling characteristics of salt spray concentration and humidity, wind speed characteristics and surface conductivity characteristics to construct a comprehensive salt deposition index.

[0023] The salt deposition index construction module receives preprocessed data and feature values, and constructs a comprehensive salt deposition index through multi-feature weighted fusion to quantify the degree of salt deposition in each region. The calculation expression for the comprehensive salt deposition index is as follows: ;

[0024] in, To normalize the cumulative salt input, Normalized salt spray concentration, To normalize humidity, To normalize the wind speed, wind speed affects the diffusion and deposition of salt spray. To normalize the surface conductivity, which directly reflects the actual amount of salt adhering, , , , Let be the weight coefficients of each feature term, satisfying The weighting coefficients are determined using a data-driven method: historical multi-source monitoring data is used as input, and manually detected salt deposition levels are used as labels. The importance of each feature to the salt deposition state is calculated, and the corresponding weight values ​​are obtained after normalization. In an exemplary embodiment, the weight values ​​are... , , , The higher the value of the salt deposition comprehensive index, the higher the degree of salt deposition in the corresponding area.

[0025] because It is a coupling term of salt spray concentration and humidity. High humidity environment will promote the condensation and adhesion of salt spray on the surface. The product of the two can quantify the environmental promoting effect of salt deposition.

[0026] The state identification and prediction module is used to identify the current salt deposition level based on monitoring data and extracted feature quantities, and to predict the salt deposition development trend based on the comprehensive salt deposition index and the rate of change; the risk assessment module is used to calculate the salt deposition risk value by combining the current comprehensive salt deposition index, the rate of change and the prediction results.

[0027] The graded early warning output module is used to classify risks according to risk values ​​and output monitoring results and graded early warning information.

[0028] Preferably, the modular monitoring units are deployed in each salt deposition sensitive area of ​​the seawater cooling tower, and each monitoring unit integrates a temperature sensor, a humidity sensor, a wind speed sensor, a salt spray concentration sensor, and a conductivity detection module.

[0029] The conductivity detection module includes a corrosion-resistant monitoring sheet and a detection electrode disposed on the surface of the monitoring sheet, used to collect surface conductivity signals of the monitoring area; the salt spray concentration is used to reflect the salt input intensity, temperature, humidity and wind speed are used to reflect the salt deposition environment, and surface conductivity is used to reflect the salt adhesion state of the monitoring surface.

[0030] Preferably, the state recognition and prediction module includes a random forest classification unit and a trend prediction unit;

[0031] The random forest classification unit takes a feature vector composed of temperature, humidity, wind speed, salt spray concentration, surface conductivity, cumulative salt input, and salt deposition change rate as input, and outputs four state identification results: low salt deposition, general salt deposition, relatively high salt deposition, and high salt deposition.

[0032] The trend prediction unit calculates the predicted salt deposition index for future times through short-cycle linear extrapolation based on the current salt deposition comprehensive index and the salt deposition change rate. Based on the numerical range of the salt deposition change rate, the salt deposition development trend is divided into four states: continuous increase, slow increase, basically stable, and gradual decrease.

[0033] Preferably, the risk assessment module has a built-in dynamic threshold adjustment unit. The dynamic threshold adjustment unit is used to calculate the mean and standard deviation of historical risk values ​​based on the salt deposition risk value data of each monitoring area during the historical normal operation phase, and generate a dynamic warning threshold by combining the safety factor of the corresponding warning level. When the historical risk value data is insufficient, a preset initial threshold is used to classify the risk level.

[0034] A smart monitoring method for salt deposition in seawater cooling towers, step S1: Simultaneously collect five types of multi-source monitoring data, namely temperature, humidity, wind speed, salt spray concentration and surface conductivity, in multiple salt deposition monitoring areas of the seawater cooling tower.

[0035] Step S2: First, outlier removal and missing value compensation are performed on the multi-source monitoring data. Then, the cumulative salt input is calculated based on the processed salt spray concentration time series data. The salt deposition change rate is calculated based on the processed surface conductivity time series data. Finally, all monitoring data and extracted feature quantities are normalized and preprocessed.

