Cooling tower intelligent self-cleaning and scale control method

Abstract

The application provides a cooling tower intelligent self-cleaning anti-fouling control method, relates to the technical field of anti-fouling maintenance control, and obtains the time sequence data of water temperature, flow velocity, conductivity and pH value of multiple monitoring points in the cooling tower, performs fouling dynamics reasoning after spatial gridding, obtains the scale layer growth characteristics of each grid unit containing instantaneous and cumulative risks, comprehensively sorts the characteristics of each grid unit, generates a scale layer growth intensity spatial distribution map, identifies a high scale accumulation area and counts a spatial aggregation degree, matches the spatial aggregation degree with historical cleaning records, constructs a mapping function to generate a differentiated partition cleaning scheme, generates the self-cleaning triggering time and cleaning duration of each area according to the partition scheme, preferentially triggers the high scale accumulation area, and converts the cleaning intensity parameters into control instructions for output. The application realizes dynamic evaluation of the fouling risk and accurate distribution of the cleaning intensity, improves the cleaning efficiency, and prolongs the service life of the equipment.

Description

Technical Field

[0001] This invention relates to the field of scale prevention and control technology, and more specifically, to an intelligent self-cleaning scale prevention and control method for cooling towers. Background Technology

[0002] With the continuous advancement of industrialization and the widespread application of heat exchange technology, cooling towers, as a common industrial device, have been widely used in industries such as power, chemical, metallurgy, and high-end manufacturing. Cooling towers remove heat from high-temperature systems through water circulation and evaporation, effectively ensuring the normal operation of industrial equipment and the thermal balance of the working environment. However, during long-term operation, cooling towers typically undergo complex physicochemical processes such as water quality changes, high-temperature evaporation, and material deposition, leading to uneven scale buildup on the inner walls of the water circulation system and components like packing. This results in increased thermal resistance, decreased flow rate, and reduced energy efficiency. To address the scaling problem in cooling towers, current research and practice have proposed various technical methods. For example, physical methods include using high-frequency oscillation and ultrasonic oscillation devices to interfere with the scaling process; chemical methods mainly utilize water quality regulators and scale inhibitors to delay scale formation; and mechanical methods directly remove scale through periodic cleaning. These methods have improved the operating efficiency of cooling towers and slowed down the scaling rate to some extent, but their versatility, economy, and intelligent features still need further optimization.

[0003] Existing scaling prevention and self-cleaning technologies for cooling towers still face a series of pressing technical challenges. On one hand, the distribution of scale inside cooling towers exhibits high spatiotemporal non-uniformity, and current technologies fail to specifically assess the spatial distribution and temporal dynamics of scaling risk, lacking effective monitoring and early warning mechanisms. On the other hand, conventional cleaning methods typically employ fixed-cycle, full-coverage cleaning operations, failing to focus on areas with high scale buildup, resulting in significant waste of cleaning resources and poor cleaning effectiveness in these areas. Furthermore, the cleaning intensity or scale inhibitor dosage schemes relied upon by some technologies lack scientific basis, making it difficult to effectively match actual scaling conditions for precise control. In particular, with the continuous increase in the scale of cooling towers and the increasing complexity of their operating environments, the technical requirements for regional dynamic control and global self-cleaning optimization are becoming increasingly prominent. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides an intelligent self-cleaning and anti-scaling control method for cooling towers. This method can, to some extent, solve the problems of wasted cleaning resources, poor cleaning effect in areas with high scale buildup, and lack of precision in cleaning control caused by the inability to accurately determine the scale buildup inside the cooling tower, the failure to dynamically assess the scale buildup risk in different areas and rationally allocate cleaning efforts.

[0005] According to one aspect of the present invention, a method for intelligent self-cleaning and scale prevention control of cooling towers is provided, comprising:

[0006] Time-series data of water temperature, flow rate, conductivity, and pH value from multiple monitoring points within the cooling tower are acquired, and after spatial gridding, scale formation kinetics are inferred to obtain scale growth characteristics of each grid cell containing instantaneous and cumulative risk information.

[0007] The peak amplitude and cumulative integral of each grid cell are sorted by magnitude to generate a spatial distribution map of scale growth intensity. High scale accumulation areas are identified and their spatial continuous distribution range and area ratio are statistically analyzed to obtain the scale spatial aggregation degree.

[0008] By matching the spatial aggregation degree of scale with historical cleaning effect records, a mapping function between cleaning intensity and scale removal rate is constructed to generate a zoned cleaning scheme that assigns different cleaning intensities to areas with different scale accumulation levels.

[0009] Based on the zonal cleaning plan, the self-cleaning trigger time and cleaning duration for each zone are generated. The self-cleaning operation of high-fouling zones is triggered first, and the cleaning intensity parameters are converted into control commands and output to the actuator.

[0010] Furthermore, the scaling kinetics reasoning is based on the principles of solution chemical equilibrium and crystallization kinetics to calculate the supersaturation of scaling ions and the driving force for crystallization;

[0011] Taking calcium carbonate as an example, the formula for calculating supersaturation is as follows:

[0012]

[0013] in, For a moment Location The supersaturation of calcium carbonate, This is the ion activity correction factor. This refers to the calcium ion concentration. This represents the carbonate ion concentration. For temperature The solubility product constant under these conditions For pH values ​​deviating from the sensitivity parameter, This is a damping parameter representing the effect of flow rate on ion activity. This represents the difference between the measured pH value and the precipitation equilibrium pH value. The velocity is the water flow velocity.

[0014] Furthermore, the formula for calculating the crystallization driving force is expressed as follows:

[0015]

[0016] in, As the reference crystallization rate constant, This is the molar mass of calcium carbonate. The nonlinear response exponent of supersaturation to crystallization rate is given. For nucleation activation energy, This is the universal gas constant. Water temperature The critical flow velocity is... The effective diffusion coefficient of ions is . This is the Peckley number correction factor for mass transfer. The characteristic length is denoted as .

[0017] Furthermore, the overall magnitude ranking includes:

[0018] The crystallization driving force calculated point by point at each time point is connected in chronological order to generate a crystallization driving force time series curve;

[0019] The peak amplitude is extracted based on the crystallization driving force time-series curve; the cumulative integral is obtained by integrating the crystallization driving force time-series curve over time.

[0020] The peak amplitude and the cumulative integral are normalized respectively, and then a weighted sum is calculated to form a scale level score; the weights of the weighted sum are determined by historical scale thickness distribution data;

[0021] The Pearson correlation coefficient was calculated between the scoring results of different weight combinations and the actual scale thickness. The weight combination with the highest correlation coefficient was selected, and all grid cells were sorted from high to low according to the comprehensive scale level score.

[0022] Furthermore, the spatial aggregation degree of the scale layer is defined as the number of continuous regions divided by the average number of grid cells contained in a single continuous region;

[0023] The continuous region is determined by spatial connectivity analysis, and the criterion is that the absolute value of the coordinate difference between adjacent grid cells in three directions does not exceed 1.5 times the size of a single grid cell.

[0024] Furthermore, the formula for calculating the mapping function is as follows:

[0025]

[0026] in, The scale removal rate predicted by the mapping function. To comprehensively measure cleaning intensity parameters, This represents the theoretical upper limit of the scale removal rate when the comprehensive cleaning intensity parameter approaches infinity. For the steepness coefficient of the logic function, The location of the midpoint of the logic function. The coefficient of the logarithmic correction term. The saturation suppression coefficient of the logarithmic correction term. This is the confidence adjustment coefficient for historical data. This is the arithmetic mean of the comprehensive cleaning intensity parameters in the historical cleaning records; This represents the standard deviation of the comprehensive cleaning intensity parameter in historical cleaning records.

[0027] Furthermore, the self-cleaning trigger time is generated by real-time monitoring of the scale growth rate in each continuous area. When the scale growth rate reaches the cleaning start threshold of the corresponding scale level, it is marked as a state to be cleaned. If multiple areas reach the threshold at the same time, they are sorted from high to low according to the scale level. The first area in the sorted order immediately triggers self-cleaning, and the trigger time of subsequent areas is successively superimposed with the cleaning duration of the previous area and the equipment transfer time.

