An operation optimization method for a compressed air and steam combined supply system

Abstract

The present invention relates to an operation optimization method for a compressed air and steam joint supply system, comprising: obtaining operation history data of the compressed air and steam joint supply system based on influencing factors; processing and screening each parameter data in the historical data; establishing an LSTM prediction model for the screened parameter data to obtain prediction data; based on the historical data and the prediction data, with the goal of minimizing the total cost of electricity and steam operation, using a PSO algorithm to optimize the operation plan and obtain the optimal operation plan; the system operates based on the optimal operation plan. The beneficial effects of the present invention are: the present invention performs LSTM prediction on the demand for compressed air and steam, considers the operating characteristics of steam-driven centrifugal air compressors and electric-driven centrifugal air compressors, with the goal of minimizing the operating cost of steam and electricity, and with the constraint that compressed air and steam meet user needs, to obtain an optimized operation strategy, which can reduce resource waste.

Description

Technical Field

[0001] The invention relates to the field of integrated energy technology, and more specifically, to an operation optimization method for a compressed air and steam combined supply system. Background Art

[0002] At present, the construction of centralized industrial parks has become an important model for scientific development and energy conservation in my country, and steam supply is an important infrastructure to ensure the industrial heat use in the park. In order to actively respond to the development of the global energy and environmental protection situation, we must take multiple measures to reduce energy waste and improve resource utilization. my country's existing steam supply network mainly uses high-parameter steam. Users need to adjust steam parameters and reduce steam quality through temperature and pressure reduction devices according to their own needs. The losses caused by steam conversion cannot be underestimated. At the same time, steam users in the park themselves or other users have compressed air needs. Generally, they build small screw air compressors by themselves, which have low compressed air conversion efficiency, high power consumption, and difficult oily wastewater discharge and treatment. The maintenance cost of the air compressor is high. During the operation of the equipment, it cannot be reasonably adjusted according to changes in user needs in a timely manner, and the secondary conversion of steam is increased, resulting in waste of resources. Summary of the invention

[0003] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method for optimizing the operation of a compressed air and steam combined supply system. The technical solution is as follows:

[0004] In a first aspect, a method for optimizing the operation of a compressed air and steam combined supply system is provided, comprising the following steps:

[0005] Step 1: Based on the influencing factors, obtain the historical operation data of the compressed air and steam joint supply system, wherein the historical data includes a plurality of parameter data;

[0006] Step 2: Normalize each parameter data in the historical data and calculate the Pearson correlation coefficient ρ X,Y Calculate and filter out parameter data that meets the screening conditions, wherein the parameter data that meets the screening conditions has a greater correlation with the optimization target;

[0007] Step 3: Establish a long short-term memory neural network (LSTM) prediction model with time series prediction characteristics for the parameter data with greater correlation screened out in step 2, and obtain the forecast data of the demand for compressed air and steam after 5 minutes, 1 hour and 24 hours;

[0008] Step 4: Based on the historical data and the forecast data in step 3, with the goal of minimizing the total cost of electricity and steam operation, the PSO algorithm is used to optimize the operation plan to obtain the optimal operation plan;

[0009] Step 5: The compressed air and steam combined supply system operates based on the optimal operation plan in step 4.

[0010] Preferably, in step 2, the normalized calculation formula is:

[0011]

[0012] Among them, X max is the maximum value of the sample data, X min is the minimum value of the sample data, X is the sample data, X 归一 is the normalized data.

[0013] As a preference, in step 2, the Pearson correlation coefficient ρ X,Y The calculation formula is:

[0014]

[0015] Among them, cov is the covariance, σ is the standard deviation, E is the expectation, and X and Y are parameters.

[0016] Preferably, step 3 specifically comprises the following steps:

[0017] Step 3.1, establish the initial prediction model;

[0018] Step 3.2, dividing the parameter data with greater correlation screened out in step 2 into a training data set, a validation data set, and a test data set;

[0019] Step 3.3, train the initial prediction model using the training data set to obtain a training prediction model;

[0020] Step 3.4: Use the validation data set to train the prediction model and adjust the parameters of the model;

[0021] Step 3.5, repeat steps 3.3 and 3.4 until the training prediction model reaches the set training termination threshold, and determine the best prediction model;

[0022] Step 3.6: Based on the best prediction model in step 3.5, obtain the forecast data of the demand for compressed air and steam after 5 minutes, 1 hour and 24 hours.