[0036] Step S3: Assign corresponding weights to the normalized cumulative salt input, the coupling characteristics of salt spray concentration and humidity, wind speed characteristics and surface conductivity characteristics, and construct a comprehensive salt deposition index by weighted fusion;

[0037] Step S4: Combine the preprocessed multi-source monitoring data, the extracted cumulative salt input, and the salt deposition change rate to form a feature vector, input it into the trained classification model, identify the current salt deposition level, and predict the salt deposition development trend based on the comprehensive salt deposition index and change rate.

[0038] Step S5: Calculate the salt deposition risk value for each monitoring area by combining the current comprehensive salt deposition index, the rate of salt deposition change, and the predicted comprehensive salt deposition index;

[0039] Step S6: Compare the salt deposition risk value with the corresponding threshold to classify the risk level, and output the salt deposition monitoring results, graded early warning information and key salt deposition areas for each region.

[0040] Preferably, step S1 specifically includes:

[0041] Based on the structural characteristics, airflow distribution, and salt deposition-prone areas of the seawater cooling tower, the cooling tower was divided into multiple independent salt deposition monitoring zones. Modular monitoring units were deployed in each monitoring zone to simultaneously collect five types of monitoring parameters, forming a multi-source monitoring data sequence. : ;

[0042] in, It is the first Each monitoring area is in Multi-source monitoring data vectors at any given time For the first One salt deposition monitoring area, For the current moment, For temperature, For humidity, For wind speed, Salt spray concentration, denoted as surface conductivity.

[0043] The weighting coefficients are determined as follows: using historical multi-source monitoring data as input and manually detected salt deposition levels as labels, the influence of each feature on the salt deposition state is calculated using a random forest model, and the corresponding weight values ​​are obtained after normalization. The larger the comprehensive salt deposition index value, the higher the degree of salt deposition in the corresponding area.

[0044] Preferably, step S4 specifically comprises:

[0045] A 7-dimensional feature vector is composed of temperature, humidity, wind speed, salt spray concentration, surface conductivity, cumulative salt input, and salt deposition change rate. This vector is then input into a pre-trained random forest model, which outputs the salt deposition level of the current area. The training labels of the random forest model are calibrated in the following way: the corrosion-resistant monitoring sheet is periodically disassembled, its surface salt deposition mass is measured, the salt deposition mass per unit area is calculated, and it is divided into four salt deposition levels: low, average, relatively high, and high, according to a preset threshold.

[0046] When the system is deployed for the first time and there is insufficient historical training data, the system uses a rule-based initialization criterion based on the salt deposition comprehensive index for state recognition. When the number of accumulated labeled training samples reaches a preset threshold, the system triggers offline training of the random forest model and smoothly switches to a data-driven recognition mode.

[0047] Preferably, steps S5 and S6 specifically involve: calculating the salt deposition risk value by combining the current comprehensive salt deposition index, the normalized salt deposition change rate, and the predicted comprehensive salt deposition index. ;

[0048] in, This represents the normalized rate of salt deposition change. For the weighting coefficients, satisfying , For predicting duration The subsequent salt deposition comprehensive index is calculated by short-period linear extrapolation, wherein the prediction duration is... 1-24 hours;

[0049] Based on historical risk value data for each monitoring area, the historical mean and standard deviation are calculated, and dynamic warning thresholds are generated by combining the safety factor of the corresponding warning level.

[0050] The salt deposition risk value is compared with the dynamic early warning threshold or the preset initial threshold, and four risk levels are divided into normal, attention, early warning and severe early warning.

[0051] The system summarizes the monitoring results of all monitored areas and outputs the current salt deposition level, salt deposition change rate, development trend, predicted status, risk level, and information on key salt deposition areas.

[0052] The technical effects and advantages of this application are as follows:

[0053] 1. Compared with existing technologies, this intelligent monitoring method for salt deposition in seawater cooling towers deploys modular monitoring units in sensitive areas of the cooling tower to simultaneously collect five types of multi-source data: salt spray concentration, temperature, humidity, wind speed, and surface conductivity. It constructs a complete monitoring system from three dimensions: salt input intensity, deposition environment conditions, and surface salt adhesion status. Furthermore, by constructing a comprehensive salt deposition index, it weights and fuses the normalized cumulative salt input, the coupling characteristics of salt spray concentration and humidity, wind speed characteristics, and surface conductivity characteristics to quantify the degree of salt deposition in each area. The coupling term of salt spray concentration and humidity quantifies the promoting effect of high humidity on salt spray condensation and adhesion. Thus, surface conductivity can directly reflect the actual amount of salt adhesion through a corrosion-resistant monitoring plate. This invention, by employing a multi-source data fusion monitoring method, avoids problems caused by judging the state based on a single parameter.

