Intelligent management and control system and coordination control method for indirect cooling and air cooling island

Abstract

The invention discloses an intelligent management and control system and coordination control method for an indirect cooling air cooling island, and belongs to the technical field of intelligent operation and maintenance of equipment, and the method comprises the steps: collecting key operation parameters in real time through building an intelligent sensing network, and calculating a fin heat exchange efficiency attenuation coefficient and a back pressure early warning value; when the attenuation coefficient of the heat exchange efficiency is lower than a threshold value or the back pressure exceeds a limit, a linkage control mechanism of the intelligent cleaning system and the intelligent spraying system is synchronously started; constructing a control effect evaluation matrix based on the back pressure-attenuation coefficient coupling degree, dynamically adjusting the spraying intensity and the number of cleaning times, and switching to a conventional monitoring mode until the parameters return to normal and are continuously stable; according to the method, cooperative intelligent control of cleaning and spraying is creatively achieved, the heat exchange efficiency of the air cooling island is improved through multi-parameter fusion evaluation and a closed loop feedback mechanism, energy consumption is reduced, meanwhile, the self-adaptive anomaly monitoring and quick response capacity is achieved, and an efficient intelligent operation and maintenance solution is provided for a thermal generator set air cooling system.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of intelligent operation and maintenance of equipment, in particular to an inter-cooled air-cooled island intelligent management and control system and a coordination control method. BACKGROUND

[0002] The traditional air-cooled island control system has significant limitations, mainly manifested in the use of a single parameter threshold triggering mechanism, which cannot fully reflect the real running state of the equipment, and the cleaning and spraying systems are independent of each other, lacking collaborative control capability. The existing technology uses a fixed control strategy, which is difficult to adapt to dynamic changes in operating conditions, especially under extreme weather conditions. At the same time, due to the lack of a closed-loop feedback mechanism, the control effect cannot be quantitatively evaluated, leading to increased energy waste and equipment wear and tear. With the advancement of smart power plant construction, this extensive control mode cannot meet the fine operation and maintenance needs, and a new type of control system that can realize multi-parameter fusion and intelligent collaboration is urgently needed. SUMMARY

[0003] In view of the deficiencies of the prior art, the present application proposes an inter-cooled air-cooled island intelligent management and control system and a coordination control method, which includes: building an intelligent sensing network to collect key operating parameters in real time, calculating the fin heat exchange efficiency decay coefficient and the back pressure early warning value; when the heat exchange efficiency decay coefficient is lower than the threshold value or the back pressure is out of limit, the linkage control mechanism of the intelligent cleaning system and the intelligent spraying system is started simultaneously; based on the back pressure-decay coefficient coupling degree, a control effect evaluation matrix is constructed, and the spraying intensity and cleaning frequency are dynamically adjusted until the parameters return to normal and remain stable, and then the system is switched to the normal monitoring mode; the present application innovatively realizes the collaborative intelligent control of cleaning and spraying, improves the heat exchange efficiency of the air-cooled island through multi-parameter fusion evaluation and closed-loop feedback mechanism, reduces energy consumption, and at the same time has the ability of self-adaptive abnormal monitoring and rapid response, providing an efficient intelligent operation and maintenance solution for the air-cooled system of the thermal power generating unit.

[0004] To achieve the above-mentioned purpose, the present application provides the following technical scheme:

[0005] An inter-cooled air-cooled island coordination control method, comprising:

[0006] S1: build an intelligent sensing network to collect key operating parameters in real time, construct an evaluation model based on the key operating parameters, calculate the fin heat exchange efficiency decay coefficient and the back pressure early warning value; the key operating parameters include fin temperature, back pressure, environmental temperature and pollutant thickness;

[0007] S2: when the fin heat exchange efficiency decay coefficient is lower than the preset value or the back pressure reaches the back pressure early warning value, the linkage control mechanism of the intelligent cleaning system and the intelligent spraying system is started simultaneously, and the initial cleaning and spraying control instruction is output;

[0008] S3: Based on the current back pressure and fin heat exchange efficiency decay coefficient, a control effect evaluation matrix based on the coupling degree of back pressure-decay coefficient is constructed, and the control effect level of cleaning and spraying is evaluated, and the spraying intensity or cleaning frequency is adjusted according to the control effect level, and the adjusted control parameters are output;

[0009] S4: Until the ambient temperature and back pressure reach the preset normal condition and last for a preset time length, a control end signal is output, and the control end signal is based on the control end signal to enter a normal monitoring mode; The normal monitoring mode keeps real-time collection and monitoring of the key operating parameters, and automatically switches back to the intelligent control mode as soon as abnormal changes in the key operating parameters are found, and restarts the intelligent cleaning system or the intelligent spraying system.

[0010] Specifically, the evaluation model includes a parameter-decay coefficient mapping relationship model and a back pressure early warning value calculation model, and the construction process of the evaluation model includes:

[0011] Obtain historical key operating parameters and perform noise filtering and time sequence alignment processing;

[0012] A multi-tree collaborative model is used to obtain a parameter-decay coefficient mapping relationship model, with fin temperature, ambient temperature, and pollutant thickness as input features, and measured fin heat exchange efficiency decay coefficient as labels for training.

[0013] Based on historical back pressure data, a sliding window method is used to calculate the back pressure mean and standard deviation at different ambient temperatures, and a back pressure early warning value calculation model is constructed in combination with the device safety operation threshold.

[0014] Specifically, the multi-tree collaborative model is used to obtain a parameter-decay coefficient mapping relationship model, with fin temperature, ambient temperature, and pollutant thickness as input features, and measured fin heat exchange efficiency decay coefficient as labels for training, including:

[0015] A1: Obtain fin temperature, ambient temperature, pollutant thickness, and measured fin heat exchange efficiency decay coefficient, and combine to form an input parameter set, wherein the fin temperature, ambient temperature, and pollutant thickness are input features, and the measured fin heat exchange efficiency decay coefficient is a label;

[0016] A2: Define three types of nodes of the model, including root nodes, internal nodes and leaf nodes; The root node is used to receive the input parameter set and start the first split; The internal node is used to store the split feature, the split threshold and the linkage relationship of the left and right child nodes, and is used for sample diversion; The leaf node is used to store the sample subset that cannot be further split and the mean value of the label of the subset as the final prediction value output;

[0017] A3: setting a node termination split condition; the node termination split condition is that the number of sample subsets of the current node is less than a preset number, or the variance of the labels in the sample subset of the current node is less than a preset maximum variance, or the tree depth reaches a preset upper limit;

[0018] A4: randomly extracting a sample subset of the same size from the input parameter set as the sample subset of the first tree, and randomly selecting part of the features from the input features as the candidate split features of the first tree; the sample subset of the first tree allows the same parameter to be repeatedly extracted during the extraction process;

[0019] A5: inputting the sample subset of the first tree into the root node, and taking the mean value of all sample labels in the root node as the initial prediction value of the node;

[0020] A6: for the sample subset of the current node, if the node termination split condition is met, the current node becomes a leaf node, and the prediction value is the mean value of all sample labels in the sample subset of the current node;

[0021] A7: for the current node that does not meet the termination condition, traverse the candidate split features of the first tree, sort the values of each candidate split feature in ascending order, and take the midpoint of adjacent values as the candidate threshold value of the candidate split feature;

[0022] A8: for each candidate split feature and the corresponding candidate threshold value, split the sample subset of the current node into a left subset and a right subset, calculate the variance of the labels in the parent node, the variance of the labels in the left subset, and the variance of the labels in the right subset, and then subtract the weighted sum of the variances of the left and right subsets from the variance of the labels in the parent node to obtain the variance reduction amount;

[0023] A9: from all combinations of candidate split features and candidate threshold values, select the combination with the largest variance reduction amount as the split feature and split threshold value of the current node, i.e., the optimal split combination;

[0024] A10: according to the optimal split combination, split the sample subset of the current node into a left subset and a right subset, generate a left child node and a right child node, respectively, and input the left subset and the right subset into the two child nodes, respectively, and repeat A6-A9 for the left child node and the right child node until all child nodes meet the node termination split condition and become leaf nodes, and the first tree model training is completed;

[0025] A11: repeat the steps of A4 to A10 to generate a preset number of tree models, and optimize the structure of each tree model using a post-pruning strategy to obtain an optimized multi-tree model;

[0026] A12: integrate all optimized multi-tree models into a multi-tree collaborative model;

[0027] A13: When it is necessary to predict the fin heat exchange efficiency decay coefficient of a new sample, the input features of the new sample are respectively input into each tree model, each tree outputs a predicted value, that is, the label mean of the leaf node where the sample finally arrives, and the mean of all tree predicted values is taken as the final output of the multi-tree collaborative model, and the obtained parameter-decay coefficient mapping relationship model.