[0028] According to another aspect of the present invention, a cooling tower intelligent self-cleaning and anti-scaling control system is provided, comprising:

[0029] The data acquisition module is used to acquire time-series data of water temperature, flow rate, conductivity, and pH value at multiple monitoring points in the cooling tower. After spatial gridding, it performs scale formation kinetics inference to obtain scale growth characteristics of each grid cell containing instantaneous and cumulative risk information.

[0030] The fouling inference module is used to sort the peak amplitude and cumulative integral of each grid cell by magnitude, generate a spatial distribution map of fouling growth intensity, identify high fouling areas and count their spatial continuous distribution range and area ratio to obtain the spatial aggregation degree of fouling.

[0031] The scheme generation module is used to match the spatial aggregation degree of scale with historical cleaning effect records, construct a mapping function between cleaning intensity and scale removal rate, and generate a zoned cleaning scheme that assigns different cleaning intensities to areas with different scale levels.

[0032] The execution control module is used to generate the self-cleaning trigger time and cleaning duration for each area according to the zoning cleaning plan, prioritize the self-cleaning operation of areas with high fouling, and convert the cleaning intensity parameters into control commands to be output to the actuator.

[0033] Compared with existing technologies, this invention acquires time-series data on water temperature, flow rate, conductivity, and pH value from multiple monitoring points within the cooling tower. It then performs spatial gridding analysis to deduce the scale growth characteristics of each grid unit. Simultaneously, it generates zoned cleaning schemes for high-scale-accumulation areas and optimizes them by combining historical cleaning performance records. This enables dynamic assessment of scaling risk and precise allocation of cleaning intensity for different areas, resulting in more rational allocation of cleaning resources, improved cleaning efficiency, and effective mitigation of the impact of scale accumulation on cooling tower performance, thus extending equipment lifespan. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0035] Figure 1 A flowchart of a cooling tower intelligent self-cleaning and scale prevention control method according to an embodiment of the present invention;

[0036] Figure 2 This is a spatial distribution diagram of scale growth intensity according to an embodiment of the present invention. Detailed Implementation

[0037] Hereinafter, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0038] As mentioned in the background section above, the existing technology has two main problems: First, due to the highly non-uniform spatial distribution of scale inside the cooling tower, the existing technology lacks dynamic sensing and accurate identification of high scale accumulation areas, resulting in blind cleaning operations and difficulty in achieving rational use of resources; Second, most existing scale prevention and cleaning control methods adopt fixed cycle and full-coverage cleaning modes, failing to reasonably match the cleaning intensity with the actual scale accumulation risk, resulting in poor cleaning effect, especially in the difficulty of effectively removing high scale accumulation areas.

[0039] Figure 1 This is a system block diagram of a cooling tower intelligent self-cleaning and scale prevention control method according to an embodiment of the present invention. Figure 1 and Figure 2 As shown, the intelligent self-cleaning and scale prevention control method for cooling towers includes:

[0040] S1: Acquire time-series data of water temperature, flow rate, conductivity, and pH value from multiple monitoring points within the cooling tower. After spatial gridding, perform scale formation kinetics inference to obtain scale growth characteristics for each grid cell, which includes information on instantaneous and cumulative risks.

[0041] The control system receives four types of time-series data—water temperature, flow rate, conductivity, and pH value—from multiple monitoring points distributed at different heights and orientations on the cooling tower. Each record is accompanied by the spatial coordinates and timestamp of the monitoring point.

[0042] Upon receiving the data, the measurements from all monitoring points at the same time are merged to form a spatial snapshot of that moment. Then, spatial partitioning and mapping are performed, pre-dividing the internal space of the cooling tower into multiple grid cells based on height, radial distance, and azimuth. Monitoring points are assigned to their corresponding grid cells according to their spatial coordinates. When multiple monitoring points exist within a grid cell, the simultaneous measurements from all monitoring points within that cell are spatially weighted and averaged to obtain the representative value for that grid. For grid cells without monitoring points, spatial interpolation is performed using known values ​​from adjacent grids. This transforms the discrete monitoring point time-series records into a four-dimensional spatial distribution time-series matrix composed of all grid cells. Each element of the matrix represents a vector of water temperature, flow velocity, conductivity, and pH value for that grid at a given moment. To ensure interpolation accuracy, monitoring points are preferentially placed in critical areas prone to scaling, such as the surface of the packing material, the distribution tank, and the walls of the collection pool. The grid granularity is adjusted based on the simulation results of the circulating water flow field.

[0043] After completing the spatial partitioning mapping, scaling kinetics generative inference is performed on the temporal data of each grid cell. The scaling kinetics generative inference is based on the principles of solution chemical equilibrium and crystallization kinetics to calculate the supersaturation of scaling ions and the crystallization driving force time-by-time.

[0044] Specifically: for any time and spatial grid position First, the concentration distribution of various scale-forming ions, such as calcium ions, magnesium ions, sulfate ions, phosphate ions, and silicate ions, is calculated from the conductivity. Then, the correlation model is dynamically corrected using online hardness monitoring or periodic sampling and testing data. Simultaneously, the water temperature is considered. Determine the solubility product constant of calcium carbonate. The difference between the measured pH values ​​and the equilibrium pH values ​​of various scale types is used to obtain the equilibrium pH value. Then, the supersaturation degree of scale-forming ions for each scale type was calculated. Taking calcium carbonate as an example, the supersaturation degree was... The calculation formula is expressed as:

[0045]

[0046] in, For a moment Location The supersaturation of calcium carbonate, This is the ion activity correction factor. This refers to the calcium ion concentration. This represents the carbonate ion concentration. For temperature The solubility product constant under these conditions For pH values ​​deviating from the sensitivity parameter, This is a damping parameter representing the effect of flow rate on ion activity. This represents the difference between the measured pH value and the precipitation equilibrium pH value. The velocity is the water flow velocity.

[0047] Similarly, for other scale types such as calcium sulfate and calcium phosphate, their supersaturation is calculated using their respective ion concentration product and solubility product constants, expressed in the same form.

[0048] It should be noted that the conductivity correction factor The ion activity coefficients are calculated using ion activity theory combined with the Debye-Hückel equation. Specifically, the total ionic strength of the solution is estimated based on conductivity measurements, and then the ionic strength is substituted into the extended Debye-Hückel formula to calculate the ion activity coefficients; the solubility product constant is also calculated. The standard solubility product values ​​of calcium carbonate at different temperatures were obtained by consulting the NIST Chemical Thermodynamics Database or the CRC Chemical Physics Handbook; pH sensitivity parameters were also obtained. Scaling rates were measured under different pH conditions through a series of pH gradient experiments, and the scaling rates were obtained by fitting a semi-logarithmic coordinate to the pH deviation; flow velocity damping parameters were also measured. The determination was made by fitting the data from the variable flow rate fouling experiment.

[0049] Based on supersaturation, the crystallization driving force, which characterizes the potential for scale mass growth per unit time and unit surface area, is calculated. Taking calcium carbonate as an example, the crystallization driving force is... The calculation formula is expressed as:

[0050]

[0051] in, As the reference crystallization rate constant, This is the molar mass of calcium carbonate. This refers to the calcium ion concentration. The nonlinear response exponent of supersaturation to crystallization rate is given. For nucleation activation energy, This is the universal gas constant. Water temperature The critical flow velocity is... The effective diffusion coefficient of ions is . This is the Peckley number correction factor for mass transfer. The characteristic length is denoted as .

[0052] Similarly, for other scale types, their respective crystallization driving forces are calculated in the same form using their corresponding molar mass, calcium ion concentration (or corresponding cation concentration), supersaturation, and kinetic parameters.

[0053] It should be noted that the reference crystallization rate constant in the crystallization driving force formula... The nonlinear response index was obtained by monitoring the decay rate of calcium ion concentration and inferring it from the specific surface area of ​​the seed crystals through a calcium carbonate crystallization kinetic experiment under isothermal and supersaturation conditions; The nucleation activation energy was obtained by double logarithmic fitting of scale formation experimental data under different supersaturation conditions. Nucleation rate was measured using a variable-temperature crystallization experiment and fitted using the Arrhenius equation; critical flow rate The abrupt change point in scale growth rate was determined by observing flow state transition experiments; characteristic length. The dimensions are determined based on the heat exchanger geometry: for tubular heat exchangers, the inner diameter of the tubes is used; for plate heat exchangers, the plate spacing is used; and for spiral heat exchangers, the equivalent diameter of the flow channel is used.