[0054] 2. Compared with existing technologies, this intelligent monitoring method for salt deposition in seawater cooling towers divides the interior of the cooling tower into multiple independent salt deposition monitoring zones based on the tower's structural characteristics, airflow distribution patterns, and areas prone to salt deposition. Each zone is equipped with modular monitoring units. Each monitoring unit integrates temperature, humidity, wind speed, salt mist concentration, and conductivity detection modules to achieve 24-hour uninterrupted multi-parameter synchronous acquisition and data upload. This modular zoning monitoring approach allows for targeted monitoring based on the salt deposition risk level of different zones and focused monitoring of different structural areas. It also reduces the need for localized manual sampling and achieves continuous coverage across the entire area. This approach can also detect abnormal changes in salt deposition in key monitoring areas.

[0055] 3. Compared with existing technologies, this intelligent monitoring method for salt deposition in seawater cooling towers identifies the current salt deposition level, calculates the rate of salt deposition change, and uses a short-cycle linear extrapolation method to predict the comprehensive salt deposition index for the next 1-24 hours. It then calculates the salt deposition risk value for each region based on the current salt deposition status, rate of change, and prediction results. Simultaneously, based on historical risk value data for each monitoring region, it generates a dynamic early warning threshold through statistical analysis of the mean and standard deviation. The risk value is compared with the dynamic threshold or a preset initial threshold to classify four risk levels: normal, watchful, warning, and severe warning. This dynamic threshold automatically adjusts with seasonal changes, sea conditions, and operating conditions, ensuring the timeliness and adaptability of the early warning. Attached Figure Description

[0056] Figure 1 This is the overall system diagram of this application;

[0057] Figure 2 This is a flowchart illustrating the overall method of this application;

[0058] Figure 3 A flowchart illustrating the construction of the comprehensive salt deposition index for this application;

[0059] Figure 4 This is a flowchart of the status identification and risk warning process for this application. Detailed Implementation

[0060] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0061] Example 1

[0062] As attached Figures 1 to 4The intelligent monitoring system for salt deposition in seawater cooling towers includes a modular monitoring unit, a data preprocessing module, a salt deposition index construction module, a status identification and prediction module, a risk assessment module, and a graded early warning output module.

[0063] The modular monitoring unit is used to collect multi-source monitoring data in multiple salt deposition monitoring areas of the seawater cooling tower and transmit it to the data preprocessing module.

[0064] Modular monitoring units are deployed in multiple salt deposition-sensitive areas of the seawater cooling tower, including the air inlet area, the periphery of the packing material, high-humidity and high-salt areas, and areas of metal structures prone to corrosion. Each monitoring unit integrates a temperature sensor, humidity sensor, wind speed sensor, salt spray concentration sensor, and conductivity detection module via an RS485 bus to achieve synchronous acquisition of multiple parameters.

[0065] The conductivity detection module includes a corrosion-resistant monitoring plate and a pair of detection electrodes disposed on the surface of the monitoring plate. The corrosion-resistant monitoring plate can be made of corrosion-resistant alloy material, uses bolt mounting, and can be periodically disassembled and replaced; after replacement, the reference value needs to be recalibrated. The detection electrodes are fixedly mounted on the monitoring bracket and are in close contact with the surface of the monitoring plate. During operation, the resistance signal between the detection electrodes is collected and converted into surface conductivity using the following formula: ;

[0066] Where, in the formula, For surface conductivity, To detect resistance, This is an electrode structure constant, determined by the electrode spacing and electrode area;

[0067] Among the five types of parameters collected, salt spray concentration is used to reflect the intensity of salt input in the area, temperature, humidity and wind speed are used to reflect the environmental conditions for salt deposition, and surface conductivity is used to directly reflect the salt adhesion state on the monitored surface. A complete monitoring system is constructed from three dimensions: input, environment and results.

[0068] The data preprocessing module is used to first remove outliers and compensate for missing values ​​in the multi-source monitoring data, then extract the cumulative salt input and salt deposition change rate based on the processed time series data, and finally normalize the original monitoring data and the extracted feature quantities. The normalized multi-source monitoring data and feature quantities are then transmitted to the salt deposition index construction module and the state identification and prediction module.