[0028] Specifically, the calculation method of the fin heat exchange efficiency decay coefficient is:

[0029] The average value of the fin heat exchange efficiency in the first 24 hours after the new equipment is first operated or completely cleaned is taken as a reference value.

[0030] The current heat exchange efficiency is calculated by the temperature difference between the fin temperature and the environment temperature and the pollutant thickness collected in real time, combined with the Newton cooling formula.

[0031] The ratio of the current heat exchange efficiency to the reference value is taken as the fin heat exchange efficiency decay coefficient.

[0032] Specifically, the linkage control mechanism of the intelligent cleaning system and the intelligent spraying system includes:

[0033] The intelligent cleaning system is started, a cleaning priority dynamic division method based on pollutant-temperature double parameter coupling is adopted, the priority area is divided and cleaned in sequence combined with the collected pollutant thickness and fin temperature, and the cleaning completion is determined according to the fin temperature field change transmitted back by the sensing network in real time through the cleaning end intelligent judgment system fed back by the built fin temperature field.

[0034] The intelligent spraying system is started synchronously, a hierarchical spraying strategy is implemented according to the fin temperature division area, the spraying water pressure is monitored and adjusted and compensated, and the adjustment and compensation results are fed back to the intelligent sensing network in real time.

[0035] Specifically, the cleaning priority dynamic division method based on pollutant-temperature double parameter coupling includes:

[0036] The surface of the air cooling island fin is divided into standard grid units, each grid unit is equipped with an independent temperature sensor and a laser pollution measuring instrument, the fin temperature and the environment temperature of the grid unit are collected in real time, and the pollutant thickness is calculated.

[0037] The pollutant thickness is divided into N grade intervals, and the fin temperature difference is divided into M grade intervals; the fin temperature difference is the temperature difference between the fin temperature and the environment temperature.

[0038] An N*M two-dimensional priority matrix is constructed; each element in the priority matrix corresponds to the cleaning priority of a combined working condition; the cleaning priority is divided into a first queue, a second queue and a third queue.

[0039] Real-time scanning all grid cells, according to the current collected pollutant thickness and fin temperature difference, the grid cells are classified into a first queue, a second queue or a third queue according to the dynamic scoring rules, and the spatial coordinates of the grid cells in each queue are recorded; the dynamic scoring rules are the corresponding relationship between the pre-defined level combination of pollutant thickness and fin temperature difference and the cleaning priority;

[0040] The improved greedy algorithm is used to generate the shortest cleaning path covering all the first queue grids.

[0041] Specifically, the process of generating the shortest cleaning path covering all the first queue grids by using the improved greedy algorithm includes:

[0042] Extract all grid cells classified as the first queue from the intelligent sensing network, record the physical coordinates of each cell, and read the pollutant thickness and fin temperature difference of each grid cell;

[0043] The pollutant thickness and fin temperature difference of all grid cells are normalized to the same scale, and the cleaning urgency is calculated according to the preset weight; the cleaning urgency is the comprehensive score of the pollutant thickness and fin temperature difference;

[0044] Get the current wet area coordinates being sprayed from the intelligent spraying system, mark it as a forbidden area, and generate a forbidden list;

[0045] Take the current position of the cleaning robot as the starting point of the path, add all the first queue grids to the cleaning list, and remove all grid cells overlapping with the forbidden area;

[0046] For each grid cell remaining in the cleaning list, calculate the weighted sum of the Euclidean distance, cleaning urgency and moving energy consumption to get the comprehensive cost;

[0047] Integrate the three indicators, select the grid cell with the minimum comprehensive cost as the next target cleaning point, and control the robot to move to the target cleaning point to perform cleaning work, and remove the grid cell from the cleaning list after cleaning is completed;

[0048] If the spraying system starts a new wet area during cleaning, update the forbidden list immediately;

[0049] Check the generated path, if three-point detours are found, delete the intermediate target point, if the cleaning list is empty, it means that all the first queue grid cells have been cleaned, and the path planning is completed.

[0050] Specifically, the determination logic of the cleaning endpoint intelligent determination system of the fin temperature field feedback is:

[0051] Real-time monitoring of the change rate and uniformity of the fin temperature field during the cleaning process, when the temperature field change rate is lower than the preset rate threshold and the temperature field distribution uniformity reaches the preset uniformity threshold, the cleaning of the region is determined to be completed, and the determination result is fed back to the intelligent cleaning system to control it to enter the cleaning operation of the next priority area.

[0052] Specifically, the construction process of the control effect evaluation matrix based on the coupling degree of back pressure and attenuation coefficient includes:

[0053] Taking the back pressure change rate and the fin heat exchange efficiency attenuation coefficient change rate as two-dimensional coordinate axes, K evaluation regions are divided, each evaluation region corresponds to a control effect level, and the coupling degree index of back pressure and attenuation coefficient is calculated to determine the level of the current control effect.

[0054] An intelligent control system for an inter-cooling air cooling island, comprising: an intelligent sensing module, an evaluation and decision module, a linkage control module, an effect evaluation module, and a mode switching module.

[0055] The intelligent sensing module is configured to collect key operation parameters of the air cooling island in real time.

[0056] The evaluation and decision module is configured to construct a multi-parameter fusion evaluation model and generate a control strategy.

[0057] The linkage control module is configured to perform collaborative control of intelligent cleaning and spraying.

[0058] The effect evaluation module is configured to quantify the control effect and dynamically optimize the strategy.

[0059] The mode switching module is configured to realize seamless conversion between intelligent control and conventional monitoring.

[0060] Compared with the prior art, the present application has the following advantages:

[0061] 1. The present application proposes an intelligent control system for an inter-cooling air cooling island, and optimizes and improves the architecture, operation steps and flow, the system has the advantages of simple flow, low investment and operation cost, and low production cost.

[0062] 2. The present application proposes an inter-cooling air cooling island coordinated control method, the present application realizes intelligent collaborative control of the cleaning and spraying system by monitoring key operation parameters of the air cooling island in real time through an intelligent sensing network, and dynamically calculating the heat exchange efficiency attenuation coefficient and back pressure warning value based on a multi-parameter fusion evaluation model, when performance degradation is detected, the system automatically triggers a linkage control mechanism, and dynamically optimizes the control strategy through a back pressure-attenuation coefficient coupling evaluation matrix, which significantly improves the accuracy and response speed of heat exchange efficiency maintenance.

