Circulating water system parameter prediction method based on machine learning

Abstract

The invention relates to the technical field of circulating water, in particular to a circulating water system parameter prediction method based on machine learning, and the method comprises the steps: collecting the characteristic parameters of a circulating water system at different moments; carrying out normal distribution and linear relation analysis on the characteristic parameters by utilizing a consistency check algorithm, and removing the characteristic parameters with negative statistical magnitude; a correlation coefficient method is adopted to calculate correlation coefficients of the characteristic parameters, and the characteristic parameters smaller than a correlation coefficient threshold value are removed to obtain correlation characteristic parameters; and constructing a random forest regression model, training the random forest regression model by using the related characteristic parameters of the training sample and the corresponding water supply flow, and verifying the trained random forest model by using the test sample. According to the method, the operation characteristics of the circulating water system are analyzed, key influence factors are identified, and a prediction model is constructed.

Description

Technical Field

[0001] The present invention relates to the technical field of circulating water, and in particular to a method for predicting circulating water system parameters based on machine learning. Background Art

[0002] With the expansion of industrial production scale and the improvement of environmental protection requirements, the circulating water system, as an indispensable part of industrial cooling and heating processes, its operating efficiency and management level are directly related to the economic benefits and environmental responsibility of the enterprise.

[0003] Traditional circulating water system management and optimization methods mainly rely on experience and static models, lack real-time data-driven capabilities, and are difficult to adapt to the complex and changing needs of modern industry.

[0004] At present, the optimization research and technical application of circulating water systems are mostly focused on direct optimization or system transformation. The former improves local performance by adjusting the operating parameters of a single device or component (such as a pump or valve); the latter physically improves or upgrades existing facilities. These methods have obvious limitations: lack of a global perspective and no in-depth analysis of the relationship between the various components of the circulating water system, resulting in inaccurate optimization measures and difficulty in achieving true full-process optimization; low data utilization, a large amount of historical operating data is not fully utilized, missing the opportunity to extract valuable information from it, limiting the possibility of further optimization;

[0005] In order to overcome the above, there is an urgent need for a new method that can deeply analyze the operating characteristics of the circulating water system, identify the key influencing factors, and build a prediction model. Summary of the invention

[0006] In view of the shortcomings of the existing methods, the present invention uses consistency test and correlation coefficient to extract relevant feature parameters, and uses random forest regression model for prediction.

[0007] The technical solution adopted by the present invention is: a method for predicting circulating water system parameters based on machine learning comprises the following steps:

[0008] Step 1: Collect characteristic parameters of the circulating water system at different times;

[0009] As a preferred embodiment of the present invention, the characteristic parameters include: total water supply temperature, total return water temperature, return water flow, water supply pressure, return water pressure, first cooling tower temperature, first cooling tower liquid level, second cooling tower temperature, second cooling tower cooling tower liquid level, third cooling tower temperature, third cooling tower liquid level, total sewage flow, total sewage pressure, rainwater replenishment flow, rainwater replenishment pressure, low-pressure production water replenishment flow, low-pressure production water replenishment pressure, circulating water pH, circulating water turbidity, circulating water conductivity, circulating water ORP, circulating water residual chlorine, circulating water corrosion rate and circulating water deposition rate.

[0010] Step 2: Use the consistency test algorithm to analyze the normal distribution and linear relationship of the characteristic parameters, and eliminate the characteristic parameters with negative statistics;

[0011] As a preferred implementation of the present invention, the consistency check algorithm includes: Shapiro-Wilk test.

[0012] Step 3: Using the correlation coefficient method, calculate the correlation coefficient of the characteristic parameters, eliminate the characteristic parameters that are less than the correlation coefficient threshold, and obtain the relevant characteristic parameters;

[0013] As a preferred embodiment of the present invention, the correlation coefficient method includes the Spearman correlation analysis method.

[0014] As a preferred embodiment of the present invention, the relevant characteristic parameters include: the liquid level of the first cooling tower, the temperature of the second cooling tower, the rainwater replenishment pressure and the circulating water ORP.

[0015] As a preferred implementation of the present invention, the relevant characteristic parameters are normalized.

[0016] Step 4: Construct a random forest regression model, use the relevant characteristic parameters of the training samples and the corresponding water supply flow to train the random forest regression model, and use the test samples to verify the trained random forest model;

[0017] As a preferred implementation of the present invention, the optimal parameters of the random forest regression model are set as: n_estimators=90, min_samples_split=2, max_depth=19, min_samples_leaf=1.

[0018] As a preferred embodiment of the present invention, explained variance, mean absolute error, mean square error and determination coefficient are used as evaluation indicators of the random forest regression model.