[0069] The data preprocessing module receives the raw monitoring data uploaded by each modular monitoring unit and sequentially performs outlier removal, missing value compensation, and minimum-maximum normalization. Outlier removal adopts... The criteria include removing outlier data that exceeds a reasonable fluctuation range; missing value compensation uses linear interpolation to fill in short-term missing data; and normalization uses the minimum-maximum normalization method to uniformly map all physical quantities to... The general formula for intervals is: ;

[0070] in, , For the historical minimum and maximum values ​​of the corresponding physical quantities, after normalization ;

[0071] After preprocessing, two core features are extracted: cumulative salt input and salt deposition rate.

[0072] Cumulative salt input: Calculated based on the time-series data of salt spray concentration within a set time window, using the following formula: ;

[0073] in, The number of samples within the time window, with a total window duration of [value missing]. , The sampling time interval, The first time before the current moment Salt spray concentration at each sampling point;

[0074] Salt deposition rate: calculated based on the temporal change in surface conductivity, using the following formula: : ;

[0075] in, The surface conductivity at the previous sampling time;

[0076] The preprocessed raw monitoring data and extracted features are transmitted together to the salt deposition index construction module and the state identification and prediction module.

[0077] The salt deposition index construction module is used to weight and fuse the cumulative salt input, the coupling characteristics of salt spray concentration and humidity, wind speed characteristics and surface conductivity characteristics to construct a comprehensive salt deposition index.

[0078] The salt deposition index construction module receives preprocessed data and feature values, and constructs a comprehensive salt deposition index through multi-feature weighted fusion to quantify the degree of salt deposition in each region. The calculation expression for the comprehensive salt deposition index is as follows: ;

[0079] in, To normalize the cumulative salt input, Normalized salt spray concentration, To normalize humidity, To normalize the wind speed, wind speed affects the diffusion and deposition of salt spray. To normalize the surface conductivity, which directly reflects the actual amount of salt adhering, , , , Let be the weight coefficients of each feature term, satisfying The weighting coefficients are determined using a data-driven method: historical multi-source monitoring data is used as input, and manually detected salt deposition levels are used as labels. The characteristic importance of each feature to the salt deposition state is calculated, and the corresponding weight values ​​are obtained after normalization. In an exemplary embodiment, the weight value is taken as... , , , The higher the value of the salt deposition comprehensive index, the higher the degree of salt deposition in the corresponding area.

[0080] because It is a coupling term of salt spray concentration and humidity. High humidity environment will promote the condensation and adhesion of salt spray on the surface. The product of the two can quantify the environmental promoting effect of salt deposition.

[0081] The status identification and prediction module is used to identify the current salt deposition level based on monitoring data and extracted features, and to predict the salt deposition development trend based on the comprehensive salt deposition index and the rate of change; the risk assessment module is used to calculate the salt deposition risk value by combining the current comprehensive salt deposition index, the rate of change, and the prediction results.

[0082] The graded early warning output module is used to classify risks based on risk values ​​and output monitoring results and graded early warning information.

[0083] As a preferred embodiment, modular monitoring units are deployed in each salt deposition sensitive area of ​​the seawater cooling tower. Each monitoring unit integrates a temperature sensor, a humidity sensor, a wind speed sensor, a salt spray concentration sensor, and a conductivity detection module.

[0084] The conductivity detection module includes a corrosion-resistant monitoring sheet and detection electrodes set on the surface of the monitoring sheet, which are used to collect surface conductivity signals of the monitoring area; salt spray concentration is used to reflect the salt input intensity, temperature, humidity and wind speed are used to reflect the salt deposition environment, and surface conductivity is used to reflect the salt adhesion state of the monitoring surface.

[0085] In a preferred embodiment, the state recognition and prediction module includes a random forest classification unit and a trend prediction unit;

[0086] The random forest classification unit adopts an offline training and online update working mode. The model input is a 7-dimensional feature vector, including temperature, humidity, wind speed, salt spray concentration, surface conductivity, cumulative salt input, and salt deposition change rate. The output is four salt deposition levels: low salt deposition, general salt deposition, relatively high salt deposition, and high salt deposition.