[0063] 3.The application provides an intermediate cooling air cooling island coordinated control method, which innovatively constructs a closed-loop regulation and control system from abnormality detection, intelligent control to effect evaluation, can not only quickly restore the device performance to the normal state, but also can realize continuous operation and maintenance through the automatic switching mechanism of the conventional monitoring mode, effectively reduces the energy consumption compared with the traditional method, prolongs the service life of the device, reduces the manual intervention, and provides an efficient, reliable and fully automatic intelligent operation and maintenance solution for the air cooling island system. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 FIG. 1 is a schematic diagram of the intermediate cooling air cooling island coordinated control method of the application;

[0065] Figure 2 FIG. 2 is a principle flowchart of the intermediate cooling air cooling island coordinated control method of the application;

[0066] Figure 3 FIG. 3 is an architecture diagram of the intelligent management and control system of the intermediate cooling air cooling island of the application. DETAILED DESCRIPTION

[0067] Embodiment 1

[0068] Please refer to Figures 1-2 The application provides an embodiment: an intermediate cooling air cooling island coordinated control method, which comprises the following steps:

[0069] S1: an intelligent sensing network is built to collect key operation parameters in real time, an evaluation model is built based on the key operation parameters, and a fin heat exchange efficiency attenuation coefficient and a back pressure early warning value are calculated; the key operation parameters include fin temperature, back pressure, environmental temperature and pollutant thickness;

[0070] Further, the construction of the intelligent sensing network comprises: temperature sensors, pressure sensors, environmental sensors and thickness sensors are arranged in the fin area, the back pressure monitoring point, the environmental monitoring area and the pollutant easy-accumulation area of the intermediate cooling air cooling island respectively, each sensor is connected with a central controller through a wireless communication module, a distributed sensing network is formed, and the data collected in real time is transmitted to the central controller after being preprocessed by an edge computing node to provide data support for the evaluation model.

[0071] Further, the determination method of the back pressure early warning value is: according to the maximum allowable back pressure value in the device design parameter, combining the influence coefficient of the real-time environmental temperature on the back pressure, dynamically adjusting the back pressure early warning value through a back pressure early warning value calculation model, and triggering the linkage control mechanism in S2 when the real-time back pressure reaches the early warning value.

[0072] Further, the influence coefficient of the real-time environmental temperature on the back pressure is gradually determined by combining data modeling and dynamic calibration, comprising:

[0073] First, the principle of the device operation is to determine how the ambient temperature affects the back pressure, that is, the change of the ambient temperature changes the density, viscosity and heat transfer efficiency of the air around the fin, which in turn affects the deposition rate of pollutants, and finally leads to the fluctuation of the resistance of the airflow through the fin. Based on this mechanism, a basic data set is constructed through experiments or historical operation data collection, covering back pressure values corresponding to different ambient temperatures, back pressure change data under the same working conditions, and the corresponding relationship between temperature change rate and back pressure change rate; secondly, the construction stage of the influence coefficient model is entered, the core of which is to quantify the relative change rate of back pressure caused by unit change of ambient temperature: first, define the influence coefficient index, which represents the relative change proportion of back pressure when the ambient temperature changes by 1℃, then fit the temperature and back pressure relationship in the data set through regression analysis, and obtain the basic influence coefficient value corresponding to different temperature intervals, which reflects the differential influence of different temperature ranges on back pressure, such as the positive and negative difference of the coefficients in high and low temperature intervals; finally, the basic influence coefficient is dynamically calibrated and updated in real time, by introducing a correction factor, combining the deviation between the real-time collected ambient temperature, back pressure measured value and model predicted value to adjust the basic coefficient, and obtaining the real-time influence coefficient, to ensure that the real-time influence coefficient can reflect the latest device running state, and avoid error accumulation caused by long-term use of fixed coefficient.

[0074] It needs to be explained that the construction of intelligent sensing network realizes the comprehensive perception and real-time monitoring of the operation state of air cooling island, solves the problem of incomplete data collection and poor timeliness in traditional methods, through the deployment of high-precision sensors and the use of data preprocessing technology, the system can accurately capture the subtle changes in the operation of air cooling island, at the same time, the real-time transmission and storage function ensures the integrity and traceability of data, which provides support for system optimization and fault diagnosis.

[0075] S2: when the fin heat transfer efficiency attenuation coefficient is lower than the preset value or the back pressure reaches the back pressure warning value, the linkage control mechanism of intelligent cleaning system and intelligent spraying system is started synchronously, and the initial cleaning and spraying control instructions are output;

[0076] Further, the preset value in S2 is the critical threshold of fin heat transfer efficiency attenuation coefficient, which is determined by analyzing the energy efficiency curve of the device under different operating conditions. When the fin heat transfer efficiency attenuation coefficient is lower than the preset value, it indicates that the fin heat transfer efficiency has decreased to the extent that affects the normal operation of the device, and the linkage control mechanism in S2 needs to be started.

[0077] S3: based on the current back pressure and fin heat transfer efficiency attenuation coefficient, a control effect evaluation matrix based on the coupling degree of back pressure-attenuation coefficient is constructed and the control effect level of cleaning and spraying is evaluated, and the spraying intensity or cleaning frequency is adjusted according to the control effect level, and the adjusted control parameters are output;

[0078] Further, the specific rule of adjusting the spraying intensity or the cleaning frequency according to the control effect level is: when the control effect level is excellent, the current parameters are kept; when the control effect level is good, the current parameters are kept or fine-tuned and optimized; when the control effect level is medium, the spraying intensity is increased or the cleaning frequency is increased; when the control effect level is poor, the spraying intensity is increased or the cleaning frequency is greatly increased, and the adjusted parameters are used as new control instructions and sent to the corresponding system.

[0079] S4: until the ambient temperature and the back pressure reach the preset normal condition and last for a preset time length, a control end point signal is output, and a normal monitoring mode is entered based on the control end point signal; the normal monitoring mode keeps real-time collection and monitoring of the key operation parameters, and once an abnormal change of the key operation parameters is found, the intelligent control mode is automatically switched back, and the intelligent cleaning system or the intelligent spraying system is restarted.

[0080] Further, the preset normal condition is that the ambient temperature is within the equipment design operation environment temperature range, the back pressure is lower than the back pressure warning value, and the fin heat exchange efficiency attenuation coefficient is greater than or equal to a preset value, the preset time length is set according to the equipment operation stability requirement, and is usually 30-60 minutes, and when the above conditions are met, the control end point signal is output.

[0081] Further, the operation mode of the normal monitoring mode is that the intelligent sensing network keeps real-time collection of the key operation parameters, but the collection frequency is lower than that in the intelligent control mode, the central controller performs normal analysis on the collected data, and when it is found that the key operation parameters are out of the normal fluctuation range, it is determined that there is an abnormal change, and the mode switching mechanism is triggered immediately.

[0082] Further, the determination standard of the abnormal change of the key operation parameters is that the fin temperature mutation amplitude exceeds a preset temperature difference threshold value, the back pressure rising rate exceeds a preset rate threshold value, the ambient temperature exceeds the normal range, or the pollutant thickness growth rate is abnormal, and when any of the above conditions is met, the intelligent control mode is automatically switched back, and the operations of S2 to S4 are re-executed.

[0083] In summary, the present application is directed to the innovative solution designed to address the problems existing in the operation and management of traditional air cooling islands, such as monitoring lag of efficiency decay, extensive cleaning and spraying control, and insufficient energy efficiency optimization. By building a high-precision intelligent sensing network, combining a multi-parameter fusion evaluation model and a dynamic linkage control mechanism, real-time sensing, accurate evaluation and intelligent regulation of the operation state of the air cooling island are realized. The core advantages are: 1) The system can collect key operating parameters such as fin temperature, back pressure, environmental temperature and pollutant thickness in real time, ensuring the comprehensiveness and timeliness of data collection, and providing reliable basis for subsequent evaluation and control; 2) Based on historical data and real-time parameters, a fin heat exchange efficiency decay coefficient and back pressure early warning value calculation model is built to identify potential efficiency decay and safety risks in advance, providing scientific basis for preventive maintenance; 3) Through the linkage control of intelligent cleaning system and intelligent spraying system, combined with the cleaning priority division and hierarchical spraying strategy of pollutant-temperature double parameter coupling, accurate execution of cleaning and spraying is realized, significantly improving control efficiency and energy efficiency level; 4) Through the control effect evaluation matrix based on the coupling degree of back pressure-decay coefficient, the control strategy is dynamically optimized to ensure that the system can maintain the optimal operating state under different working conditions, while supporting seamless switching between intelligent control and conventional monitoring, improving system stability and reliability.