[0023] Preferably, the influencing factors include: enterprise benefits, enterprise attributes, market conditions, and climate factors.

[0024] Preferably, in step 2, the screening condition is X,Y ≥0.6.

[0025] In a second aspect, the present invention further provides a compressed air and steam combined supply system, comprising: a steam turbine, a steam-driven centrifugal air compressor, an electrically-driven centrifugal air compressor, and operated by any of the operation optimization methods described in the first aspect.

[0026] The beneficial effects of the present invention are as follows: the present invention predicts the quality (pressure, temperature) and pressure demand of downstream compressed air and steam by using a long short-term memory neural network (LSTM), obtains the estimated value at the future moment, considers the operating characteristics of steam-driven centrifugal air compressors and electric-driven centrifugal air compressors, takes minimizing the steam and electric energy operating costs as the goal, takes compressed air and steam meeting user needs as the constraint, adopts a particle swarm optimization algorithm (PSO) to adjust the steam intake volume and air compressor operation, obtains an optimized operating strategy, provides guidance for the optimized operation of the compressed air and steam joint supply system, can reduce resource waste, and solve the problem of untimely system adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 Schematic diagram of the structure of the compressed air and steam combined supply system provided for this application;

[0028] Figure 2 A flow chart of the operation optimization method of the compressed air and steam combined supply system provided for this application;

[0029] Figure 3 A flow chart of the steam load prediction method provided for this application;

[0030] Figure 4 A flowchart for obtaining the optimal operating solution provided for this application. DETAILED DESCRIPTION

[0031] The present invention is further described below in conjunction with embodiments. The description of the following embodiments is only used to help understand the present invention. It should be noted that for ordinary persons in the art, without departing from the principle of the present invention, the present invention can also be modified in some ways, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

[0032] Embodiment 1:

[0033] The present application proposes a compressed air and steam combined supply system, comprising: a steam turbine, a steam-driven centrifugal air compressor, and an electrically-driven centrifugal air compressor.

[0034] like Figure 1As shown, the high-pressure steam supplied by the centralized steam supply pipeline is adjusted to the working conditions suitable for the steam turbine. The adjusted high-pressure steam enters the steam turbine, drives the turbine impeller to rotate, and the kinetic energy of the steam is converted into mechanical energy. Through further adjustment of the steam turbine, the steam parameters are adjusted to meet the needs of the user and then supplied to the heat user. The centrifugal air compressor is directly driven by the steam-driven radial turbine. A part of the low-pressure steam is supplied to the downstream steam users, and the other part is generated by the centrifugal air compressor to generate compressed air for the downstream compressed air users. Since the steam-driven centrifugal air compressor is determined by the steam consumption of the downstream user's process point and cannot be adjusted by itself, this system uses a steam-driven centrifugal air compressor in parallel with an electric-driven centrifugal air compressor. When the pressure of the compressed air in the pipeline network exceeds the set pressure of the pipeline network, the electric-driven centrifugal air compressor is adjusted to reduce the gas production, thereby achieving the purpose of energy saving. At the same time, in order to ensure that the power of the centrifugal air compressor and the steam turbine fluctuate synchronously, this unit adopts constant speed regulation.

[0035] The present application can reduce the energy loss caused by the secondary conversion of steam, can more efficiently recover steam pressure energy, and adopts both steam drive and electric drive to improve the safety and economy of the system. In addition, the system structure of the present application is simple, occupies a small area, and is suitable for steam pressure energy recovery for end users.

[0036] Embodiment 2:

[0037] This application proposes an operation optimization method for a compressed air and steam joint supply system, which uses a long short-term memory neural network (LSTM) to predict the demand for compressed air and steam and combines the PSO algorithm to optimize the operation plan. Based on the parallel operation mode of steam-driven centrifugal air compressors and electric-driven centrifugal air compressors, the optimal operation plan is recommended to solve the problem of untimely system adjustment, such as Figure 2 As shown, the following steps are included:

[0038] Step 1: Based on influencing factors, historical operation data of the compressed air and steam combined supply system is obtained, wherein the historical data includes multiple parameter data.