[0087] The labels for the training samples were obtained through manual calibration: the corrosion-resistant monitoring plates in each area were periodically disassembled, the mass before and after deposition was weighed, and the mass of salt deposition per unit area was calculated. : ;

[0088] in, To monitor the quality of the deposited sample, To monitor the initial quality of the film, The effective area of ​​the monitoring patch;

[0089] Based on the salt deposition quality per unit area, four levels are defined, with an example threshold of: low salt deposition. Salt deposits High salt deposition High-salinity sediments .

[0090] When the system is deployed for the first time and historical training data is insufficient, a rule-based initialization criterion is used for state identification. The system is classified based on a simplified comprehensive salt deposition index. An example initialization formula is as follows: ;

[0091] in, To normalize the cumulative salt input, Normalized salt spray concentration, To normalize humidity, To normalize wind speed, To normalize the surface conductivity, , , , Let be the weight coefficients of each feature term, satisfying ;

[0092] Once the number of accumulated labeled training samples reaches a preset threshold, the system automatically triggers offline training of the random forest model. After training is completed, it smoothly switches to a data-driven recognition mode, and can continuously receive new samples for incremental updates.

[0093] The trend prediction unit calculates the predicted salt deposition index for future times through short-cycle linear extrapolation based on the current comprehensive salt deposition index and the rate of salt deposition change. Based on the numerical range of the rate of salt deposition change, the salt deposition development trend is divided into four states: continuous increase, slow increase, basically stable, and gradual decrease.

[0094] The trend prediction unit calculates the predicted salt deposition index for future times using short-period linear extrapolation based on the current comprehensive salt deposition index and the rate of salt deposition change. The formula is as follows: ;

[0095] in, For the predicted duration, the value range is 1–24 hours. Since the salt deposition process in seawater cooling towers is time-varying due to environmental influences, the salt spray input changes gradually and the deposition process is continuous within a short period, resulting in approximately stable conductivity changes. Therefore, linear extrapolation within a 1–24 hour timescale is engineeringly reasonable. The system classifies the salt deposition rate into four development trends based on the numerical range: a rate of change higher than a first preset threshold indicates a continuously increasing state; a rate of change greater than 0 and less than or equal to the first preset threshold indicates a slowly increasing state; an absolute value of the rate of change lower than a second preset threshold indicates a basically stable state; and a rate of change less than 0 indicates a gradually decreasing state.

[0096] The risk assessment module calculates the salt deposition risk value by combining the current salt deposition status, rate of change, and prediction results. The expression for calculating the risk value is: ;

[0097] in, The normalized salt deposition rate was obtained through minimum-maximum normalization. For the weighting coefficients, satisfying These correspond to the weighting percentages of the current state, the trend of change, and the future prediction, respectively. For predicting duration The comprehensive salt deposition index was obtained by short-period linear extrapolation, where the prediction duration was... 1-24 hours;

[0098] The risk assessment module incorporates a dynamic threshold adjustment unit, generating dynamic early warning thresholds based on risk value data from historical normal operation phases in each monitoring area. Specifically, it takes the historical risk values ​​of each monitoring area over a fixed time window and calculates their average. with standard deviation ; ; ;

[0099] Through formula Generate dynamic thresholds for corresponding levels, where k is the safety factor. Different warning levels correspond to different k values. The dynamic thresholds can be automatically adjusted according to the season, sea state, and operating conditions to adapt to the salt deposition patterns under different environments.

[0100] When historical risk data is insufficient, a preset initial static threshold is used to classify risk levels. Once sufficient data has been accumulated, the system will automatically switch to a dynamic threshold.

[0101] The graded early warning output module compares the calculated salt deposition risk value with the dynamic early warning threshold, classifying it into four risk levels: normal state, attention state, early warning state, and severe early warning state. When historical data is insufficient during the initial system deployment, a preset initial static threshold can be used for grading; once sufficient data has accumulated, it will automatically switch to a dynamic threshold.

[0102] Finally, the monitoring results of all monitored areas are summarized, and the current salt deposition level, salt deposition change rate, salt deposition development trend, predicted salt deposition status, salt deposition risk level, and key salt deposition areas are output. This forms the regional distribution results of salt deposition status in seawater cooling towers, locates key areas with high deposition levels or rapid growth, and provides data support for operation and maintenance decisions.