[0084] For example, the evaluation model includes a parameter-decay coefficient mapping relationship model and a back pressure early warning value calculation model, and the construction process of the evaluation model includes:

[0085] Obtain historical key operating parameters and perform noise filtering and time series alignment processing;

[0086] Further, the specific process of noise filtering includes:

[0087] (1) Draw a trend chart of the original historical key operating parameters according to time sequence, and observe the data fluctuation characteristics: if the data presents irregular high-frequency small amplitude oscillation, it is determined as random noise; if there is an isolated large amplitude jump, such as a sudden rise in fin temperature followed by an instant drop, it is determined as impulse noise; if the data as a whole presents a slow, non-physical law rising or falling trend, such as an hourly increase of 0.5℃ in environmental temperature under stable working conditions, which exceeds the natural variation range, it is determined as drift noise;

[0088] (2) Identify the abnormal value corresponding to the impulse noise by adopting three criteria, including: first, calculate the mean and standard deviation of the historical data of any parameter; data points beyond the threshold range of the outliers are marked as outliers; the marked outliers are combined with the physical operation rules of the equipment for secondary verification, such as the back pressure cannot jump from 0.2 MPa to 0.8 MPa in an instant, and can be directly determined as an abnormal value, and after confirmation, the abnormal value is temporarily stored for processing, and the operation data after removing the abnormal value is obtained;

[0089] (3) The operation data after removing the abnormal value is processed by using a sliding window mean filtering, including: setting a window size; in sequence along a time axis, all data points in the window are taken to calculate an arithmetic mean value, and the mean value is used to replace the data point at the center position of the window; the window is slid point by point from the starting position of the data until all data are covered, and random noise is eliminated by smoothing high-frequency fluctuations; the window size is determined according to the parameter characteristics, such as the fin temperature changes rapidly, and the window size is set to 5 sampling points; the back pressure changes slowly, and the window size is set to 10 sampling points;

[0090] (4) A linear regression model is established according to the time sequence of the operation data processed above, and whether there is drift noise is determined according to the slope of the linear regression model, wherein the linear regression model is a prior art content in the field and is not the creative scheme of the present application, and will not be described here;

[0091] (5) If the absolute value of the slope exceeds a preset physical threshold, it is determined that there is drift noise, and the corrected drift-free data is obtained by subtracting the linear model calculation value from the original operation data;

[0092] (6) The outliers in (2) are completed by using a linear interpolation method of adjacent effective data points, and the noise-filtered historical key operation parameters are obtained, wherein the linear interpolation method is a prior art content in the field and is not the creative scheme of the present application, and will not be described here.

[0093] Further, the specific process of the time sequence alignment processing includes:

[0094] (1) Extracting original time stamps from the historical key operation parameters of each parameter and performing unified conversion to ensure consistent time expression, and calculating the time stamp according to the starting time and sampling interval for data without explicit time stamp;

[0095] (2) Counting the sampling frequencies of all historical key operation parameters, selecting the highest sampling frequency in the sampling frequencies of the historical key operation parameters as a reference frequency, and taking the parameter time sequence corresponding to the reference frequency as a reference time axis, and if there are at least two highest frequency parameters, taking the parameter time axis with the largest data amount as the reference to ensure the time range covering all parameters;

[0096] (3) For parameters with sampling frequency lower than the reference frequency, based on the reference time axis and reference frequency in (2), linear interpolation method is used to improve the sampling frequency to the reference frequency, including: on the reference time axis, find out all the reference time points between the adjacent two original data points of the historical key operating parameter, calculate the interpolation value, so that the historical key operating parameter has a corresponding value at each time point on the reference time axis;

[0097] (4) For parameters with sampling frequency equal to the reference frequency, directly match the reference time axis;

[0098] (5) For time offset caused by sensor response delay, the offset is calculated by cross-correlation analysis, including: taking the time series of two parameters, calculating the correlation coefficient under different time offset, and taking the offset when the correlation coefficient is maximum as the correction value, offsetting the time stamp of the lagging parameter by the corresponding time length, realizing time synchronization, wherein the cross-correlation analysis method is the prior art content in the field, and is not the inventive scheme of the present application, which is not described here;

[0099] (6) After alignment, check whether there is missing value on the reference time axis, if there is no parameter value at any time point, use the average of the previous and next 10 valid data points to complete, finally draw the superposition graph of multi-parameter time sequence to intuitively check whether the values of each parameter at the same time point match the physical logic, and ensure the logical consistency of the aligned data.

[0100] A multi-tree collaborative model is used to take fin temperature, environmental temperature and pollutant thickness as input features, and the measured fin heat exchange efficiency decay coefficient as label to train, and obtain a parameter-decay coefficient mapping relationship model.

[0101] Based on the historical back pressure data, the sliding window method is used to calculate the mean and standard deviation of the back pressure under different environmental temperatures, and combined with the safe operation threshold of the equipment, a back pressure early warning value calculation model is constructed.

[0102] Further, the equipment safe operation threshold refers to the maximum allowable limit of back pressure set by the equipment in design and actual operation to ensure its stable, safe and efficient operation, which can also be called the maximum allowable back pressure value or the upper limit of safe back pressure. It is a core parameter determined by the equipment manufacturer according to hardware performance, operation principle, life consumption and other factors, and is a critical standard for judging whether the equipment is in a safe operation state. For example, taking the fin heat exchanger of an air conditioner outdoor unit as an example, the equipment safe operation threshold is set to 500 Pa, that is, when the back pressure exceeds 500 Pa, the fan load is too high, which may burn the motor or cause the heat exchange efficiency to decrease. Based on historical data, the mean value of back pressure under different environmental temperatures is 200-400 Pa, and the standard deviation is 30-50 Pa. Therefore, the early warning value calculation model will set the early warning value dynamically to 350-450 Pa based on the safety bottom line of 500 Pa, the influence coefficient of environmental temperature on back pressure, the mean value and the standard deviation, to ensure that the early warning is triggered before the back pressure approaches the dangerous value. Pa is the unit of pressure, which represents Pascal.

[0103] The multi-tree collaborative model is trained by taking the fin temperature, the environmental temperature and the pollutant thickness as input features and the measured fin heat exchange efficiency decay coefficient as a label to obtain a parameter-decay coefficient mapping relationship model, which includes:

[0104] A1: Obtain the fin temperature, the environmental temperature, the pollutant thickness and the measured fin heat exchange efficiency decay coefficient, and combine them to form an input parameter set, wherein the fin temperature, the environmental temperature and the pollutant thickness are input features, and the measured fin heat exchange efficiency decay coefficient is a label;

[0105] A2: Define three types of nodes of the model, including a root node, an internal node and a leaf node; the root node is used to receive the input parameter set and start the first split; the internal node is used to store the split feature, the split threshold and the linkage relationship of the left and right child nodes for sample diversion; and the leaf node is used to store the sample subset that cannot be further split and the mean value of the labels of the subset as the final prediction value output;

[0106] A3: Set a node termination split condition; the node termination split condition is that the number of sample subsets of the current node is less than a preset number, or the variance of the labels in the sample subset of the current node is less than a preset maximum variance, or the tree depth reaches a preset upper limit;

[0107] A4: Randomly extract a sample subset of the same size from the input parameter set as the sample subset of the first tree, and randomly select part of the features from the input features as the candidate split features of the first tree; the sample subset of the first tree allows the same parameters to be repeatedly extracted during the extraction process;

[0108] Further, a sample subset of the same size is randomly selected from the input parameter set as the sample subset of the first tree, where the same size means that the number of sample subsets selected is consistent with the number of samples in the original input parameter set, and repeated selection is allowed, that is, part of the samples may be included in the subset multiple times as long as the number is the same.