[0039] In step 1, the influencing factors may include: enterprise benefits, enterprise attributes, market conditions, and climate factors. Optionally, the influencing factors may also include: steam pressure, steam flow, steam temperature, steam usage, and compressed air usage. It should be noted that the influencing factors determine the parameters selected for historical data and the parameter data included in the historical data. For example, when the influencing factors are steam pressure and steam flow, the parameters selected for historical data are steam pressure and steam flow. At this time, the historical data includes steam pressure data and steam flow data.

[0040] Step 2: normalize each parameter data in the historical data, calculate the Pearson correlation coefficient ρ_(X,Y), and filter out parameter data that meet the screening conditions. The parameter data that meet the screening conditions has a greater correlation with the optimization target.

[0041] Under actual operating conditions, the system is in a variable operating state. During the variable operating conditions, the system's operating parameters change asynchronously and cannot reflect the static characteristics of the corresponding equipment. Therefore, when collecting historical data on the demand for low-pressure steam and compressed air on the user side, it is necessary to preprocess the modeling data and extract the data of the system under stable conditions from the historical data for correlation analysis. In order to improve the model prediction accuracy and the convergence speed of the prediction process, the data is normalized.

[0042] In step 2, the normalization calculation formula is:

[0043]

[0044] Among them, X max is the maximum value of the sample data, X min is the minimum value of the sample data, X is the sample data, X 归一 is the normalized data.

[0045] Since the fluctuation values ​​of the monitoring data are similar, it is difficult to determine the degree of correlation between each group of data, so the Pearson correlation coefficient is introduced to conduct Pearson correlation analysis.

[0046] Pearson correlation coefficient ρ X,Y The calculation formula is:

[0047]

[0048] Among them, cov is the covariance, σ is the standard deviation, E is the expectation, and X and Y are parameters.

[0049] In addition, the screening condition in step 2 can be X,Y ≥0.6.

[0050] Step 3: A long short-term memory neural network (LSTM) prediction model with time series prediction characteristics is established for the parameter data with greater correlation screened out in step 2 to obtain the forecast data of the demand for compressed air and steam after 5 minutes, 1 hour and 24 hours.

[0051] In this research field, wavelet transform, variational mode decomposition and empirical mode decomposition are commonly used methods for time series signal analysis and processing. However, wavelet transform and variational mode decomposition are easily affected by initialization parameters; when decomposing the original data, empirical mode decomposition decomposes multiple IMF modal components and a residual component, which may lead to modal stacking problems, thereby affecting the results of the prediction analysis. In order to solve the problems existing in traditional algorithms, and combining the volatility and time series characteristics of steam, the present invention uses a CEEMDAN-LSTM steam load prediction method.

[0052] The CEEMDAN algorithm can optimize the mode stacking problem caused by the traditional EMD algorithm and produce better decomposition results. Figure 3 As shown in the figure, first, the historical data of compressed air and steam demand are imported into the CEEMDAN algorithm to decompose the complex historical data of compressed air and steam demand into multiple IMF modal components and a residual component. The IMF modal components obtained after decomposition are modeled and analyzed, and the prediction results are obtained by further analysis and prediction through the LSTM neural network. At the same time, in order to improve the learning efficiency of the model and the generalization ability of the prediction, layer standardization is used to improve the convergence speed and further optimize the network model.

[0053] In addition, if Figure 4 As shown, step 3 specifically includes the following steps:

[0054] Step 3.1, establish the initial prediction model;

[0055] Step 3.2, dividing the parameter data with greater correlation screened out in step 2 into a training data set, a validation data set, and a test data set;

[0056] Step 3.3, train the initial prediction model using the training data set to obtain a training prediction model;

[0057] Step 3.4: Use the validation data set to train the prediction model and adjust the parameters of the model;

[0058] Step 3.5, repeat steps 3.3 and 3.4 until the training prediction model reaches the set training termination threshold, and determine the best prediction model;

[0059] Step 3.6: Based on the best prediction model in step 3.5, obtain the forecast data of the demand for compressed air and steam after 5 minutes, 1 hour and 24 hours.