[0103] Example 2

[0104] Based on Example 1, the solution in Example 1 will be further described in detail below with reference to the specific working method, such as... Figures 1 to 4 As shown below, see details:

[0105] A smart monitoring method for salt deposition in seawater cooling towers includes the following step S1: Based on the structural characteristics, airflow distribution patterns, and salt deposition-prone areas of the seawater cooling tower, the interior of the cooling tower is divided into multiple independent salt deposition monitoring zones; modular monitoring units are deployed in each monitoring zone to simultaneously collect five types of monitoring parameters: temperature, humidity, wind speed, salt spray concentration, and surface conductivity, forming a multi-source monitoring data sequence. : ;

[0106] in, It is the first Each monitoring area is in Multi-source monitoring data vectors at any given time For the first One salt deposition monitoring area, For the current moment, Air temperature, unit , Humidity, in % Wind speed, unit , Salt spray concentration, unit , Surface conductivity, in units .

[0107] Step S2: First, outlier removal and missing value compensation are performed on the multi-source monitoring data. Then, the cumulative salt input is calculated based on the processed salt spray concentration time series data. The salt deposition change rate is calculated based on the processed surface conductivity time series data. Finally, all monitoring data and extracted feature quantities are normalized and preprocessed.

[0108] Outlier removal, missing value compensation, and minimum-maximum normalization were performed on the raw multi-source monitoring data; the cumulative salt input was calculated based on the salt spray concentration data within the set time window. : ;

[0109] in, The number of samples within the time window. The sampling time interval, The first time before the current moment Salt spray concentration at each sampling point;

[0110] Then, based on the temporal variation of surface conductivity, the rate of salt deposition is calculated. : ;

[0111] in, The surface conductivity at the previous sampling time is denoted as .

[0112] Step S3: Assign corresponding weights to the normalized cumulative salt input, the coupling characteristics of salt spray concentration and humidity, wind speed characteristics and surface conductivity characteristics, and construct a comprehensive salt deposition index by weighted fusion;

[0113] ;

[0114] in, To normalize the cumulative salt input, Normalized salt spray concentration, To normalize humidity, To normalize wind speed, To normalize the surface conductivity, , , , Let be the weight coefficients of each feature term, satisfying The higher the comprehensive salt deposition index, the higher the degree of salt deposition in the corresponding area.

[0115] Step S4: Combine the preprocessed multi-source monitoring data, the extracted cumulative salt input, and the salt deposition change rate to form a feature vector, input it into the trained classification model, identify the current salt deposition level, and predict the salt deposition development trend based on the comprehensive salt deposition index and change rate.

[0116] The pre-processed temperature, humidity, wind speed, salt spray concentration, and surface conductivity, along with the extracted cumulative salt input and salt deposition rate, form a 7-dimensional feature vector. This vector is then input into a pre-trained random forest model, which outputs the salt deposition level of the current area, categorized into four types: low-salt deposition, moderate-salt deposition, relatively high-salt deposition, and high-salt deposition. Simultaneously, based on the current comprehensive salt deposition index and its rate of change, a predicted comprehensive salt deposition index for a short period of 1–24 hours is calculated using linear extrapolation.

[0117] ;

[0118] The prediction results are used to determine whether salt deposits are in a state of continuous growth.

[0119] Step S5: Calculate the salt deposition risk value for each monitoring area by combining the current comprehensive salt deposition index, the rate of salt deposition change, and the predicted comprehensive salt deposition index;

[0120] ;

[0121] in, For the weighting coefficients, satisfying ;

[0122] Simultaneously, based on historical risk value data for each monitoring area, the historical mean and standard deviation are calculated, and dynamic early warning thresholds are generated by combining the safety coefficients of the corresponding early warning levels.

[0123] Step S6: Compare the salt deposition risk value with the dynamic early warning threshold or the preset initial threshold, and classify it into four risk levels: normal, attention, early warning, and severe early warning. Summarize the monitoring data of all monitoring areas, and output the current salt deposition degree, salt deposition change rate, salt deposition development trend, predicted salt deposition status, salt deposition risk level, and information on key salt deposition areas to form a complete salt deposition monitoring and early warning result.