[0054] After completing the above time-by-time calculations, for each grid cell, the values ​​of scale belonging to the same scale type at each time point throughout the entire observation period are calculated. By connecting them in chronological order, independent supersaturation time-series curves for each scale type are generated; similarly, for each scale type... They were also connected as independent crystallization driving force time series curves.

[0055] If only one type of scale has significant supersaturation within a grid cell, the crystallization driving force time series curve of that scale type is directly used as the final result for that cell.

[0056] If the supersaturation time series curves of multiple scale types all exhibit significant supersaturation characteristics, the inference engine further analyzes the crystallization competition relationship among the scale types to generate a comprehensive crystallization driving force time series curve. The specific competition judgment and processing methods are as follows:

[0057] First, a precipitation priority index is defined for each type of scale, which is defined as the supersaturation of that scale at the current moment. Its critical supersaturation The ratio, where the critical supersaturation is the minimum supersaturation required for the scale to crystallize detectably under given temperature and flow rate conditions, is obtained through preliminary experiments or literature data;

[0058] When the precipitation priority index of the dominant scale species is the highest and exceeds a preset threshold, the dominant scale species is determined to precipitate preferentially and occupy most of the surface active sites. For each of the other scale species, the difference between its precipitation priority index and that of the dominant scale species is compared. If the difference is within a preset range, the scale species is determined to have a co-precipitation competition relationship with the dominant scale species and is included in the co-precipitation weight allocation process. If the difference exceeds the preset range, the scale species is determined to be an inhibited scale species, and its driving force is adjusted using an attenuation coefficient. This attenuation coefficient is not a fixed constant but is dynamically calculated. ,in and These are the precipitation priority indices for dominant and suppressed scale species, respectively. To characterize the positive index of the inhibition intensity's sensitivity to dominance, it was obtained by fitting coprecipitation competition experimental data; the comprehensive crystallization driving force is obtained by linearly superimposing the total driving force of the dominant scale species, the driving force of the coprecipitation competition scale species allocated according to weights, and the driving force of other inhibited scale species after discounting according to their respective attenuation coefficients.

[0059] If the precipitation priority indices of multiple scale species differ within a preset range and there is no obvious single dominant scale species, it is determined to be co-precipitation competition. At this time, a weight is assigned to each competing scale species. The weight is equal to the supersaturation value of the scale species divided by the sum of the supersaturation values ​​of all competing scale species. Then, the crystallization driving force calculated separately for each scale species is multiplied by the corresponding weight and summed to obtain the comprehensive crystallization driving force.

[0060] For small-scale users lacking experimental calibration conditions, the inference engine can use the solubility product constants and crystallization kinetic parameters of calcium carbonate and calcium sulfate reported in the literature as initial values. In the initial stage of system operation, online adaptive calibration is performed through the following steps: Periodically acquire the actual scale thickness measurements at fixed monitoring points on the cooling tower's water collection tank wall or packing surface. Compare the measured values ​​with the integral of the comprehensive crystallization driving force calculated by the inference engine within the corresponding time period. Based on the comparison results, use a Kalman filter to recursively update the undetermined parameters in the empirical correlation model using least squares. After each update, correct the subsequent predicted values. After several calibration cycles, until the deviation between the predicted and measured values ​​stably converges within the preset error range, the adaptive tuning of the model parameters is achieved. If offline water sample analysis capabilities are available, collect representative water samples for small-scale laboratory scaling experiments, measuring the scaling rate under different combinations of temperature, pH, and ion concentration. Use the minimum root mean square error between the model's predicted and measured values ​​as the objective function, and adjust the baseline crystallization rate constant accordingly. Nonlinear response index Critical flow velocity Multivariate nonlinear least squares fitting is performed on key dynamic parameters to obtain a set of specific empirical parameters, which are then used to replace the initial values ​​in the literature to improve the prediction accuracy of the inference engine under specific water quality conditions.

[0061] After completing the above calculations for single or multiple scale species competitive coupling, each grid cell ultimately obtains a defined time-series curve of the comprehensive crystallization driving force, characterizing its scale growth potential. The supersaturation time-series curve is scanned to identify all local peaks. The amplitude of the highest peak is extracted as the peak amplitude, and the timestamp of its occurrence is recorded as the corresponding time. Numerical integration is performed on the comprehensive crystallization driving force time-series curve to calculate the cumulative integral over the entire observation period. If multiple scale species exist, the peak amplitude, corresponding time, and cumulative integral for each species are extracted, and the dominant scale species type is labeled.

[0062] Finally, the peak amplitude of each scale type, the time corresponding to the peak, and the cumulative integral of each grid cell are combined to form a scale growth feature vector of the cell, which contains information on instantaneous extreme risk and long-term cumulative risk, as well as scale type identification.

[0063] S2: The peak amplitude and cumulative integral of each grid cell are sorted by magnitude to generate a spatial distribution map of scale growth intensity, identify high scale accumulation areas and count their spatial continuous distribution range and area ratio to obtain the scale spatial aggregation degree.

[0064] Based on the scale growth feature vectors of each output spatial grid cell, the peak amplitude and cumulative integral are extracted as evaluation dimensions for comprehensive magnitude ranking.

[0065] Peak amplitude reflects the strongest instantaneous fouling risk experienced by the grid cell during the observation period. The larger the value, the more severe the oversaturation state has occurred, and the higher the possibility of rapid fouling growth in a short period of time. Cumulative integral reflects the continuous accumulation of fouling growth potential of the grid cell throughout the entire observation period. The larger the value, the longer the high fouling risk state is in the long term, and the greater the final accumulated fouling thickness.

[0066] If the feature vector contains information on multiple scale types, the peak amplitude is taken as the peak amplitude corresponding to the dominant scale type, and the cumulative integral is directly taken as the cumulative integral of the comprehensive crystallization driving force curve, so that the sorting input is consistent with the multi-scale coupling calculation logic.

[0067] To eliminate the dimensional and numerical scale differences between the peak amplitude and the cumulative integral, all grid cells are normalized, and their numerical ranges are mapped to the interval between zero and one. The same normalization operation is performed on the cumulative integral of all grid cells to obtain the normalized cumulative integral.

[0068] Based on this, the comprehensive fouling level score for each grid cell is calculated by weighted summation of the normalized peak amplitude and the normalized cumulative integral. The weights of the peak amplitude and the cumulative integral in the weighted summation are equal. The specific values ​​of these two weights are determined as follows: Actual fouling thickness distribution data for each grid cell within at least one historical operating cycle of the cooling tower are obtained as a calibration benchmark. The normalized peak amplitude and normalized cumulative integral generated within the same historical cycle are weighted and summed with different weight combinations to obtain multiple sets of candidate comprehensive fouling level scores. The Pearson correlation coefficient between each set of candidate scores and the actual fouling thickness of the corresponding grid cell is calculated. The weight combination with the highest Pearson correlation coefficient is selected as the weight setting value for formal operation. The traversal of weight combinations is achieved by exhaustively searching within the zero-to-one range with a set weight step size, or by using a gradient descent optimization algorithm to find the optimal weight combination. If historical fouling thickness distribution data is unavailable, the two weights are taken as equal values, ensuring that the comprehensive score simultaneously considers both instantaneous extreme risks and long-term cumulative risks.

[0069] The Pearson correlation coefficient is calculated as follows: when only some grid cells have actual scale thickness measurements, only the grid cells with actual scale thickness measurements are used to form a scoring sequence and a thickness sequence. Both sequences are arranged according to the same subset of grid numbers to ensure a one-to-one correspondence. All candidates of grid cells with actual scale thickness measurements are combined into a scale level score and arranged in order of grid number to form a scoring sequence. The actual scale thickness of the corresponding grid cells is arranged in order of the same grid number to form a thickness sequence. The two sequences are of equal length and correspond one-to-one. Calculate the arithmetic mean of the scoring sequence and the thickness sequence separately. Subtract the average of the scoring sequence from each value in the scoring sequence to obtain the scoring deviation sequence, and subtract the average of the thickness sequence from each value in the thickness sequence to obtain the thickness deviation sequence. Calculate the product of the scoring deviation and the thickness deviation for each grid cell, and sum the products of all grid cells to obtain the covariance numerator. Calculate the sum of squares of the deviation values ​​in the scoring deviation sequence and the sum of squares of the deviation values ​​in the thickness deviation sequence, multiply the two sums, and take the square root to obtain the covariance denominator. Divide the covariance numerator by the covariance denominator to obtain the Pearson correlation coefficient, which ranges from -1 to +1. The closer the absolute value of the correlation coefficient is to 1, the stronger the linear correlation between the candidate score and the actual thickness. Select the weight combination with the largest correlation coefficient as the optimal weight setting value. If multiple weight combinations have equal and maximum correlation coefficients, select the combination with the smallest difference between the peak amplitude weight and the cumulative integral weight to achieve a better balance between the two types of risks in the overall score.

