Automatic heat supply regulation and control system based on waste heat of power plant
By using adaptive smoothing windows and weighted analysis, the problem of data noise interference in the power plant waste heat heating system was solved, enabling highly accurate prediction and control of heating status data and improving the automatic control capability of the heating system.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-30
- Publication Date
- 2026-03-24
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In existing power plant waste heat heating systems, heating status data is easily affected by electromagnetic interference and sensor errors, leading to reduced prediction accuracy of the ARIMA algorithm. Furthermore, the fixed window and time weight cannot adapt to the changing characteristics of power plant waste heat heating, affecting the accuracy of heating status data prediction and control.
The heating status data sequence is obtained through the data acquisition module. The local fluctuation coefficient and data cluster are obtained based on the recent historical change characteristics of the data points using the window analysis module. The smoothing weight coefficient is calculated based on the consistency of change and distance characteristics of adjacent data points using the weight analysis module. The data is smoothed using an adaptive smoothing window. Finally, the heating status is predicted and controlled using the ARIMA algorithm.
It improves the smoothing effect and prediction accuracy of heating status data, ensuring the accuracy of heating regulation, while preserving the local details and trend characteristics of the data, and enhancing the automatic regulation capability of the heating system.
Smart Images
Figure CN121720153A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and more specifically to an automatic heating control system based on waste heat from power plants. Background Technology
[0002] Currently, common methods for utilizing waste heat from power plants include combined heat and power (CHP) and waste heat heating. Waste heat heating involves recovering waste heat emitted from power plants through heat exchangers and other equipment and transmitting it to the city's heating network. This method not only improves energy efficiency but also reduces heating costs, resulting in significant economic and environmental benefits. Traditional prediction algorithms, such as the ARIMA autoregressive moving average model, can predict heating status data such as temperature, pressure, and exhaust steam flow, thereby achieving automatic heating control and improving heating efficiency. However, in heating systems, parameters such as temperature and exhaust steam flow are easily affected by factors such as electromagnetic interference from power plants and sensor errors during data acquisition, leading to noise interference in the heating status data and affecting the prediction accuracy of the ARIMA algorithm.
[0003] Therefore, in order to improve the accuracy of prediction, it is necessary to smooth the collected heating status data. The existing weighted moving average algorithm can smooth noise and extract long-term trends by assigning different weights to data at different times. However, the fixed window and time weight during smoothing may not be able to adapt to the changing characteristics of power plant waste heat heating, and noise smoothing may be inaccurate, thus affecting the accuracy of heating status data prediction and heating regulation. Summary of the Invention
[0004] To address the aforementioned technical problems, the present invention aims to provide an automatic heating control system based on waste heat from power plants. The specific technical solution adopted is as follows: The data acquisition module is used to acquire the status data sequence for monitoring the heating status; The window analysis module is used to obtain the local fluctuation coefficient of the data points based on the recent historical data change characteristics of the data points in the state data sequence; to obtain different data clusters in the state data sequence based on the data points and the local fluctuation coefficients; to obtain the average fluctuation trend coefficient based on the fluctuation and trend characteristics of continuous data segments in the data clusters; and to obtain the adaptive smoothing window of the data points based on the data smoothing characteristics of the continuous data segments and the average fluctuation trend coefficient. The weight analysis module is used to obtain the consistency of change of adjacent data points based on the data discrete characteristics within the adaptive smoothing window of the data points; and to obtain the smoothing weight coefficient of adjacent data points based on the distance characteristics and consistency of change between the data points and the adjacent data points. The prediction and control module is used to smooth the data based on the smoothing weight coefficients of adjacent data points within the adaptive smoothing window of the data points to obtain a smoothed data sequence; and to predict and control the heating status based on the smoothed data sequence.
[0005] Furthermore, the step of obtaining the local fluctuation coefficient of the data point based on the change characteristics of the recent historical data of the data point in the state data sequence includes: In the formula, R represents the local fluctuation coefficient of the data point, and T represents the amount of data within the preset recent historical range of the data point. This represents the value of the t-th data point. Indicates the first The value of each data point. Represents the average difference characteristic value. This represents the slope between the t-th data point and the previous data point. Indicates the first The slope of each data point relative to the previous data point This represents the characteristic value of the average change difference.
[0006] Furthermore, the step of obtaining different data clusters in the state data sequence based on the data points and the local fluctuation coefficients includes: A planar coordinate system is constructed for the data points, with the horizontal axis representing the numerical value of the data points and the vertical axis representing the local fluctuation coefficient of the data points. The data points are then clustered in the planar coordinate system using the K-nearest neighbor clustering algorithm to obtain different data clusters.