[0019] As a preferred embodiment of the present invention, a circulating water system parameter prediction system based on machine learning includes: a memory for storing instructions executable by a processor; and a processor for executing instructions to implement a circulating water system parameter prediction method based on machine learning.

[0020] As a preferred embodiment of the present invention, a computer-readable medium storing computer program code is characterized in that the computer program code, when executed by a processor, implements the method for predicting circulating water system parameters based on machine learning as described in any one of claims 1-8.

[0021] Beneficial effects of the present invention:

[0022] 1. Through in-depth analysis of the characteristic parameter data of the circulating water system, the Shapiro-Wilk test and Spearman correlation analysis method are used to dig out the relevant characteristic parameters related to the water supply flow, and finally the random forest regression model is used for training to predict the water supply flow value. The system model is simple and the prediction accuracy is high;

[0023] 2. The prediction model of the present invention can effectively allocate the water supply of each water plant, obtain the water supply pressure through the water supply flow, thereby guiding the operation of the circulating water system and reducing the operating cost;

[0024] 3. Provide forecast water supply pressure to provide decision support for production scheduling in advance. By implementing comprehensive scheduling management of the outlet pump stations of each water plant in the pipeline network, the balanced distribution of pipeline network pressure can be achieved, and ultimately the energy-saving and consumption-reduction goals of the entire production system can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 It is a flow chart of the circulating water system parameter prediction method based on machine learning of the present invention;

[0026] Figure 2 is a scatter plot of water supply flow and characteristic parameters of the present invention;

[0027] Figure 3 Schematic diagram of the MSE score of the random forest regression model of the present invention;

[0028] Figure 4 It is a schematic diagram of the structure of the existing circulating water system;

[0029] Figure 5 It is a comparison chart of the predicted value and the actual value of the present invention. DETAILED DESCRIPTION

[0030] The present invention is further described below in conjunction with the accompanying drawings and embodiments. This figure is a simplified schematic diagram, which only illustrates the basic structure of the present invention in a schematic manner, and therefore it only shows the components related to the present invention.

[0031] like Figure 1 As shown, a circulating water system parameter prediction method based on machine learning includes the following steps:

[0032] Step 1: Collect characteristic parameters of the circulating water system at different times;

[0033] Use the DCS system to collect characteristic parameters of the circulating water system in petrochemical plants at different time periods and under different environmental conditions;

[0034] The characteristic parameters include: total supply water temperature, total return water temperature, return water flow, supply water pressure, return water pressure, first cooling tower temperature, first cooling tower liquid level, second cooling tower temperature, second cooling tower cooling tower liquid level, third cooling tower temperature, third cooling tower liquid level, total sewage flow, total sewage pressure, rainwater replenishment flow, rainwater replenishment pressure, low-pressure production water replenishment flow, low-pressure production water replenishment pressure, circulating water pH, circulating water turbidity, circulating water conductivity, circulating water ORP, circulating water residual chlorine, circulating water corrosion rate and circulating water deposition rate, as well as the corresponding scatter plot of water supply flow; the scatter plot is as follows: Figure 2 As shown;

[0035] The model of the first cooling tower is C2001A, the model of the second cooling tower is C2001B, and the model of the third cooling tower is C2001C;

[0036] like Figure 4 This is a schematic diagram of the existing circulating water system structure: the DCS system installed in the circulating water plant collects the total supply water temperature, total return water temperature, supply water flow, return water flow, supply water pressure, return water pressure, total sewage flow, total sewage pressure, rainwater makeup flow, rainwater makeup pressure, low-pressure production water makeup flow, low-pressure production water makeup pressure, circulating water pH, circulating water turbidity, circulating water conductivity, circulating water ORP, circulating water residual chlorine, circulating water corrosion rate and circulating water deposition rate.

[0037] Step 2: Use the consistency test algorithm to analyze the normal distribution and linear relationship of the characteristic parameters, and eliminate the characteristic parameters with negative statistics;

[0038] The consistency test algorithm uses the Shapiro-Wilk test, and the formula is:

[0039]

[0040] Where n represents the total number of data samples; a i is the coefficient related to the expected value and standard deviation of the normal distribution, which can be found in the Shapiro-Wilk test table and is related to the sorting order of sample data and the concept of regression analysis; i Collect sample values ​​for the i-th data, is the average value of the data collection samples.

[0041] The normal distribution test is shown in Table 1

[0042] Table 1: Shapiro-Wilk normality test

[0043]

[0044] It can be seen from Table 1 that all characteristic parameter statistics are greater than 0.

[0045] Step 3: Using the correlation coefficient method, calculate the correlation coefficient of the characteristic parameters, eliminate the characteristic parameters that are less than the correlation coefficient threshold, and obtain the relevant characteristic parameters;

[0046] The correlation coefficient method is the Spearman correlation analysis method, and the formula is:

[0047]

[0048] Where n represents the total number of data samples, ∑D 2 represents the sum of squares of all rank differences.