[0070] After weighted summation, all grid cells are sorted in descending order of their overall scale level score. Grid cells at the top of the sort list represent areas with the most severe scale growth and require priority cleaning, while grid cells at the bottom represent areas with less severe scale growth and can be cleaned later or not at all.

[0071] To visually identify the spatial distribution pattern of scale growth within the cooling tower from the numerical ranking results, the overall magnitude ranking results are correlated and mapped with the spatial coordinate information of each grid unit. The spatial coordinates of each grid unit include three-dimensional parameters: tower height, radial position, and azimuth angle. These are combined into a scale magnitude score, which serves as the attribute value for that coordinate point. Spatial visualization maps the three-dimensional spatial coordinates and the overall score into a graphical representation, generating a spatial distribution map of scale growth intensity. This map uses a color-gradient approach to represent the spatial variation of the overall score: the maximum and minimum values ​​of the overall score for all grid units are statistically analyzed, and the range between the maximum and minimum values ​​is divided into five color-gradient intervals, corresponding to light blue, blue, yellow, orange, and red from low to high. Grid units with an overall score in the highest color-gradient interval (the fifth interval) are marked in red, grid units with an overall score in the lowest color-gradient interval (the first interval) are marked in light blue, and the middle three color-gradient intervals are marked in blue, yellow, and orange, respectively.

[0072] like Figure 2 The diagram illustrates the spatial distribution characteristics of scale growth in a cooling tower during operation. The entire cooling tower space is divided into 8 height levels, 6 radial zones, and 8 azimuth zones, totaling 384 grid cells. Each small square represents a grid cell, with color depth reflecting the degree of scale formation risk at that location. Observation shows that the bottom area of ​​the tower (lower vertical layer) exhibits more red and orange hues, indicating a higher risk of scale formation at the bottom, consistent with the physical characteristics of lower temperatures and higher supersaturation at the bottom during actual operation. The radially outer area (right side horizontally) also shows a darker color, reflecting the increased tendency for scale formation due to lower flow velocities and longer ion retention times at the edges. The statistical box on the right shows that in this example, the comprehensive score ranges from 0.385 to 0.989, the color level interval width is 0.121, the fifth color level threshold is 0.868, and 76 grid cells belong to the high-scale accumulation area, accounting for 19.8% of the total.

[0073] High-fouling regions are identified based on the spatial distribution map of fouling growth intensity. During the screening process, both the cutoff ratio and the scoring threshold are used simultaneously. Grid cells satisfying either method are selected to form a candidate set of high-fouling regions to avoid omissions by a single screening method.

[0074] Regarding the cutoff ratio, if historical cleaning records exist, the ratio of the number of grid cells that actually needed cleaning during the historical cleaning to the total number of grid cells is calculated, and this historical ratio is directly used as the cutoff ratio for the current operating cycle. If no historical cleaning records exist, the cutoff ratio is determined based on the ratio of the cooling tower's designed heat exchange load to the actual operating load. Specifically, the actual operating load is calculated by dividing the designed heat exchange load to obtain the load ratio. For example, if the baseline value for the load ratio is set to 0.8 and the baseline value for the cutoff ratio is set to 0.25, when the load ratio is greater than 0.8, the cutoff ratio is increased by multiplying the difference between the load ratio and 0.8 by 0.5. When the load ratio is less than 0.8, the cutoff ratio is decreased by multiplying the difference between 0.8 and the load ratio by 0.5. The final value of the cutoff ratio is limited to between 0.1 and 0.5. If the calculated result is less than 0.1, then 0.1 is used; if the calculated result is greater than 0.5, then 0.5 is used.

[0075] To determine the scoring threshold, we statistically analyze the distribution characteristics of the comprehensive scores of all grid cells, calculate the arithmetic mean of the comprehensive scores, then calculate the sum of squares of the differences between the comprehensive scores of each grid cell and the arithmetic mean. We divide the sum of squares by the total number of grid cells and take the square root to obtain the standard deviation. We then add the arithmetic mean and the standard deviation to obtain the comprehensive scoring threshold.

[0076] According to the extraction ratio, a corresponding number of grid cells are extracted from the top of the comprehensive ranking results. The number of extractions is the integer obtained by multiplying the total number of grid cells by the extraction ratio and rounding up. Alternatively, all grid cells with a comprehensive score exceeding the comprehensive score threshold are selected. Grid cells that meet either condition are included in the candidate set of high fouling areas. After the selection is completed, all grid cells in the candidate set of high fouling areas are marked on the spatial distribution map with a highlighted border or an overlaid semi-transparent color block to form the spatial labeling of high fouling areas.

[0077] For the identified candidate set of high-fouling regions, spatial connectivity analysis is used to determine the connectivity between grid cells. The criteria for determining spatial connectivity are as follows: extract the three-dimensional spatial coordinates of any two high-fouling grid cells, including tower height coordinates, radial position coordinates, and azimuth coordinates. Calculate the absolute values ​​of the coordinate differences between the two grid cells in the tower height direction, the radial position direction, and the azimuth direction. If all three absolute values ​​do not exceed 1.5 times the size of a single grid cell in their respective directions, then the two grid cells are determined to have a common boundary or a common vertex, and are thus considered spatially connected. The size of a single grid cell in each direction is determined according to the grid step size set during mesh generation.

[0078] Furthermore, the grid cells in the candidate set are merged into several spatially continuous patches. The merging process is as follows: Select any grid cell from the candidate set of high-fouling regions as the starting cell. Search for all grid cells in the candidate set that are spatially connected to the starting cell. Add these connected cells to the visited list. Then, continue searching for other grid cells spatially connected to each connected cell in the visited list. Add the newly found connected cells to the visited list. Repeat this process until no new cells are added to the visited list. Merge all grid cells in the visited list into the first continuous region. Select a grid cell from the candidate set that has not yet been added to any visited list as a new starting cell. Repeat the above search and merging process to generate the second continuous region. After processing all grid cells in the candidate set, several non-connected continuous regions are obtained. The number of continuous regions is denoted as N.

[0079] Based on this, the number of grid cells contained in each continuous region is counted. The maximum and minimum values ​​of the coordinates of all grid cells in the continuous region along the tower height direction are calculated, and the difference between the two values ​​is taken as the span of the continuous region along the tower height direction. Similarly, the spans in the radial and azimuth directions are calculated, and the maximum value of the spans in the three directions is taken as the maximum span of the continuous region. The arithmetic mean of the tower height coordinates, the arithmetic mean of the radial coordinates, and the arithmetic mean of the azimuth coordinates of all grid cells in the continuous region are calculated. These three averages constitute the centroid coordinates of the continuous region.

[0080] After completing the continuous region division, the area proportion and spatial clustering of high fouling regions are statistically analyzed. The area proportion is calculated by dividing the total number of grid cells in the candidate set of high fouling regions by the total number of grid cells in the effective heat exchange area of ​​the cooling tower, with a value ranging from zero to one. The average number of grid cells in a single continuous region is calculated by dividing the sum of the number of grid cells in all continuous regions by the number of continuous regions N, denoted as M.

[0081] The spatial aggregation degree of scale is calculated based on N and M. The spatial aggregation degree of scale is defined as N divided by M. The smaller the value, the higher the aggregation degree, that is, the high scale area is concentrated in a few large patches. The larger the value, the lower the aggregation degree, that is, the high scale area is dispersed into many small patches.

[0082] S3: Match the spatial aggregation degree of scale with historical cleaning effect records, construct a mapping function between cleaning intensity and scale removal rate, and generate a zoned cleaning scheme that assigns different cleaning intensities to areas with different scale levels.