[0007] Further, the step of obtaining the average fluctuation trend coefficient based on the fluctuation characteristics and trend characteristics of continuous data segments in the data cluster includes: The range of data points in the data cluster with a number of consecutive points exceeding a constant 3 is taken as a continuous data segment; the coefficient of variation of the continuous data segment is calculated to obtain the first fluctuation characteristic value; the absolute value of the fitting slope of the continuous data segment is calculated to obtain the trend characteristic value; the sum of the absolute values of the differences between each data point and the previous data point in the continuous data segment is calculated and positively correlated to obtain the second fluctuation characteristic value; the product of the first fluctuation characteristic value, the trend characteristic value and the second fluctuation characteristic value is calculated and normalized to obtain the average fluctuation trend coefficient of the continuous data segment.
[0008] Further, the step of obtaining the adaptive smoothing window for the data points based on the data smoothing characteristics of the continuous data segments and the average fluctuation trend coefficient includes: The continuous data segment is fitted, and the absolute value of the difference between the last term of the continuous data segment and the corresponding fitted value is calculated to obtain the smoothness difference of the continuous data segment; the average fluctuation trend coefficient corresponding to the continuous data segment with the smallest smoothness difference is used as the judgment threshold; the length of the continuous data segment with the smallest smoothness difference is used as the benchmark length. The adjustment degree is obtained by calculating the difference between the average fluctuation trend coefficient of any continuous data segment in the data cluster and the judgment threshold; the adjustment amount is obtained by multiplying the adjustment degree by the baseline length; the adaptive window length of any data point in the continuous data segment is obtained by calculating the sum of the baseline length and the adjustment amount and rounding it up; for other data points in the data cluster that are not in continuous data segments, the adaptive smoothing window of the other data points is the average of the adaptive window lengths of the preceding and following continuous data segments and rounded up; the adaptive smoothing window of the data points is constructed based on the adaptive window length of the data points.
[0009] Furthermore, the step of obtaining the consistency of change of adjacent data points based on the data discrete features within the adaptive smoothing window of the data points includes: In the formula This represents the consistency of change of the m-th adjacent data point within the adaptive smoothing window. This represents an exponential function with the natural constant as its base. Indicates the first The values of the adjacent data points, where H represents the fitting slope of the data within the adaptive smoothing window. This represents the value of the m-th adjacent data point.
[0010] Further, the step of obtaining the smoothing weight coefficients of adjacent data points based on the distance characteristics and consistency of change between the data point and its neighboring data points includes: In the formula This represents the smoothing weight coefficient of the m-th adjacent data point. This indicates the consistency of change among the m-th adjacent data points. This represents an exponential function with the natural constant as its base. This indicates the maximum position of the adaptive smoothing window. This indicates the position of the m-th adjacent data point.
[0011] Further, the step of smoothing the data according to the smoothing weight coefficients of adjacent data points within the adaptive smoothing window to obtain a smoothed data sequence includes: In the formula This represents the final smoothing weight of the m-th adjacent data point. This represents the smoothing weight coefficient of the m-th adjacent data point. This represents the sum of the smoothing weight coefficients of all adjacent data points within the adaptive smoothing window; the sum of the products of all adjacent data points and their corresponding smoothing weight coefficients within the adaptive smoothing window is calculated to obtain the smoothing value of the data point, and the smoothed data sequence is constructed based on the smoothing values of different data points.
[0012] Furthermore, the step of predicting and regulating the heating status based on the smoothed data sequence includes: The smoothed data sequence is predicted using the ARIMA algorithm to obtain state prediction results; heating is then regulated based on these state prediction results.
[0013] The present invention has the following beneficial effects: In this invention, obtaining the local fluctuation coefficient can determine the local data fluctuation characteristics of data points, and thus enable the division of data points with similar fluctuation characteristics based on the local fluctuation coefficient; obtaining data clusters can group data points with similar values and local fluctuation characteristics. Obtaining the average fluctuation trend coefficient can characterize the volatility and trend of continuous data segments in the data cluster. Obtaining an adaptive smoothing window for data points based on the smoothing characteristics of continuous data segments and the average fluctuation trend coefficient allows the window size during data smoothing to better match the data fluctuation trend characteristics of the smoothing object, improving the smoothing effect. Obtaining the consistency of changes among adjacent data points can characterize whether the data change trends of adjacent data points within the smoothing window are consistent, further improving the accuracy of smoothing weight settings. Obtaining the smoothing weight coefficient can accurately assign the corresponding smoothing weight to each adjacent data point within the adaptive smoothing window, improving the accuracy of the smoothing result. Finally, obtaining the smoothed data sequence can retain local data details and trend characteristics while denoising, making the prediction and control of heating status based on the smoothed data sequence more accurate. Attached Figure Description
[0014] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0015] Figure 1 This is a block diagram of an automatic heating control system based on waste heat from a power plant, provided as an embodiment of the present invention. Detailed Implementation
[0016] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an automatic heating control system based on power plant waste heat proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0018] The following description, in conjunction with the accompanying drawings, details a specific scheme for an automatic heating control system based on power plant waste heat provided by the present invention.