[0109] A5: The sample subset of the first tree is transmitted to the root node, and the mean of all sample labels calculated in the root node is taken as the initial prediction value of the node;

[0110] A6: For the sample subset of the current node, if the node termination splitting condition is met, the current node becomes a leaf node, and the prediction value is the mean of all sample labels in the sample subset of the current node;

[0111] A7: For the current node that does not meet the termination condition, the candidate splitting features of the first tree are traversed, and the midpoint of adjacent values is taken as the candidate threshold value of the candidate splitting feature after sorting the values of each candidate splitting feature in ascending order;

[0112] A8: For each candidate splitting feature and the corresponding candidate threshold value, the sample subset of the current node is split into a left subset and a right subset, and the variance of the labels in the parent node, the variance of the labels in the left subset, and the variance of the labels in the right subset are calculated. Then, the variance reduction amount is obtained by subtracting the weighted sum of the variances of the left and right subsets from the variance of the labels in the parent node;

[0113] A9: From all combinations of candidate splitting features and candidate threshold values, the combination with the largest variance reduction amount is selected as the splitting feature and splitting threshold value of the current node, that is, the optimal splitting combination;

[0114] A10: According to the optimal splitting combination, the sample subset of the current node is split into a left subset and a right subset, and a left child node and a right child node are generated, respectively. The left subset and the right subset are transmitted to the two child nodes, respectively, and A6-A9 are repeated for the left child node and the right child node until all child nodes meet the node termination splitting condition and become leaf nodes, and the first tree model training is completed;

[0115] A11: Repeat steps A4 to A10 to generate a predetermined number of tree models, and use a post-pruning strategy to optimize the structure of each tree model to obtain an optimized multi-tree model;

[0116] Further, the post-pruning strategy is used to optimize the structure of each tree model, including:

[0117] Backtracking from the leaf node to the root node, for each internal node, two kinds of errors are calculated: one is the error after pruning, that is, if the node is converted into a leaf node and the mean value of the labels of all the sample nodes of the node is taken as the predicted value, the error is calculated; the other is the error before pruning, that is, the total error when the node is retained; if the error after pruning is less than or equal to the error before pruning, the child nodes of the node are removed and the node is converted into a leaf node; the process is repeated until all the prunable nodes are processed, and an optimized multi-tree model is obtained.

[0118] A12: all the optimized multi-tree models are integrated into a multi-tree collaborative model;

[0119] A13: when it is necessary to predict the fin heat exchange efficiency attenuation coefficient of a new sample, the input features of the new sample are respectively input into each tree model, each tree outputs a predicted value, that is, the mean value of the labels of the leaf nodes finally reached by the sample, and the mean value of all the predicted values of the trees is taken as the final output of the multi-tree collaborative model, so that the parameter-attenuation coefficient mapping relationship model is obtained.

[0120] It can be seen that the parameter-attenuation coefficient mapping relationship model in the application is a branch structure model capable of autonomously learning the correlation between the fin temperature, the environmental temperature, the pollutant thickness and the heat exchange efficiency attenuation coefficient, the core of which is composed of a unique branch network of three types of nodes, wherein the root node is the starting point of the model and will first receive all the training samples, which contain the above three feature parameters and the corresponding measured attenuation coefficients, in the initial state, the mean value of the attenuation coefficients of all the samples is calculated as the initial predicted value, and after starting the splitting, the root node is converted into a core node for planning the branch direction; the branch node is the decision center of the model, each node carries key attributes: one is the selected splitting feature, the other is the corresponding critical value, and the third is the link path of the left and right two child nodes, wherein the samples with a feature value less than the critical value flow to the left child node, and the samples with a feature value greater than the critical value flow to the right child node, so as to realize accurate sample shunting; finally, the terminal node is the result output end of the model, when the sample subset satisfies the condition of stopping splitting, it converges here, and the core attribute is all the samples and the corresponding average attenuation coefficient of the subset, and finally the average value is directly output as the predicted result.

[0121] Further, in the application, the root node first completes the first splitting through the optimal splitting feature and the critical value, and derives two branch nodes which respectively undertake samples with different feature ranges; each branch node will repeat this process and split again according to the feature distribution of the subset, until the number of samples in the subset is insufficient to support subdivision, the fluctuation range of the attenuation coefficient reaches the preset purity anchor point, that is, the fluctuation is extremely small, or the branch depth reaches the upper limit, at this time, the branch node is converted into a terminal node and stops splitting. This progressive splitting logic can make the model gradually capture the change rule of the attenuation coefficient under different feature combinations and form a hierarchical correlation network.

[0122] Further, for the prediction of new samples, the multi-tree collaborative model will start the path tracing mechanism: starting from the root node, according to the splitting features and critical values of the node, it judges the range of the feature value of the new sample, enters the corresponding branch node; after reaching the branch node, it continues to flow according to its features and critical values, layer by layer, until it reaches the terminal node, and the average attenuation coefficient of this node is the final prediction result. This model is particularly suitable for three features: for fin temperature, it can capture the feature sensitivity of the sudden drop in attenuation coefficient after the temperature breaks through any critical value; for ambient temperature, it can learn the response difference of the attenuation coefficient to the fin temperature under different ambient temperatures; for the thickness of pollutants, it can quantify the decline gradient of the attenuation coefficient with each certain increase in thickness, and finally through the unique branch network, it constructs a precise feature-attenuation coefficient mapping relationship.

[0123] The calculation method of the fin heat exchange efficiency attenuation coefficient is:

[0124] The average value of the fin heat exchange efficiency for 24 consecutive hours after the first operation of the new equipment or after complete cleaning is taken as the reference value;

[0125] The current heat exchange efficiency is calculated by the temperature difference between the fin temperature and the ambient temperature, the thickness of the pollutants, and the Newton cooling formula;

[0126] The ratio of the current heat exchange efficiency to the reference value is taken as the fin heat exchange efficiency attenuation coefficient.

[0127] It needs to be explained that the construction of the evaluation model realizes the accurate evaluation and early warning of the operation state of the air cooling island. The parameter-attenuation coefficient mapping relationship model reveals the complex relationship between fin temperature, ambient temperature, pollutant thickness, and fin heat exchange efficiency through machine learning algorithm, providing a scientific basis for quantitative evaluation of system efficiency attenuation. The back pressure early warning value calculation model sets the back pressure safety threshold under different ambient temperatures based on historical data and statistical methods, providing a strong guarantee for the safe operation of the system. The real-time calculation of the fin heat exchange efficiency attenuation coefficient enables the system to dynamically track the changes in the fin heat exchange efficiency, providing timely feedback for cleaning and spraying control. The synergistic effect of these models improves the accuracy of system evaluation and the timeliness of early warning.

[0128] The linkage control mechanism of the intelligent cleaning system and the intelligent spraying system includes:

[0129] The intelligent cleaning system is started, a cleaning priority dynamic division method based on the coupling of pollutants-temperature double parameters is adopted, the priority area is divided and cleaned in sequence according to the collected pollutant thickness and fin temperature, and the cleaning end intelligent judgment system fed back by the constructed fin temperature field is used to determine whether the cleaning is completed according to the fin temperature field change transmitted back by the perception network in real time.

[0130] The intelligent sprinkler system is started synchronously, a hierarchical sprinkling strategy is implemented according to the temperature of the fins, the water pressure of the sprinkling is monitored and adjusted and compensated, and the adjustment and compensation results are fed back to the intelligent sensing network in real time.

[0131] Further, the adjustment and compensation results refer to the specific operation results and data of the intelligent sprinkler system when the water pressure of the sprinkling deviates from the target value, and the water pressure is restored to the target range through adjustment and control means. These results are fed back to the intelligent sensing network in real time, providing a basis for subsequent control effect evaluation and parameter adjustment, including operation parameters of water pressure adjustment, actual values after water pressure correction, deviation correction conditions, and state identification of the adjustment process.