[0060] Step 4: Based on the historical data and the predicted data in step 3, with the goal of minimizing the total cost of electricity and steam operation, the PSO algorithm is used to optimize the operation plan to obtain the optimal operation plan.

[0061] The steps of the PSO algorithm are as follows:

[0062] Step 1: Read the load side operating data, including pressure, temperature, flow, etc.

[0063] Step 2: Encode the steam pressure, steam flow, and steam temperature.

[0064] Step 3: Set the population size popsize, the maximum number of iterations Iteration, the learning factors c1, c2 of the PSO algorithm, and randomly generate the initialization population according to the value range of the optimization variable.

[0065] Step 4: According to the fitness function, calculate the fitness of individual particles, and calculate the position and optimal value of the local optimal particle and the global optimal particle.

[0066] Step5: Update particle speed and position.

[0067] Step 6: Calculate fitness and update position and speed.

[0068] Step7: If gen>Iteration, save the optimal solution; otherwise gen=gen+1, go to Step4.

[0069] Step 5: The compressed air and steam combined supply system operates based on the optimal operation plan in step 4.

[0070] In summary, this application establishes an operation optimization model, combines the predicted values ​​of compressed air and steam demand, takes the lowest total cost of electricity and steam operation as the goal, optimizes the adjustment means, and obtains the operation strategy. This application can reduce costs, reduce resource waste, and solve the problem of untimely system adjustment.

Claims

1. A method for optimizing the operation of a compressed air and steam combined supply system, characterized in that: The method comprises: Step 1: Based on influencing factors, historical operation data of the compressed air and steam joint supply system is obtained, wherein the historical data includes multiple parameter data; the influencing factors include: enterprise benefits, enterprise attributes, market conditions, and climate factors; Step 2: Normalize each parameter data in the historical data and calculate the Pearson correlation coefficient ρ X,Y Calculate and filter out parameter data that meet the screening conditions, and the parameter data that meet the screening conditions has a greater correlation with the optimization target; the screening condition is ρ X,Y ≥0.6; Step 3: Establish a long short-term memory neural network (LSTM) prediction model with time series prediction characteristics for the parameter data with greater correlation screened out in step 2, and obtain the forecast data of the demand for compressed air and steam after 5 minutes, 1 hour and 24 hours; Step 3 specifically includes the following steps: Step 3.1, establish the initial prediction model; Step 3.2, dividing the parameter data with greater correlation screened out in step 2 into a training data set, a validation data set, and a test data set; Step 3.3, train the initial prediction model using the training data set to obtain a training prediction model; Step 3.4: Use the validation data set to adjust the parameters of the training prediction model; Step 3.5, repeat steps 3.3 and 3.4 until the training prediction model reaches the set training termination threshold, and determine the best prediction model; Step 3.6: Based on the best prediction model in step 3.5, obtain the forecast data of the demand for compressed air and steam after 5 minutes, 1 hour and 24 hours; Step 4: Based on the historical data and the forecast data in step 3, the PSO algorithm is used to optimize the operation plan with the goal of minimizing the total cost of electricity and steam operation to obtain the optimal operation plan; Step 5: The compressed air and steam combined supply system operates based on the optimal operation plan in step 4.

2. The operation optimization method of the compressed air and steam combined supply system according to claim 1, characterized in that: In step 2, the normalized calculation formula is: Among them, X max is the maximum value of the sample data, X min is the minimum value of the sample data, X is the sample data, X 归一 is the normalized data.

3. The operation optimization method of the compressed air and steam combined supply system according to claim 1, characterized in that: In step 2, the Pearson correlation coefficient ρ X,Y The calculation formula is: Among them, cov is the covariance, σ is the standard deviation, E is the expectation, and X and Y are parameters.

4. A compressed air and steam combined supply system, characterized in that: include: A steam turbine, a steam-driven centrifugal air compressor, and an electrically-driven centrifugal air compressor are operated by the operation optimization method described in any one of claims 1 to 3.

Images