Claims

1. A smart monitoring system for salt deposition in seawater cooling towers, characterized in that, It includes a modular monitoring unit, a data preprocessing module, a salt deposition index construction module, a state identification and prediction module, a risk assessment module, and a graded early warning output module; The modular monitoring unit is used to collect multi-source monitoring data in multiple salt deposition monitoring areas of the seawater cooling tower and transmit it to the data preprocessing module. The data preprocessing module is used to first remove outliers and compensate for missing values ​​in the multi-source monitoring data, then extract the cumulative salt input and salt deposition change rate based on the processed time series data, and finally perform normalization preprocessing on the original monitoring data and the extracted feature quantities. The normalized multi-source monitoring data and feature quantities are then transmitted to the salt deposition index construction module and the state recognition and prediction module. The salt deposition index construction module is used to weight and fuse the cumulative salt input, the coupling characteristics of salt spray concentration and humidity, wind speed characteristics and surface conductivity characteristics to construct a comprehensive salt deposition index. The state identification and prediction module is used to identify the current salt deposition level based on monitoring data and extracted feature quantities, and to predict the salt deposition development trend based on the comprehensive salt deposition index and the rate of change; the risk assessment module is used to calculate the salt deposition risk value by combining the current comprehensive salt deposition index, the rate of change and the prediction results. The graded early warning output module is used to classify risks according to risk values ​​and output monitoring results and graded early warning information.

2. The intelligent monitoring system for salt deposition in seawater cooling towers according to claim 1, characterized in that: The modular monitoring units are deployed in each salt deposition sensitive area of ​​the seawater cooling tower. Each monitoring unit integrates a temperature sensor, a humidity sensor, a wind speed sensor, a salt spray concentration sensor, and a conductivity detection module. The conductivity detection module includes a corrosion-resistant monitoring sheet and a detection electrode disposed on the surface of the monitoring sheet, used to collect surface conductivity signals of the monitoring area; the salt spray concentration is used to reflect the salt input intensity, temperature, humidity and wind speed are used to reflect the salt deposition environment, and surface conductivity is used to reflect the salt adhesion state of the monitoring surface.

3. The intelligent monitoring system for salt deposition in seawater cooling towers according to claim 1, characterized in that: The state recognition and prediction module includes a random forest classification unit and a trend prediction unit; The random forest classification unit takes a feature vector composed of temperature, humidity, wind speed, salt spray concentration, surface conductivity, cumulative salt input, and salt deposition change rate as input, and outputs four state identification results: low salt deposition, general salt deposition, relatively high salt deposition, and high salt deposition. The trend prediction unit calculates the predicted salt deposition index for future times through short-cycle linear extrapolation based on the current salt deposition comprehensive index and the salt deposition change rate. Based on the numerical range of the salt deposition change rate, the salt deposition development trend is divided into four states: continuous increase, slow increase, basically stable, and gradual decrease.

4. The intelligent monitoring system for salt deposition in seawater cooling towers according to claim 1, characterized in that: The risk assessment module has a built-in dynamic threshold adjustment unit. The dynamic threshold adjustment unit is used to calculate the mean and standard deviation of historical risk values ​​based on the salt deposition risk value data of each monitoring area during the historical normal operation phase, and generate a dynamic warning threshold by combining the safety factor of the corresponding warning level. When historical risk data is insufficient, a preset initial threshold is used to classify risk levels.

5. A smart monitoring method for salt deposition in seawater cooling towers, characterized in that: Step S1: Simultaneously collect five types of multi-source monitoring data, namely temperature, humidity, wind speed, salt spray concentration and surface conductivity, in multiple salt deposition monitoring areas of the seawater cooling tower; Step S2: First, outlier removal and missing value compensation are performed on the multi-source monitoring data. Then, the cumulative salt input is calculated based on the processed salt spray concentration time series data. The salt deposition change rate is calculated based on the processed surface conductivity time series data. Finally, all monitoring data and extracted feature quantities are normalized and preprocessed. Step S3: Assign corresponding weights to the normalized cumulative salt input, the coupling characteristics of salt spray concentration and humidity, wind speed characteristics and surface conductivity characteristics, and construct a comprehensive salt deposition index by weighted fusion; Step S4: Combine the preprocessed multi-source monitoring data, the extracted cumulative salt input, and the salt deposition change rate to form a feature vector, input it into the trained classification model, identify the current salt deposition level, and predict the salt deposition development trend based on the comprehensive salt deposition index and change rate. Step S5: Calculate the salt deposition risk value for each monitoring area by combining the current comprehensive salt deposition index, the rate of salt deposition change, and the predicted comprehensive salt deposition index; Step S6: Compare the salt deposition risk value with the corresponding threshold to classify the risk level, and output the salt deposition monitoring results, graded early warning information and key salt deposition areas for each region.