[0083] Based on the scale spatial aggregation determination results output from step S2, the spatial distribution information of each continuous region, and the comprehensive scale formation level score of each grid cell, a differentiated cleaning intensity allocation scheme is formulated for each continuous region, including:

[0084] Historical cleaning effect records were retrieved from the cooling tower operation and maintenance database. Each record included information such as cleaning time, tower height coordinates, radial position coordinates, azimuth coordinates, pre-cleaning scale thickness measurement, post-cleaning scale thickness measurement, cleaning agent concentration, cleaning duration, and cleaning water flow rate. To obtain historical cleaning cases similar to the spatial characteristics of the current high-scale area, the centroid coordinates of each continuous area output by S2 were used as a benchmark. Records in the historical cleaning effect records where the absolute value of the difference between the cleaning location and the centroid coordinates of any continuous area in the three directions of tower height, radial position, and azimuth did not exceed three cell grid sizes were selected as matching historical cleaning data.

[0085] If the number of selected records is less than five, the coordinate difference tolerance will be gradually increased from three-cell grid size to five-cell grid size until the number of selected records reaches five or more or all historical cleaned records in the database are included in the selection range.

[0086] If a sufficient number of historical cleaning records cannot be obtained after using the above spatial matching method, or if the database itself does not contain historical cleaning effect records, then the recommended cleaning intensity parameter range and theoretical scale removal rate curve provided in the cooling tower design document shall be used as an alternative data source to ensure that the construction of the mapping function has sufficient data support.

[0087] Based on the selected historical cleaning data, the correlation between cleaning intensity parameters and scale removal rate was extracted. For each record, cleaning agent concentration, cleaning duration, and cleaning water flow rate were used as the three components of cleaning intensity. To eliminate dimensional differences, the maximum and minimum values ​​of cleaning agent concentration were calculated for all selected records. The minimum value was subtracted from the minimum value, and then divided by the difference between the maximum and minimum values ​​to obtain the normalized cleaning agent concentration. The same normalization operation was performed on the cleaning duration and cleaning water flow rate to obtain the normalized cleaning duration and normalized cleaning water flow rate. The weighted sum of the three normalized values ​​was used as the comprehensive cleaning intensity parameter, where the sum of the weights of cleaning agent concentration, cleaning duration, and cleaning water flow rate was equal to one. The specific values ​​of each weight are determined as follows: If historical cleaning effect records exist, a mapping function is constructed with the comprehensive cleaning intensity parameter as the independent variable and the scale removal rate as the dependent variable. The least squares method is used for fitting, and the weights are optimized with the minimum root mean square error as the objective function. The weight combination that minimizes the root mean square error is selected as the final weight setting value. If no historical cleaning effect records exist and the parameters recommended in the design document are used as a substitute data source, the three weights are equal, i.e., each is one-third. The scale residue rate after cleaning is calculated by dividing the measured scale thickness after cleaning by the measured scale thickness before cleaning. The scale removal rate is defined as one minus the scale residue rate.

[0088] The normalized cleaning agent concentration, normalized cleaning duration, and normalized cleaning water flow rate of each selected record are weighted and summed according to the determined weights to obtain the comprehensive cleaning intensity parameter for that record. This comprehensive cleaning intensity parameter is then further normalized to the range of zero to one to obtain the dimensionless comprehensive cleaning intensity parameter. The scale removal rate corresponding to this record forms a data pair of independent and dependent variables.

[0089] After obtaining several groups After comparing the data with the scale removal rate, a mapping function between the two is constructed. The formula for the mapping function is expressed as:

[0090]

[0091] in, The scale removal rate predicted by the mapping function; This represents the theoretical upper limit of the scale removal rate when the comprehensive cleaning intensity parameter approaches infinity. The steepness coefficient of the logic function represents the rate of increase in the scale removal rate near the midpoint. The midpoint of the logic function corresponds to the value of the comprehensive cleaning intensity parameter when the scale removal rate reaches half of the maximum removal rate. The coefficient for the logarithmic correction term is used to capture the logarithmic growth characteristic where the rate of scale removal gradually decreases as the intensity increases when the initial cleaning intensity is low. This is the saturation suppression coefficient for the logarithmic correction term, used to ensure that the logarithmic correction term tends to a constant rather than diverges when the cleaning intensity is extremely high; This is a confidence adjustment coefficient for historical data, used to control the degree to which the prediction results depend on the concentrated distribution area of ​​historical cleaned records; This is the arithmetic mean of the comprehensive cleaning intensity parameters in the historical cleaning records; This represents the standard deviation of the comprehensive cleaning intensity parameter in historical cleaning records.

[0092] The above mapping function describes the physical laws of the cleaning process through the product of three factors: the first term is a logistic function term, which depicts the S-shaped growth trend of the removal rate as the cleaning intensity increases; the second term is a logarithmic correction term, which captures the detailed feature of the rapid increase in the removal rate but the gradual slowdown in the low cleaning intensity area, and automatically tends to a constant in the high intensity area through the saturation suppression coefficient; the third term is a historical data confidence adjustment term, which gives a higher confidence level to the predicted value in the area with dense historical data, and automatically reduces the predicted value in the extrapolation area.

[0093] It should be noted that the values ​​of each parameter to be determined in the mapping function are determined in the following way: all historical data pairs or data pairs provided by alternative data sources are used as fitting samples, and the model prediction values ​​are used as the model predictions. The objective function is to minimize the root mean square error between the actual scale removal rate and the target value. A nonlinear least squares method or the Levenberg-Marquardt algorithm is used to analyze the results. , , , , , The six parameters are jointly optimized and fitted to obtain the optimal estimates for each parameter. If... and If the parameters are not treated as free parameters during the fitting process, they are directly calculated from the arithmetic mean and standard deviation of the comprehensive cleaning intensity parameters in the historical cleaning records. After fitting, the root mean square error (RMSE) is calculated. If the RMS error is greater than 0.1, it indicates that the fitting effect is not good, and the quality of the historical data needs to be checked or the amount of data increased before refitting. If the RMS error is less than or equal to 0.1, the fitted parameter values ​​are substituted into the mapping function as the mapping function between cleaning intensity and scale removal rate during formal operation.

[0094] Based on the mapping function and the scale characteristics of each continuous region, a zoned cleaning intensity allocation scheme is generated, specifically:

[0095] For each continuous region identified in step S2, extract the comprehensive fouling level score of all grid cells contained in the continuous region that has been calculated in S2, and calculate the arithmetic mean of these scores as the average fouling level of the continuous region.

[0096] All continuous regions are sorted from highest to lowest average scale level. Based on the ratio of the sorting results to the total number of continuous regions, the continuous regions are divided into three scale levels: continuous regions with an average scale level in the first third of the sort are judged as high scale regions, continuous regions with an average scale level in the middle third of the sort are judged as medium scale regions, and continuous regions with an average scale level in the last third of the sort are judged as low scale regions.

[0097] If there are only one or two consecutive areas, the area with the highest average scale formation level is directly identified as a high scale formation area, and the remaining areas are identified as medium or low scale formation areas in descending order of average scale formation level.

[0098] Different target scale removal rates are set for different scale buildup levels based on the current operating conditions of the cooling tower. Specifically: the target scale removal rate range for high scale buildup areas is 0.90 to 0.98; for medium scale buildup areas, it is 0.80 to 0.90; and for low scale buildup areas, it is 0.65 to 0.80. The specific values ​​for each level are determined based on the heat exchange load requirements of the cooling tower. The higher the load requirement, the upper limit of the corresponding range is used, and the lower the load requirement, the lower limit of the corresponding range is used. When there are no special operating requirements, the default values ​​of 0.95, 0.85, and 0.7 are used for the three levels, respectively.

[0099] Substituting the target scale removal rate of each continuous region into the mapping function, the corresponding comprehensive cleaning intensity parameter is solved by numerical iteration. Numerical iteration can be performed using the bisection method or the Newton-Raphson method, with the absolute value of the difference between the target scale removal rate and the predicted value of the mapping function being less than 0.001 as the convergence condition for iteration. The comprehensive cleaning intensity parameter obtained by the solution is limited to the range of zero to one. If the calculation result is less than zero, it is taken as zero, and if the calculation result is greater than 1, it is taken as 1.

[0100] The obtained comprehensive cleaning intensity parameters are then mapped inversely to specific values ​​for cleaning agent concentration, cleaning duration, and cleaning water flow rate.