[0049] The correlation coefficient threshold was set to 0.5;

[0050] Table 2 shows the relevant characteristic parameters obtained by Spearman correlation analysis method

[0051] Table 2 Related characteristic parameters and corresponding correlation coefficient values

[0052]

[0053] The relevant feature parameters are normalized, and the normalization formula is:

[0054]

[0055] Among them, x i Indicates that a target data with a large influence weight is obtained through correlation analysis, x min Indicates the minimum data in the target data collection list, x max Indicates the maximum data in the target data collection list, x i 'Updated value after normalization of this data.

[0056] The data set of relevant feature parameters is divided into 70% training sample data and 30% testing sample data.

[0057] Step 4: Construct a random forest regression model, use the relevant characteristic parameters of the training samples and the corresponding water supply flow to train the random forest regression model, and use the test samples to verify the trained random forest model;

[0058] The optimal parameters of the random forest regression model are shown in Table 3;

[0059] Table 3 Optimal parameters of random forest model

[0060]

[0061] The explained variance (EV), mean absolute error (MAE), mean square error (MSE), and coefficient of determination (R 2 ) is used as the evaluation index to evaluate the random forest regression model;

[0062] Through the hyperparameter optimization of the random forest regression model, the water supply flow can be accurately predicted; it can be applied to the subsequent circulating water system of the petrochemical plant, such as Figure 3 As shown, the MSE scores of the random forest regression model for different parameter combinations during the grid search process.

[0063] Figure 5 The curve graph of the predicted value and the actual value shows that the predicted value is basically consistent with the actual value, indicating that the model of the present invention is effective.

[0064] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A method for predicting circulating water system parameters based on machine learning, characterized in that: The following steps are involved: Step 1: Collect characteristic parameters of the circulating water system at different times; Step 2: Use the consistency test algorithm to analyze the normal distribution and linear relationship of the characteristic parameters, and eliminate the characteristic parameters with negative statistics; Step 3: Using the correlation coefficient method, calculate the correlation coefficient of the characteristic parameters, eliminate the characteristic parameters that are less than the correlation coefficient threshold, and obtain the relevant characteristic parameters; Step 4: Construct a random forest regression model, use the relevant feature parameters of the training samples and the corresponding water supply flow to train the random forest regression model, and use the test samples to verify the trained random forest model.

2. The method for predicting circulating water system parameters based on machine learning according to claim 1, characterized in that: Consistency test algorithms include: Shapiro-Wilk test.

3. The method for predicting circulating water system parameters based on machine learning according to claim 1, characterized in that: The correlation coefficient method includes the Spearman correlation analysis method.

4. The method for predicting circulating water system parameters based on machine learning according to claim 1, characterized in that: The characteristic parameters include: total supply water temperature, total return water temperature, return water flow, supply water pressure, return water pressure, first cooling tower temperature, first cooling tower liquid level, second cooling tower temperature, second cooling tower liquid level, third cooling tower temperature, third cooling tower liquid level, total sewage flow, total sewage pressure, rainwater make-up flow, rainwater make-up pressure, low-pressure production water make-up flow, low-pressure production water make-up pressure, circulating water pH, circulating water turbidity, circulating water conductivity, circulating water ORP, circulating water residual chlorine, circulating water corrosion rate and circulating water deposition rate.

5. The method for predicting circulating water system parameters based on machine learning according to claim 1, characterized in that: The relevant characteristic parameters include: the first cooling tower liquid level, the second cooling tower temperature, the rainwater replenishment pressure and the circulating water ORP.

6. The method for predicting circulating water system parameters based on machine learning according to claim 5, characterized in that: Normalize the relevant feature parameters.

7. The method for predicting circulating water system parameters based on machine learning according to claim 1, characterized in that: The optimal parameter settings for the random forest regression model are: n_estimators=90, min_samples_split=2, max_depth=19, min_samples_leaf=1.

8. The method for predicting circulating water system parameters based on machine learning according to claim 7, characterized in that: Explained variance, mean absolute error, mean square error and coefficient of determination were used as evaluation indicators of the random forest regression model.

9. A circulating water system parameter prediction system based on machine learning, characterized in that: include: a memory for storing instructions executable by a processor; A processor, used to execute instructions to implement the method for predicting circulating water system parameters based on machine learning as described in any one of claims 1-8.

10. A computer readable medium storing computer program code, characterized in that: When the computer program code is executed by a processor, the method for predicting circulating water system parameters based on machine learning as described in any one of claims 1 to 8 is implemented.

Images