[0132] Further, the hierarchical sprinkling strategy according to the temperature of the fins includes:

[0133] The difference between the fin temperature and the ambient temperature is divided into Q sprinkling levels, and the sprinkling range is determined according to the area of the fin region. The hierarchical sprinkling is achieved by adjusting the number of open sprinkler heads, water pressure and water spraying time. Real-time data during the sprinkling process are fed back to the intelligent sensing network for control effect evaluation.

[0134] The cleaning priority dynamic division method based on the coupling of the pollutant-temperature double parameters includes:

[0135] The surface of the air-cooled island fin is divided into standard grid units, each grid unit is equipped with an independent temperature sensor and a laser pollution measuring instrument, the fin temperature and the ambient temperature of the grid unit are collected in real time, and the pollutant thickness is calculated;

[0136] The pollutant thickness is divided into N level intervals, and the fin temperature difference is divided into M level intervals; the fin temperature difference is the temperature difference between the fin temperature and the ambient temperature;

[0137] An N×M two-dimensional priority matrix is constructed; each element in the priority matrix corresponds to the cleaning priority of a combined working condition; the cleaning priority is divided into a first queue, a second queue and a third queue;

[0138] All grid units are scanned in real time, and according to the currently collected pollutant thickness and fin temperature difference, the grid units are classified into the first queue, the second queue or the third queue according to the dynamic scoring rule, and the spatial coordinates of the grid units in each queue are recorded; the dynamic scoring rule is the correspondence between the level combination of the pollutant thickness and the fin temperature difference and the cleaning priority, which is predefined;

[0139] An improved greedy algorithm is used to generate the shortest cleaning path covering all the first queue grids.

[0140] The process of generating the shortest cleaning path covering all the first-level queue grids by using the improved greedy algorithm comprises:

[0141] Extract all grid cells classified as first-level queues from the intelligent sensing network, record the physical coordinates of each cell, and read the pollutant thickness and fin temperature difference of each grid cell;

[0142] Normalize the pollutant thickness and fin temperature difference of all grid cells to the same scale, and calculate their cleaning urgency according to the preset weight; the cleaning urgency is the comprehensive score of the pollutant thickness and fin temperature difference;

[0143] Obtain the wet area coordinates currently being sprayed from the intelligent spraying system, mark them as forbidden areas, and generate a forbidden list;

[0144] Take the current position of the cleaning robot as the starting point of the path, add all first-level queue grids to the cleaning list, and remove all grid cells overlapping with the forbidden areas;

[0145] For each grid cell remaining in the cleaning list, calculate the weighted sum of the Euclidean distance, cleaning urgency, and moving energy consumption to obtain the comprehensive cost, wherein the Euclidean distance calculation formula is a prior art content in the field and is not the inventive scheme of the present application, and is not described here.

[0146] Among them, the moving energy consumption refers to the energy cost consumed by the intelligent cleaning device in moving from the current grid cell to any target grid cell in the cleaning list. Specifically, after the intelligent cleaning device completes the cleaning of a grid cell, if it needs to move to the next grid cell to be cleaned, it needs to plan a moving path according to the relative position between the two. The moving energy consumption is calculated based on the actual length of the path, combined with the average power and moving time of the device during movement. For example, the farther apart the two grid cells are, the longer the device needs to move, and the higher the energy consumption. If there are obstacles in the path that need to be bypassed, the actual moving distance increases, and the energy consumption also increases accordingly.

[0147] By integrating the three indicators, the grid cell with the smallest comprehensive cost is selected as the next target cleaning point, and the robot is controlled to move to the target cleaning point to perform cleaning work. After cleaning is completed, the grid cell is removed from the cleaning list.

[0148] If the spraying system opens a new wet area during cleaning, update the forbidden list immediately;

[0149] Check the generated path. If three-point bypassing is found, delete the intermediate target point. If the cleaning list is empty, it means that all first-level queue grid cells have been cleaned, and the path planning is complete.

[0150] The determination logic of the fin temperature field feedback cleaning end point intelligent determination system is:

[0151] The change rate and uniformity of the fin temperature field in the real-time monitoring cleaning process are monitored, when the temperature field change rate is lower than the preset rate threshold and the temperature field distribution uniformity reaches the preset uniformity threshold, it is determined that the cleaning of the region is completed, and the determination result is fed back to the intelligent cleaning system to control it to enter the cleaning operation of the next priority region.

[0152] The construction process of the control effect evaluation matrix based on the coupling degree of back pressure and attenuation coefficient includes:

[0153] Taking the back pressure change rate and the fin heat exchange efficiency attenuation coefficient change rate as two-dimensional coordinate axes, K evaluation regions are divided, each evaluation region corresponds to a control effect level, and the level of the current control effect is determined by calculating the coupling degree index of back pressure and attenuation coefficient.

[0154] The coupling degree index is a quantitative value that comprehensively measures the cooperative effect of back pressure change and attenuation coefficient change.

[0155] Further, the construction process of the two-dimensional coordinate axis includes: taking the back pressure change rate as the horizontal axis and the fin heat exchange efficiency attenuation coefficient change rate as the vertical axis, a two-dimensional evaluation coordinate system is established:

[0156] Horizontal axis direction: left side is the back pressure drop region, right side is the back pressure rise region;

[0157] Vertical axis direction: upper side is the attenuation coefficient rise region, lower side is the attenuation coefficient drop region;

[0158] The origin of the coordinate system corresponds to the initial state that the back pressure change rate is 0 and the attenuation coefficient change rate is 0, i.e. no effect of control measures.

[0159] Further, the division of the evaluation region and the corresponding control effect level includes:

[0160] According to the actual needs and historical data rules of the equipment operation, K evaluation regions are divided in the two-dimensional coordinate system, each region corresponds to a control effect level, such as four levels of excellent, good, medium and poor, i.e. K=4, and the division logic needs to reflect the principle that the more obvious the back pressure drop and the more significant the attenuation coefficient rise, the better the control effect:

[0161] Excellent region: located at the upper left of the coordinate system, i.e. the back pressure change rate is negative and the attenuation coefficient change rate is positive;

[0162] Good region: located at the upper left of the coordinate system but close to the origin, i.e. the back pressure change rate is a small negative value and the attenuation coefficient change rate is a small positive value;

[0163] Middle region: located at the right lower corner of the coordinate system or close to the origin, that is, the back pressure change rate is close to 0 or slightly positive, and the decay coefficient change rate is close to 0;

[0164] Poor region: located at the right lower corner of the coordinate system, that is, the back pressure change rate is positive and the decay coefficient change rate is negative.

[0165] Further, the process of calculating the coupling degree index of back pressure and decay coefficient includes:

[0166] (1) Determine the standardized contribution value of the back pressure change rate and the decay coefficient change rate: take the absolute value of the back pressure change rate, because a negative value indicates improvement, and the larger the absolute value, the greater the contribution; directly take the original value of the decay coefficient change rate, because a positive value indicates improvement, and the larger the value, the greater the contribution;

[0167] (2) Assign different weights to the back pressure change rate and the decay coefficient change rate, such as a back pressure weight of 0.6 and a decay coefficient weight of 0.4, and the total weight is 1. In the present application, the weight needs to be determined based on the influence of the two on the energy efficiency of the equipment in the historical data;

[0168] (3) Calculate the final coupling degree index; the final coupling degree index is equal to the product of the standardized contribution value of the back pressure change rate and the back pressure weight, plus the sum value obtained by multiplying the standardized contribution value of the decay coefficient change rate and the decay coefficient weight.