[0101] During the reverse mapping, the comprehensive cleaning intensity parameter is distributed to the three dimensions with the same weight ratio as the forward weighted summation to obtain the normalized target value of each dimension.

[0102] The actual cleaning intensity parameter for each dimension is obtained by multiplying the normalized target value of each dimension by the difference between the maximum and minimum values ​​in the historical records of that dimension, and then adding the minimum value in the historical records of that dimension. The actual value of the cleaning agent concentration is expressed as a percentage by mass, the actual value of the cleaning duration is expressed as minutes, and the actual value of the cleaning water flow rate is expressed as cubic meters per hour.

[0103] The cleaning agent concentration, cleaning duration, and cleaning water flow rate corresponding to the high scale accumulation area are recorded as the high cleaning intensity parameter combination; the parameters corresponding to the medium scale accumulation area are recorded as the medium cleaning intensity parameter combination; and the parameters corresponding to the low scale accumulation area are recorded as the low cleaning intensity parameter combination.

[0104] Finally, a detailed record table is formed by summarizing the unique identification number of each continuous area, the average scale formation level, the scale accumulation level determination result, the target scale removal rate, the comprehensive cleaning intensity parameters, the cleaning agent concentration, the cleaning duration, and the cleaning water flow rate to form a zoned cleaning intensity allocation plan.

[0105] S4: Generate the self-cleaning trigger time and cleaning duration for each area according to the zoned cleaning plan, prioritize triggering the self-cleaning operation of areas with high fouling, and convert the cleaning intensity parameters into control commands to be output to the actuator.

[0106] Based on the output of the zonal cleaning intensity allocation scheme, the self-cleaning trigger time and cleaning duration parameters of each zone are generated, and the cleaning intensity parameters in the scheme are converted into control commands that can be directly executed by the cooling tower self-cleaning actuator.

[0107] Before generating the self-cleaning trigger moment, a scale growth rate monitoring system is first established to retrieve scale thickness change data for each continuous area in the most recent operating cycle from the cooling tower operation status database in real time. The time span of the operating cycle is determined according to the cooling tower's operating mode: if the cooling tower adopts a continuous operation mode, the operating cycle is set to 24 hours; if the cooling tower adopts an intermittent operation mode, the operating cycle is set to the complete time interval between two starts.

[0108] For each continuous region, the scale thickness measurements at the centroid of the continuous region at the start and end of the operating cycle are extracted. The scale thickness increment is obtained by subtracting the scale thickness measurement at the start from the end value. The scale thickness increment is then divided by the operating cycle duration to obtain the measured scale growth rate, measured in millimeters per hour. This measured scale growth rate is positively correlated with the cumulative integral of the comprehensive crystallization driving force calculated in step S1. The cumulative integral of the comprehensive crystallization driving force reflects the theoretical accumulation of scale growth potential, while the measured scale growth rate reflects the actual scale growth. The two are cross-validated: if the ratio of the measured scale growth rate to the cumulative integral of the comprehensive crystallization driving force deviates abnormally from the historical statistical range, it indicates that the scale formation kinetic model parameters in step S1 need to be calibrated. This triggers the adaptive update process of the mapping function in step S3, ensuring that the cleaning strategy is driven by both theoretical predictions and measured feedback.

[0109] It should be noted that if no scale thickness sensor is installed at the centroid of the continuous region, the scale thickness measurement values ​​of the grid cells with scale thickness sensors installed in all grid cells included in the continuous region are extracted at the start and end of the operation cycle. The measured scale growth rate of each grid cell is calculated, and the arithmetic mean of the measured scale growth rates of each grid cell is taken as the measured scale growth rate of the continuous region.

[0110] If no scale thickness sensors are deployed in any grid cell within a continuous region, the measured scale growth rate of adjacent grid cells in the continuous region is used. The measured scale growth rate of the continuous region is estimated by inverse distance weighted interpolation, which uses the reciprocal of the Euclidean distance between the centroid coordinates of the continuous region and the centroid coordinates of each adjacent grid cell as the weight to perform a weighted average of the measured scale growth rates of each adjacent grid cell.

[0111] After obtaining the measured scale growth rate for each continuous region, it is determined whether the measured scale growth rate meets the preset cleaning start-up conditions. The cleaning start-up conditions are set separately according to the scale buildup level of each continuous region, and the cleaning start-up threshold corresponding to each scale buildup level is dynamically determined in the following way:

[0112] If the system has accumulated historical operating data for at least one complete scale growth cycle, then the measured scale growth rate data of each scale level region of the cooling tower during at least one historical operating cycle when a detectable decrease in heat exchange efficiency occurs is obtained. The lower quartile of the measured scale growth rate corresponding to each scale level is calculated, and the lower quartile of each scale level is used as the cleaning start threshold for that level region. For example, the threshold for a high scale region in a certain statistical analysis may be 0.05 mm per hour.

[0113] If no historical data exists, a unified initial cleaning start threshold is first set as a transitional solution. The unified threshold is directly determined by the ratio of the maximum allowable scale thickness to the operating cycle provided in the cooling tower design document, without distinguishing between scale levels. After the system has run for at least one complete scale growth cycle and accumulated sufficient scale thickness monitoring data, the differentiated thresholds corresponding to each scale level are determined according to the aforementioned lower quartile method, and the unified initial threshold is replaced.

[0114] When the measured scale growth rate of a continuous area reaches or exceeds the cleaning initiation threshold of the corresponding scale level, the continuous area is marked as a state to be cleaned, and the time when the threshold is reached is recorded as the time when the cleaning requirement is triggered.

[0115] If multiple consecutive areas simultaneously meet the cleaning activation conditions, the areas to be cleaned are sorted in descending order of scale buildup level. Areas with the same scale buildup level are then sorted in descending order of measured scale growth rate. The sorting result is used to determine the self-cleaning trigger time for each area. Priority is given to triggering the self-cleaning operation in areas with high scale buildup. Specifically, the self-cleaning trigger time for the first-ranked consecutive area is set to the current time. The self-cleaning trigger time for the second-ranked consecutive area is set to the self-cleaning trigger time of the first-ranked area plus the cleaning duration parameter of the first-ranked area, plus the cleaning equipment transfer time. The cleaning equipment transfer time is calculated based on the spatial distance between the centroid coordinates of the two consecutive areas and the moving speed of the cleaning equipment. The moving speed of the cleaning equipment is determined based on the technical parameters of the cooling tower's self-cleaning actuator. The self-cleaning trigger times for the third-ranked and subsequent consecutive areas follow the same pattern, ensuring that the self-cleaning operations of each area are executed sequentially according to priority and without time conflicts.

[0116] Furthermore, when generating the cleaning duration parameters for each region, the cleaning duration values ​​corresponding to each continuous region are extracted from the zonal cleaning intensity allocation scheme.

[0117] If the cleaning duration value output by the zonal cleaning intensity allocation scheme has already taken into account the spatial scale of the continuous area, then the cleaning duration value can be directly used as the cleaning duration parameter for the continuous area.

[0118] If the cleaning duration value output by the partition cleaning intensity allocation scheme is the cleaning duration per unit area, then the cleaning duration value is multiplied by the projected area of ​​the continuous region to obtain the total cleaning duration parameter of the continuous region. The projected area of ​​the continuous region is calculated by multiplying the number of grid cells contained in the continuous region by the area of ​​a single grid cell.

[0119] If the spatial distribution of a continuous area is elongated or irregular, making it impossible for the cleaning equipment to cover the entire continuous area in a single cleaning operation, the continuous area is divided into several cleaning sub-areas according to the effective operating radius of the cleaning equipment. A transition overlap zone with a width of 5% to 10% of the effective operating radius is set between adjacent cleaning sub-areas. The cleaning duration of the transition overlap zone is calculated only once and is not included in the additional sum of the independent cleaning durations of each cleaning sub-area. The cleaning duration parameter of each cleaning sub-area is calculated by multiplying the cleaning duration value by the projected area of ​​the cleaning sub-area. The total cleaning duration parameter of the continuous area is the sum of the cleaning duration parameters of each cleaning sub-area plus the sum of the transfer time for the cleaning equipment to move between the cleaning sub-areas.