6. The intelligent monitoring method for salt deposition in seawater cooling towers according to claim 5, characterized in that: Step S1 specifically involves: Based on the structural characteristics, airflow distribution, and salt deposition-prone areas of the seawater cooling tower, the cooling tower was divided into multiple independent salt deposition monitoring zones. Modular monitoring units were deployed in each monitoring zone to simultaneously collect five types of monitoring parameters, forming a multi-source monitoring data sequence. : ； in, It is the first Each monitoring area is in Multi-source monitoring data vectors at any given time For the first One salt deposition monitoring area, For the current moment, For temperature, For humidity, For wind speed, Salt spray concentration, denoted as surface conductivity.

7. The intelligent monitoring method for salt deposition in seawater cooling towers according to claim 5, characterized in that: Step S2 specifically involves: Outlier removal, missing value compensation, and minimum-maximum normalization were performed on the raw multi-source monitoring data; the cumulative salt input was calculated based on the salt spray concentration data within the set time window. : ； in, The number of samples within the time window. The sampling time interval, The first time before the current moment Salt spray concentration at each sampling point; The rate of salt deposition is calculated based on the temporal variation of surface conductivity. : ； in, The surface conductivity at the previous sampling time; After feature extraction, the original monitoring data, cumulative salt input, and salt deposition change rate were uniformly normalized using minimum and maximum methods.

8. The intelligent monitoring method for salt deposition in seawater cooling towers according to claim 5, characterized in that: In step S3, the formula for calculating the comprehensive salt deposition index is: ； in, To normalize the cumulative salt input, Normalized salt spray concentration, To normalize humidity, To normalize wind speed, To normalize the surface conductivity, , , , Let be the weight coefficients of each feature term, satisfying ; The weighting coefficients are determined as follows: using historical multi-source monitoring data as input and manually detected salt deposition levels as labels, the influence of each feature on the salt deposition state is calculated using a random forest model, and the corresponding weight values ​​are obtained after normalization. The larger the comprehensive salt deposition index value, the higher the degree of salt deposition in the corresponding area.

9. The intelligent monitoring method for salt deposition in seawater cooling towers according to claim 5, characterized in that: Step S4 specifically involves: A 7-dimensional feature vector is composed of temperature, humidity, wind speed, salt spray concentration, surface conductivity, cumulative salt input, and salt deposition change rate. This vector is then input into a pre-trained random forest model, which outputs the salt deposition level of the current area. The training labels of the random forest model are calibrated in the following way: the corrosion-resistant monitoring sheet is periodically disassembled, its surface salt deposition mass is measured, the salt deposition mass per unit area is calculated, and it is divided into four salt deposition levels: low, average, relatively high, and high, according to a preset threshold. When the system is deployed for the first time and there is insufficient historical training data, the system uses a rule-based initialization criterion based on the comprehensive salt deposition index for state identification. When the number of accumulated labeled training samples reaches a preset threshold, the system triggers offline training of the random forest model and smoothly switches to the data-driven identification mode. Meanwhile, based on the current comprehensive salt deposition index and the rate of salt deposition change, the predicted comprehensive salt deposition index for future times is calculated through short-period linear extrapolation. Based on the numerical range of the rate of salt deposition change, the development trend of salt deposition is divided into four states: continuous increase, slow increase, basically stable, and gradual decrease.

10. The intelligent monitoring method for salt deposition in seawater cooling towers according to claim 5, characterized in that: Steps S5 and S6 specifically involve: calculating the salt deposition risk value by combining the current comprehensive salt deposition index, the normalized salt deposition change rate, and the predicted comprehensive salt deposition index. ； in, This represents the normalized rate of salt deposition change. Let be the weighting coefficient, satisfying , For predicting duration The subsequent salt deposition comprehensive index is calculated by short-period linear extrapolation, wherein the prediction duration is... 1-24 hours; Based on historical risk value data for each monitoring area, the historical mean and standard deviation are calculated, and dynamic warning thresholds are generated by combining the safety factor of the corresponding warning level. The salt deposition risk value is compared with the dynamic early warning threshold or the preset initial threshold, and the risk is divided into four levels: normal, attention, early warning and severe early warning. The system summarizes the monitoring results of all monitored areas and outputs the current salt deposition level, salt deposition change rate, development trend, predicted status, risk level, and information on key salt deposition areas.

Images