[0019] Please see Figure 1 The diagram illustrates a block diagram of an automatic heating control system based on waste heat from a power plant, according to an embodiment of the present invention. The system includes the following steps: The data acquisition module S1 is used to acquire the status data sequence for monitoring the heating status.
[0020] In this embodiment of the invention, the implementation scenario is to automatically regulate the heating supply of waste heat from power plants, thereby improving the accuracy of regulation. First, a state data sequence for monitoring the heating status is acquired. In this embodiment, the turbine exhaust flow rate, exhaust temperature, and pressure are monitored at the heat source, and state data sequences for each monitored object are collected. The implementer can determine the objects to be monitored and the collection frequency according to the implementation scenario. Since noise may exist in the state data sequence, affecting the accuracy of sequence prediction and regulation, it is necessary to smooth the state data sequence to remove noise while preserving the true characteristics of the state data.
[0021] The window analysis module S2 is used to obtain the local fluctuation coefficient of the data points based on the recent historical data change characteristics of the data points in the state data sequence; to obtain different data clusters in the state data sequence based on the data points and the local fluctuation coefficient; to obtain the average fluctuation trend coefficient based on the fluctuation and trend characteristics of continuous data segments in the data cluster; and to obtain the adaptive smoothing window of the data points based on the data smoothing characteristics of continuous data segments and the average fluctuation trend coefficient.
[0022] Different operating conditions exist when waste heat is generated in thermal power plants, such as normal heating conditions, extreme weather conditions, and equipment malfunction conditions. The state data under these different conditions will vary, primarily in stability, rate of change, and magnitude. The size of the smoothing window is affected by data stability. For example, when the data is relatively stable, a smaller window can capture subtle changes and avoid over-smoothing leading to trend lag; while when the data is unstable, a larger window is needed to smooth short-term noise and prevent noise from affecting state judgment. Furthermore, the local fluctuation coefficient of the data points can be obtained based on the recent historical data change characteristics of the data points in the state data sequence, and the data points can be classified according to the local fluctuation coefficient and their values. Preferably, in this embodiment of the invention, the step of obtaining the local fluctuation coefficient includes: In the formula, R represents the local fluctuation coefficient of the data point, and T represents the amount of data within the preset recent historical range of the data point. In the embodiment of the present invention, the preset recent historical range is the range of the data point and the previous 9 adjacent data points. This represents the value of the t-th data point. Indicates the first The value of each data point. This represents the average difference characteristic value; the smaller the average difference characteristic value, the smaller the local data fluctuation of that data point. This represents the slope between the t-th data point and the previous data point. Indicates the first The slope of each data point relative to the previous data point This represents the average variation difference characteristic value; the smaller the average variation difference characteristic value, the more similar the local data change rates at that data point, and the smaller the fluctuation coefficient. The local fluctuation coefficient of a data point reflects the local stability characteristics of that data point, and thus, different data clusters in the state data sequence can be obtained based on the data point and the local fluctuation coefficient.
[0023] Preferably, in this embodiment of the invention, the step of obtaining data clusters includes: constructing a planar coordinate system about the data points, with the horizontal axis representing the numerical value of the data points and the vertical axis representing the local fluctuation coefficient of the data points; and clustering the data points in this planar coordinate system using the K-nearest neighbor clustering algorithm to obtain different data clusters. It should be noted that the K-nearest neighbor clustering algorithm is an existing technology, and the numerical values and local data change characteristics of the data points in each cluster are similar.
[0024] Furthermore, the selection of the smoothing window plays a dual role in time series analysis: responding to short-term fluctuations and identifying long-term trends. Smaller windows are sensitive to instantaneous changes in data and can quickly capture local fluctuation characteristics; larger windows filter short-term fluctuations and highlight long-term trends. Therefore, optimizing the window size requires considering both data volatility and trend, with the main principle being to suppress noise interference while preserving the effective trend. The performance of the ARIMA prediction algorithm depends on the clarity of the data trend. If the window is too large, it may filter out weak but crucial trend signals, leading to prediction lag; if the window is too small, it may misjudge noise as a trend. Therefore, when the data trend is not significant, a smaller window should be used to avoid over-smoothing and losing local change characteristics; when a stable trend exists, the window can be enlarged to aggregate more historical information to strengthen trend characteristics while suppressing the interference of short-term fluctuations on the prediction. The average volatility trend coefficient can then be obtained based on the volatility and trend characteristics of continuous data segments within the data cluster.