[0120] It is important to note that the generated cleaning duration parameter must meet the cooling tower operation continuity constraint. If the cleaning duration parameter of a certain continuous area exceeds the maximum time window allowed for a single shutdown of the cooling tower, the cleaning duration parameter will be adjusted to the value corresponding to the maximum time window, and the cleaning operation will be marked as incomplete in the cleaning operation execution record. The remaining cleaning operation will continue to be executed during the next shutdown maintenance of the cooling tower.

[0121] When converting the zoned cleaning intensity parameters into control commands for spray water pressure, chemical dosage concentration, and rinsing water flow rate, the conversion rules are as follows:

[0122] The spray water pressure control command is determined based on the correspondence between the cleaning water flow rate parameters in the zoned cleaning intensity allocation scheme and the nozzle characteristic curve of the cleaning equipment. The nozzle characteristic curve describes the functional relationship between the spray water pressure and the water flow rate through the nozzle. The parameters of the nozzle characteristic curve, including the nozzle flow coefficient and the effective cross-sectional area, are retrieved from the cleaning equipment's technical documents. The cleaning water flow rate parameters output from the zoned cleaning intensity allocation scheme are substituted into the inverse function of the nozzle characteristic curve to calculate the corresponding spray water pressure value, in megapascals (MPa). If the calculated spray water pressure value exceeds the maximum allowable working pressure of the cleaning equipment, the spray water pressure value is adjusted to the value corresponding to the maximum working pressure, and a spray water pressure limitation flag is marked in the cleaning operation execution record. If the calculated spray water pressure value is lower than the minimum working pressure at which the cleaning equipment starts, the spray water pressure value is adjusted to the value corresponding to the minimum working pressure, and a spray water pressure deficiency flag is marked in the cleaning operation execution record.

[0123] The agent dosing concentration control command directly extracts the cleaning agent concentration parameter from the zoning cleaning intensity allocation plan and converts it into a proportioning control signal for the agent storage tank and cleaning water mixing system. The proportioning control signal is calculated based on the cleaning agent concentration parameter and the cleaning water flow rate parameter. Multiplying the cleaning agent concentration parameter by the cleaning water flow rate parameter yields the required cleaning agent mass flow rate per unit time. Dividing the cleaning agent mass flow rate by the agent density at the cleaning agent storage tank outlet pipe yields the cleaning agent volumetric flow rate. This volumetric flow rate is output as the flow control command for the agent dosing pump to the agent dosing system. If the remaining cleaning agent volume in the storage tank is insufficient to support the total cleaning duration of the continuous area, a cleaning agent replenishment process is triggered before the self-cleaning trigger time. This replenishment process includes transferring cleaning agent from the standby storage tank to the main storage tank or replenishing the storage tank from an external supply system. The replenishment time ensures that the cleaning agent reserves are replenished before the self-cleaning trigger time arrives.

[0124] The flushing water flow control command and the spray water pressure control command are coupled, with the flushing water flow rate determined by both the spray water pressure and the nozzle characteristic curve. Simultaneously with generating the spray water pressure control command, the actual flushing water flow rate is calculated based on the command and the nozzle characteristic curve. This actual flow rate is then compared with the cleaning water flow rate parameters output by the zoned cleaning intensity allocation scheme. If the deviation between the actual flushing water flow rate and the cleaning water flow rate parameters exceeds 5%, the deviation is compensated for by adjusting the nozzle opening or increasing / decreasing the number of nozzles involved in the cleaning operation. Nozzle opening adjustment is achieved by controlling the opening angle of the nozzle valve. The nozzle opening and the water flow rate through the nozzle exhibit an approximately linear relationship. The adjustment amount of the nozzle opening is obtained by dividing the deviation by the flow sensitivity of a single nozzle under the current spray water pressure. Flow sensitivity is defined as the change in water flow rate when the nozzle opening changes by one degree. If adjusting the nozzle opening cannot completely eliminate the deviation, calculate the number of nozzles that need to be added or removed. Divide the deviation by the water flow rate of a single nozzle at the current spray pressure and opening to obtain the number of nozzles that need to be adjusted. A positive integer indicates that more nozzles need to be added, while a negative integer indicates that nozzles need to be turned off. The adjusted spray pressure, nozzle opening, and a list of nozzle numbers involved in the cleaning operation together constitute the complete content of the flushing water flow control instruction.

[0125] When outputting the generated control commands to the cooling tower self-cleaning actuator, a hierarchical distributed control architecture is adopted. The control commands are first output to the self-cleaning system's main controller. The main controller generates a cleaning operation timing schedule table based on the self-cleaning trigger time, cleaning duration parameters, and spatial distribution information for each continuous area. The timing schedule table includes information such as the cleaning start time, cleaning end time, cleaning equipment movement path, spray water pressure control commands, chemical dosing concentration control commands, and flushing water flow control commands for each continuous area. The main controller sequentially sends the control commands from each moment in the timing schedule table to the distributed execution units. The distributed execution units include a spray water pressure regulation unit, a chemical dosing unit, a flushing water flow regulation unit, and a cleaning equipment movement unit. After receiving the spray water pressure control commands, the spray water pressure regulation unit achieves precise control of the spray water pressure by adjusting the speed of the cleaning water supply pump or adjusting the opening of the pressure regulating valve in the water supply pipeline. The control accuracy requirement is that the deviation between the actual spray water pressure value and the control command set value does not exceed three percent. After receiving the chemical dosing concentration control command, the chemical dosing unit precisely controls the chemical dosing flow rate by adjusting the speed of the chemical dosing pump or the opening of the flow regulating valve in the chemical dosing pipeline. The control accuracy requires that the deviation between the actual chemical dosing concentration and the set value of the control command does not exceed two percent. After receiving the flushing water flow rate control command, the flushing water flow rate regulating unit drives each nozzle valve to rotate to the specified opening degree and controls the opening or closing of the specified nozzle number through the actuator, according to the nozzle opening parameters and nozzle number list in the control command. The control accuracy requires that the deviation between the actual flushing water flow rate and the set value of the control command does not exceed five percent. After receiving the cleaning equipment movement path command, the cleaning equipment movement unit drives the cleaning equipment support motor or hydraulic actuator to move the cleaning equipment in three-dimensional space in the tower height direction, radial direction, and azimuth direction. The movement positioning accuracy requires that the deviation between the actual position and the target position of the cleaning equipment does not exceed one grid cell size.

[0126] During the execution of control commands, a real-time feedback monitoring mechanism is established. Real-time data is collected from spray water pressure sensors, chemical dosing flow meters, flushing water flow meters, and cleaning equipment position sensors. The real-time data is compared with the control command setpoints to calculate the control deviation. If the control deviation exceeds the control accuracy requirements of the corresponding actuator, a deviation correction process is triggered. The deviation correction process adjusts the control parameters of the actuator according to the magnitude and direction of the deviation. The adjustment range is calculated using a proportional-integral (PI) control law. The proportional gain and integral time constant of the PI control law are calibrated as follows: During the installation and commissioning phase of the actuator, a step control signal is input to the actuator, and the rise time and overshoot of the actuator's output response are recorded. The proportional gain and integral time constant are determined according to the Ziegler-Nichols tuning rule or through simulation optimization to ensure that the actuator's step response meets the performance indicators of minimum rise time and overshoot not exceeding 10%. If the control deviation still exceeds twice the control accuracy requirement after three consecutive deviation corrections, the execution unit is determined to be faulty, triggering a fault alarm and reporting the fault information to the main controller. The main controller adjusts the timing schedule according to the fault type and scope of impact, transferring the cleaning task of the faulty execution unit to the standby execution unit or postponing it until the fault is resolved.

[0127] After the cleaning operation is completed, the thickness of the scale layer in each continuous area after cleaning is measured, the actual scale removal rate is calculated, and the actual scale removal rate is compared with the target scale removal rate set in the zonal cleaning intensity allocation plan. The comparison results, control command execution records, real-time feedback data, and fault information are summarized to form a cleaning operation execution report, which is stored in the cooling tower operation and maintenance database as a historical cleaning effect record. This historical cleaning effect record forms a closed-loop feedback with step S3: the construction of the mapping function and parameter fitting in step S3 depends on the data pairs provided by the historical cleaning effect record. After each cleaning operation, the newly added historical records are automatically included in the fitting sample set for the next mapping function update, so that the prediction accuracy of the mapping function continues to improve with the increase of the number of cleaning operations; at the same time, the determination of the cleaning start threshold in step S4 also depends on historical operating data. With the accumulation of historical records, the thresholds corresponding to each scale level will more accurately reflect the actual scaling characteristics of the cooling tower, realizing adaptive and continuous improvement of the cleaning strategy.