[0169] It should be noted that in the present application, the design of the intelligent linkage control mechanism realizes the precise execution and efficient cooperation of cleaning and spraying operations; the cleaning priority dynamic division method scientifically divides the priority of the cleaning area through double-parameter coupling analysis, ensures the priority cleaning of high-pollution and high-temperature-difference areas, and improves the pertinence and effectiveness of the cleaning operation; the improved greedy algorithm path planning optimizes the moving path of the cleaning robot, reduces invalid movement and energy consumption, and improves the efficiency of the cleaning operation; the implementation of the grading spraying strategy flexibly adjusts the spraying parameters according to different levels of fin temperature, ensuring the precision and adaptability of the spraying operation; the introduction of the cleaning endpoint intelligent judgment system scientifically judges whether the cleaning is completed by monitoring the change of the fin temperature field in real time, avoiding the problems of over-cleaning or insufficient cleaning, and improving the quality and reliability of the cleaning operation. The synergistic effect of these mechanisms improves the intelligent level of air-cooled island cleaning and spraying operations.

[0170] Embodiment 2

[0171] Please refer to Figure 3 The present application provides another embodiment: an intelligent management and control system for air-cooled island, comprising:

[0172] Intelligent sensing module, evaluation and decision module, linkage control module, effect evaluation module, mode switching module;

[0173] Intelligent sensing module for real-time collection of key operating parameters of air cooling island to provide data support for system decision-making;

[0174] Evaluation decision-making module for building a multi-parameter fusion evaluation model to generate a control strategy;

[0175] Linkage control module for performing coordinated control of intelligent cleaning and spraying;

[0176] Effect evaluation module for quantifying control effect and dynamically optimizing strategy;

[0177] Mode switching module for seamless conversion between intelligent control and conventional monitoring.

[0178] The intelligent sensing module includes a temperature monitoring unit, a back pressure monitoring unit, an environment monitoring unit, and a dirt detection unit.

[0179] The temperature monitoring unit monitors the temperature distribution on the fin surface in real time by deploying an infrared sensor array.

[0180] The back pressure monitoring unit collects inlet and outlet back pressure data of the condenser through a pressure transmitter.

[0181] The environment monitoring unit is used to integrate temperature and humidity sensors and an anemometer to obtain environmental temperature and humidity.

[0182] The dirt detection unit uses a laser thickness gauge to measure the thickness of the accumulated dirt on the fin surface.

[0183] The evaluation decision-making module includes a preprocessing unit, an efficiency calculation unit, a pre-warning analysis unit, and a decision-making generation unit.

[0184] The preprocessing unit is used to filter and normalize the original data.

[0185] The efficiency calculation unit is used to calculate the fin heat exchange efficiency decay coefficient based on a heat transfer model.

[0186] The pre-warning analysis unit is used to dynamically generate a back pressure pre-warning value through a machine learning algorithm.

[0187] The decision-making generation unit is used to trigger cleaning or spraying instructions according to preset rules.

[0188] The linkage control module includes a path planning unit, a spraying control unit, a control unit, and a safety obstacle avoidance unit.

[0189] The path planning unit uses an improved greedy algorithm to generate an optimal cleaning path.

[0190] The spraying control unit controls the start and stop of the spray head and water pressure according to temperature difference zoning.

[0191] A control unit is configured to drive the cleaning robot to operate along the planned path.

[0192] A safety obstacle avoidance unit is configured to detect and avoid the wet area in real time by using a laser radar.

[0193] The effect evaluation module comprises a coupling degree analysis unit, an effect grading unit and a parameter adjustment unit.

[0194] The coupling degree analysis unit is configured to construct a back pressure-attenuation coefficient two-dimensional evaluation matrix.

[0195] The effect grading unit is configured to divide the control effect.

[0196] The parameter adjustment unit is configured to automatically adjust the spraying intensity or the cleaning frequency according to the grade.

[0197] The mode switching module comprises a steady state determination unit, an abnormality monitoring unit and a mode switching unit.

[0198] The steady state determination unit is configured to detect the stable duration of the back pressure and the ambient temperature.

[0199] The abnormality monitoring unit is configured to identify parameter mutation.

[0200] The mode switching unit is configured to automatically trigger the switching between the intelligent control mode and the conventional monitoring mode.

[0201] The embodiments of the present application are described above with reference to the drawings, but the present application is not limited to the specific embodiments described above, and the specific embodiments described above are merely illustrative rather than limiting, and any person of ordinary skill in the art can make changes, modifications, replacements and variations to the above-described embodiments without departing from the purpose of the present application and the scope of protection, and these are all within the protection of the present application.

Claims

1. A coordinated control method for an indirect cooling air cooling island, characterized in that: include: S1: Build an intelligent sensing network to collect key operating parameters in real time, construct an evaluation model based on the key operating parameters, and calculate the fin heat exchange efficiency attenuation coefficient and back pressure warning value; the key operating parameters include fin temperature, back pressure, ambient temperature and contaminant thickness; S2: When the fin heat exchange efficiency attenuation coefficient is lower than a preset value or the back pressure reaches the back pressure warning value, the linkage control mechanism of the intelligent cleaning system and the intelligent spraying system is synchronously started to output initial cleaning and spraying control instructions; S3: Based on the current back pressure and fin heat transfer efficiency attenuation coefficient, a control effect evaluation matrix based on the back pressure-attenuation coefficient coupling degree is constructed to evaluate the control effect levels of cleaning and spraying, and the spray intensity or number of cleaning times is adjusted according to the control effect levels, and the adjusted control parameters are output; S4: Until the ambient temperature and back pressure reach the preset normal conditions and last for a preset time, a control endpoint signal is output, and based on the control endpoint signal, the system enters the conventional monitoring mode; the conventional monitoring mode maintains real-time collection and monitoring of the key operating parameters, and once abnormal changes are found in the key operating parameters, it automatically switches back to the intelligent control mode and restarts the intelligent cleaning system or the intelligent spraying system.

2. The coordinated control method for indirect cooling and air cooling island according to claim 1, characterized in that: The evaluation model includes a parameter-attenuation coefficient mapping relationship model and a backpressure warning value calculation model. The construction process of the evaluation model includes: Obtain historical key operating parameters and perform noise filtering and timing alignment; A multi-tree collaborative model is used to train the fin temperature, ambient temperature, and contaminant thickness as input features and the measured fin heat transfer efficiency attenuation coefficient as a label to obtain a parameter-attenuation coefficient mapping relationship model. Based on historical back pressure data, the sliding window method is used to calculate the mean and standard deviation of back pressure under different ambient temperatures. Combined with the equipment safety operation threshold, a back pressure warning value calculation model is constructed.

3. The coordinated control method for indirect cooling and air cooling island according to claim 2, characterized in that: The multi-tree collaborative model is trained with fin temperature, ambient temperature, and pollutant thickness as input features and the measured fin heat transfer efficiency attenuation coefficient as a label to obtain a parameter-attenuation coefficient mapping relationship model, including: A1: Obtain fin temperature, ambient temperature, contaminant thickness, and the measured fin heat transfer efficiency attenuation coefficient, and combine them to form an input parameter set. Fin temperature, ambient temperature, and contaminant thickness serve as input features, and the measured fin heat transfer efficiency attenuation coefficient serves as a label. A2: Defines three types of nodes in the model, including root nodes, internal nodes, and leaf nodes. The root node is used to receive the input parameter set and initiate the first split. The internal nodes are used to store split features, split thresholds, and the link relationship between left and right child nodes for sample diversion. The leaf nodes are used to store the subset of samples that cannot be split further and the mean of the subset label, which is output as the final prediction value. A3: Set the node termination splitting condition; the node termination splitting condition is that the number of sample subsets of the current node is less than a preset number, or the variance of the labels in the sample subset of the current node is less than a preset maximum variance, or the tree depth reaches a preset upper limit; A4: Randomly extract a sample subset of the same size from the input parameter set as the sample subset of the first tree. At the same time, randomly select some features from the input features as candidate splitting features for the first tree. The sample subset of the first tree can be repeatedly extracted with the same parameters during the extraction process. A5: Pass the sample subset of the first tree to the root node and use the calculated mean of all sample labels in the root node as the initial prediction value of the node; A6: For the sample subset of the current node, if the node termination split condition is met, the current node becomes a leaf node, and its predicted value is the mean of all sample labels in the sample subset of the current node; A7: For the current node that does not meet the termination condition, traverse the candidate split features of the first tree, sort the values ​​of each candidate split feature in ascending order, and take the midpoint of adjacent values ​​as the candidate threshold of the candidate split feature; A8: For each candidate split feature and corresponding candidate threshold, split the sample subset of the current node into a left subset and a right subset. Calculate the variance of the label in the parent node, the variance of the label in the left subset, and the variance of the label in the right subset. Then, subtract the weighted sum of the variances of the labels in the left and right subsets from the variance of the label in the parent node to obtain the variance reduction. A9: From all combinations of candidate split features and candidate thresholds, select the combination with the largest variance reduction as the split feature and split threshold for the current node, i.e., the optimal split combination; A10: Split the sample subset of the current node into a left subset and a right subset based on the optimal split combination, generating a left child node and a right child node respectively. Pass the left and right subsets into the two child nodes respectively. Repeat A6-A9 for the left and right child nodes until all child nodes meet the node termination splitting conditions and become leaf nodes. The first tree model training is completed. A11: Repeat steps A4 to A10 to generate a preset number of tree models, and optimize the structure of each tree model using a post-pruning strategy to obtain optimized multiple tree models; A12: Integrate all optimized multi-tree models into a multi-tree collaborative model; A13: When predicting the fin heat transfer efficiency attenuation coefficient of a new sample, the input features of the new sample are passed to each tree model separately. Each tree outputs a prediction value, which is the label mean of the leaf node where the sample ultimately reaches. The mean of all tree prediction values ​​is then taken as the final output of the multi-tree collaborative model to obtain the parameter-attenuation coefficient mapping relationship model.

4. The coordinated control method for indirect cooling and air cooling island according to claim 3, characterized in that: The calculation method of the fin heat transfer efficiency attenuation coefficient is: The average fin heat exchange efficiency of a new device during the first operation or after complete cleaning for 24 consecutive hours is used as the benchmark value; The current heat transfer efficiency is calculated by combining the real-time collected temperature difference between the fin temperature and the ambient temperature, the thickness of the pollutant, and the Newton cooling formula; The ratio of the current heat transfer efficiency to the reference value is taken as the fin heat transfer efficiency attenuation coefficient.

5. The coordinated control method for indirect cooling and air cooling island according to claim 1, characterized in that: The linkage control mechanism for starting the intelligent cleaning system and the intelligent spraying system includes: The intelligent cleaning system is activated, using a dynamic cleaning priority division method based on the contaminant-temperature dual parameter coupling. The collected contaminant thickness and fin temperature are combined to divide the priority areas and clean them in sequence. Furthermore, an intelligent cleaning endpoint determination system based on fin temperature field feedback is constructed to determine whether the cleaning is complete based on the fin temperature field changes returned in real time by the sensing network. The intelligent spraying system is started simultaneously, and a graded spraying strategy is implemented according to the fin temperature division area. At the same time, the spraying water pressure is monitored and adjusted and compensated, and the adjustment and compensation results are fed back to the intelligent sensing network in real time.

6. The coordinated control method for indirect cooling and air cooling island according to claim 5, characterized in that: The method for dynamically dividing cleaning priorities based on the pollutant-temperature dual parameter coupling includes: The fin surface of the air-cooling island is divided into standard grid cells. Each grid cell is equipped with an independent temperature sensor and laser pollution detector to collect the fin temperature and ambient temperature of the grid cell in real time and calculate the thickness of the pollutants. The pollutant thickness is divided into N level intervals, and the fin temperature difference is divided into M level intervals; the fin temperature difference is the temperature difference between the fin temperature and the ambient temperature; Constructing an N×M two-dimensional priority matrix; each element in the priority matrix corresponds to a cleaning priority of a combined working condition; the cleaning priority is divided into a primary queue, a secondary queue, and a tertiary queue; Scan all grid cells in real time and classify them into primary, secondary, or tertiary queues according to dynamic scoring rules based on the currently collected contaminant thickness and fin temperature difference, and record the spatial coordinates of the grid cells in each queue; the dynamic scoring rules are based on the correspondence between pre-defined combinations of contaminant thickness and fin temperature difference levels and cleaning priorities; An improved greedy algorithm is used to generate the shortest cleaning path covering all the first-level queue grids.

7. The coordinated control method for indirect cooling and air cooling island according to claim 6, characterized in that: The process of using the improved greedy algorithm to generate the shortest cleaning path covering all the first-level queue grids includes: Extract all grid cells classified as the first-level queue from the intelligent perception network, record the physical coordinates of each cell, and read the contaminant thickness and fin temperature difference of each grid cell; Normalize the contaminant thickness and fin temperature difference of all grid cells to the same scale, and calculate their cleaning urgency according to the preset weight; the cleaning urgency is a comprehensive score of the contaminant thickness and fin temperature difference; Obtain the coordinates of the wet area currently being sprayed from the smart sprinkler system, mark it as a prohibited area, and generate a prohibited area list; Taking the cleaning robot's current position as the path starting point, all first-level queue grids are added to the list to be cleaned. At the same time, all grid cells that overlap with the prohibited area are removed. For each grid cell remaining in the cleaning list, the weighted sum of the three indicators, namely, Euclidean distance, cleaning urgency, and movement energy consumption, is calculated to obtain the comprehensive cost. Based on the three indicators, the grid unit with the smallest comprehensive cost is selected as the next target cleaning point, and the robot is controlled to move to the target cleaning point to perform the cleaning operation. After the cleaning is completed, the grid unit is removed from the list of cells to be cleaned. If the spray system opens a new wet area during the cleaning process, the prohibited area list will be updated immediately; Check the generated path. If a three-point detour is found, delete the intermediate target point. If the list to be cleaned is empty, it means that all the first-level queue grid cells have been cleaned and the path planning is completed.

8. The coordinated control method for indirect cooling and air cooling island according to claim 5, characterized in that: The decision logic of the cleaning endpoint intelligent decision system based on the fin temperature field feedback is as follows: The rate of change and uniformity of the fin temperature field during the cleaning process are monitored in real time. When the temperature field change rate is lower than the preset rate threshold and the temperature field distribution uniformity reaches the preset uniformity threshold, the cleaning of the area is determined to be completed. The judgment result is fed back to the intelligent cleaning system to control it to proceed to the cleaning operation of the next priority area.

9. The coordinated control method for indirect cooling and air cooling island according to claim 1, characterized in that: The construction process of the control effect evaluation matrix based on the back pressure-attenuation coefficient coupling degree includes: The back pressure change rate and the fin heat transfer efficiency attenuation coefficient change rate are used as two-dimensional coordinate axes to divide K evaluation areas. Each evaluation area corresponds to a control effect level. The current control effect level is determined by calculating the coupling index of back pressure and attenuation coefficient.

10. An intelligent management and control system for indirect cooling and air cooling islands, which is used to implement the coordinated control method for indirect cooling and air cooling islands according to any one of claims 1 to 9, characterized in that: include: Intelligent perception module, evaluation and decision-making module, linkage control module, effect evaluation module, mode switching module; The intelligent sensing module is used to collect key operating parameters of the air cooling island in real time; The evaluation and decision-making module is used to build a multi-parameter fusion evaluation model and generate a control strategy; The linkage control module is used to perform coordinated control of intelligent cleaning and spraying; The effect evaluation module is used to quantify the control effect and dynamically optimize the strategy; The mode switching module is used to achieve seamless conversion between intelligent control and conventional monitoring.

Images