[0128] Those skilled in the art will understand that the specific operations of each step in the above-described intelligent self-cleaning and anti-scaling control method for cooling towers have been referenced above. Figures 1 to 2 The intelligent self-cleaning and anti-scaling control method for cooling towers has been described in detail, and therefore, its repeated description will be omitted.

[0129] According to another aspect of the present invention, a cooling tower intelligent self-cleaning and anti-scaling control system is provided, comprising:

[0130] The data acquisition module is used to acquire time-series data of water temperature, flow rate, conductivity, and pH value at multiple monitoring points in the cooling tower. After spatial gridding, it performs scale formation kinetics inference to obtain scale growth characteristics of each grid cell containing instantaneous and cumulative risk information.

[0131] The fouling inference module is used to sort the peak amplitude and cumulative integral of each grid cell by magnitude, generate a spatial distribution map of fouling growth intensity, identify high fouling areas and count their spatial continuous distribution range and area ratio to obtain the spatial aggregation degree of fouling.

[0132] The scheme generation module is used to match the spatial aggregation degree of scale with historical cleaning effect records, construct a mapping function between cleaning intensity and scale removal rate, and generate a zoned cleaning scheme that assigns different cleaning intensities to areas with different scale levels.

[0133] The execution control module is used to generate the self-cleaning trigger time and cleaning duration for each area according to the zoning cleaning plan, prioritize the self-cleaning operation of areas with high fouling, and convert the cleaning intensity parameters into control commands to be output to the actuator.

[0134] In summary, the intelligent self-cleaning and scale prevention control method for cooling towers based on embodiments of the present invention has been clarified. It acquires time-series data on water temperature, flow rate, conductivity, and pH value from multiple monitoring points within the cooling tower, performs spatial gridding analysis, and infers the scale growth characteristics of each grid unit. Simultaneously, it generates zoned cleaning schemes for high-scale-accumulation areas and optimizes them by combining historical cleaning effect records. This achieves dynamic assessment of scaling risk and precise allocation of cleaning intensity for different areas, resulting in more rational allocation of cleaning resources, improved cleaning efficiency, and effective mitigation of the impact of scale accumulation on cooling tower performance, thus extending equipment lifespan.

Claims

1. A method for intelligent self-cleaning and scale prevention control of cooling towers, characterized in that, include: Time-series data of water temperature, flow rate, conductivity, and pH value from multiple monitoring points within the cooling tower are acquired, and after spatial gridding, scale formation kinetics are inferred to obtain scale growth characteristics of each grid cell containing instantaneous and cumulative risk information. The peak amplitude and cumulative integral of each grid cell are sorted by magnitude to generate a spatial distribution map of scale growth intensity. High scale accumulation areas are identified and their spatial continuous distribution range and area ratio are statistically analyzed to obtain the scale spatial aggregation degree. By matching the spatial aggregation degree of scale with historical cleaning effect records, a mapping function between cleaning intensity and scale removal rate is constructed to generate a zoned cleaning scheme that assigns different cleaning intensities to areas with different scale accumulation levels. Based on the zonal cleaning plan, the self-cleaning trigger time and cleaning duration for each zone are generated. The self-cleaning operation of high-fouling zones is triggered first, and the cleaning intensity parameters are converted into control commands and output to the actuator.

2. The intelligent self-cleaning and scale prevention control method for cooling towers according to claim 1, characterized in that, The scaling kinetics reasoning is based on the principles of solution chemical equilibrium and crystallization kinetics to calculate the supersaturation of scaling ions and the driving force of crystallization. Taking calcium carbonate as an example, the formula for calculating supersaturation is as follows: in, For a moment Location The supersaturation of calcium carbonate, This is the ion activity correction factor. This refers to the calcium ion concentration. This represents the carbonate ion concentration. For temperature The solubility product constant under these conditions For pH values ​​deviating from the sensitivity parameter, This is a damping parameter representing the effect of flow rate on ion activity. This represents the difference between the measured pH value and the precipitation equilibrium pH value. The velocity is the water flow velocity.

3. The intelligent self-cleaning and anti-scaling control method for cooling towers according to claim 2, characterized in that, The formula for calculating the crystallization driving force is expressed as follows: in, As the reference crystallization rate constant, This is the molar mass of calcium carbonate. The nonlinear response exponent of supersaturation to crystallization rate is given. For nucleation activation energy, This is the universal gas constant. Water temperature The critical flow velocity is... The effective diffusion coefficient of ions is . This is the Peckley number correction factor for mass transfer. The characteristic length is denoted as .

4. The intelligent self-cleaning and scale prevention control method for cooling towers according to claim 3, characterized in that, The overall magnitude ranking includes: The crystallization driving force calculated point by point at each time point is connected in chronological order to generate a crystallization driving force time series curve; The peak amplitude is extracted based on the crystallization driving force time-series curve; the cumulative integral is obtained by integrating the crystallization driving force time-series curve over time. The peak amplitude and the cumulative integral are normalized respectively, and then a weighted sum is calculated to form a scale level score; the weights of the weighted sum are determined by historical scale thickness distribution data; The Pearson correlation coefficient was calculated between the scoring results of different weight combinations and the actual scale thickness. The weight combination with the highest correlation coefficient was selected, and all grid cells were sorted from high to low according to the comprehensive scale level score.

5. The intelligent self-cleaning and scale prevention control method for cooling towers according to claim 4, characterized in that, The spatial aggregation degree of the scale layer is defined as the number of continuous regions divided by the average number of grid cells contained in a single continuous region. The continuous region is determined by spatial connectivity analysis, and the criterion is that the absolute value of the coordinate difference between adjacent grid cells in three directions does not exceed 1.5 times the size of a single grid cell.

6. The intelligent self-cleaning and scale prevention control method for cooling towers according to claim 5, characterized in that, The formula for calculating the mapping function is: in, The scale removal rate predicted by the mapping function. To comprehensively measure cleaning intensity parameters, This represents the theoretical upper limit of the scale removal rate when the comprehensive cleaning intensity parameter approaches infinity. For the steepness coefficient of the logic function, The location of the midpoint of the logic function. The coefficient of the logarithmic correction term. The saturation suppression coefficient of the logarithmic correction term. This is the confidence adjustment coefficient for historical data. This is the arithmetic mean of the comprehensive cleaning intensity parameters in the historical cleaning records; This represents the standard deviation of the comprehensive cleaning intensity parameter in historical cleaning records.

7. The intelligent self-cleaning and scale prevention control method for cooling towers according to claim 6, characterized in that, The self-cleaning trigger time is generated by real-time monitoring of the scale growth rate in each continuous area. When the scale growth rate reaches the cleaning start threshold of the corresponding scale level, it is marked as a state to be cleaned. If multiple areas reach the threshold at the same time, they are sorted from high to low scale level. The first area in the sorted order immediately triggers self-cleaning, and the trigger time of subsequent areas is successively superimposed with the cleaning duration of the previous area and the equipment transfer time.

8. A smart self-cleaning and anti-scaling control system for cooling towers, characterized in that, include: The data acquisition module is used to acquire time-series data of water temperature, flow rate, conductivity, and pH value at multiple monitoring points in the cooling tower. After spatial gridding, it performs scale formation kinetics inference to obtain scale growth characteristics of each grid cell containing instantaneous and cumulative risk information. The fouling inference module is used to sort the peak amplitude and cumulative integral of each grid cell by magnitude, generate a spatial distribution map of fouling growth intensity, identify high fouling areas and count their spatial continuous distribution range and area ratio to obtain the spatial aggregation degree of fouling. The scheme generation module is used to match the spatial aggregation degree of scale with historical cleaning effect records, construct a mapping function between cleaning intensity and scale removal rate, and generate a zoned cleaning scheme that assigns different cleaning intensities to areas with different scale levels. The execution control module is used to generate the self-cleaning trigger time and cleaning duration for each area according to the zoning cleaning plan, prioritize the self-cleaning operation of areas with high fouling, and convert the cleaning intensity parameters into control commands to be output to the actuator.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent self-cleaning and anti-scaling control method for cooling towers as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent self-cleaning and anti-scaling control method for cooling towers as described in any one of claims 1 to 7.

Images