[0025] Preferably, in this embodiment of the invention, the step of obtaining the average fluctuation trend coefficient includes: taking the range of data points in the data cluster with a number of consecutive moments exceeding a constant 3 as a continuous data segment; calculating the coefficient of variation of the continuous data segment to obtain a first fluctuation characteristic value; the larger the coefficient of variation, the greater the volatility of the continuous data segment. Calculating the absolute value of the fitting slope of the continuous data segment to obtain a trend characteristic value; the larger the absolute value of the fitting slope, the more obvious the trend. Calculating the sum of the absolute values of the differences between each data point and the previous data point in the continuous data segment and mapping them positively to obtain a second fluctuation characteristic value; in this embodiment of the invention, using... Perform a positive correlation mapping, where 'a' represents the mapping object. This represents an exponential function with the natural constant as its base. A larger second volatility eigenvalue and a larger trend eigenvalue indicate that the continuous data segment is more likely to exhibit a single, clear trend; conversely, a smaller value suggests multiple trends or no clear trend. The average volatility trend coefficient of the continuous data segment is obtained by calculating and normalizing the product of the first volatility eigenvalue, the trend eigenvalue, and the second volatility eigenvalue. Greater volatility and trend indicate a larger average volatility trend coefficient and a more pronounced trend in the continuous data segment.
[0026] Further, after obtaining the average fluctuation trend coefficient of different continuous data segments in the data cluster, an adaptive smoothing window for the data points can be obtained based on the data smoothing characteristics and the average fluctuation trend coefficient of the continuous data segments. Preferably, in this embodiment of the invention, the step of obtaining the adaptive smoothing window includes: fitting the continuous data segments, with the last term not participating in the fitting; calculating the absolute value of the difference between the last term of the continuous data segment and the corresponding fitted value to obtain the smoothing difference of the continuous data segment; the smaller the smoothing difference, the closer the actual value of the last term is to the corresponding fitted value, and the more the length of the continuous data segment matches the length of the smoothing window, ensuring prediction accuracy while smoothing. The average fluctuation trend coefficient corresponding to the continuous data segment with the smallest smoothing difference is then used as a judgment threshold; since the data fluctuation characteristics are similar in the same data cluster, the average fluctuation trend coefficient of the continuous data segment can be used as a judgment threshold to measure the smoothing window corresponding to other continuous sequence segments. The length of the continuous data segment with the smallest smoothing difference is used as the baseline length. The adjustment degree is obtained by calculating the difference between the average fluctuation trend coefficient of any continuous data segment in the data cluster and the judgment threshold. When the adjustment degree is positive, it means that the fluctuation trend of the continuous data segment is larger and the data changes more drastically. Therefore, in order to capture the long-term trend and avoid the influence of short-term fluctuations, the smoothing window can be appropriately expanded. When the adjustment degree is negative, it means that the data change of the continuous data segment is gentle. The smoothing window needs to be appropriately reduced to improve the sensitivity of the prediction model to local details. Then, the product of the adjustment degree and the baseline length is calculated to obtain the adjustment amount. The sum of the baseline length and the adjustment amount is calculated and rounded up to obtain the adaptive window length of any data point in the continuous data segment. Since all data points in the continuous data segment are in the same data cluster, the historical and recent data fluctuation characteristics of each data point are similar. Therefore, the smoothing window length of all data points in the continuous data segment is the same adaptive window length. For other data points in the data cluster that are not in continuous data segments, the adaptive smoothing window of the other data points is the average of the adaptive window lengths of the preceding and following continuous data segments and rounded up, so that its smoothing window is consistent with that of nearby data points. An adaptive smoothing window is constructed based on the adaptive window length of the data points. This adaptive smoothing window is the range of data points within the historically adjacent adaptive window length range.
[0027] The weight analysis module S3 is used to obtain the consistency of change of adjacent data points based on the data discrete characteristics within the adaptive smoothing window of the data points; and to obtain the smoothing weight coefficient of adjacent data points based on the distance characteristics and consistency of change between the data points and adjacent data points.
[0028] Because the status data of heating monitoring exhibits regular changes—for example, during off-peak electricity consumption periods, power plant waste heat supply is sufficient, and with reduced power generation during these periods, more steam can be used for heating, thereby increasing the waste heat supply and potentially raising the heating temperature—the heating temperature may decrease during peak electricity consumption periods or when waste heat is insufficient. Furthermore, factors such as power generation load and waste heat recovery efficiency may fluctuate randomly due to grid dispatching, equipment failures, or changes in fuel quality, leading to unstable waste heat supply. Therefore, it cannot be guaranteed that adjacent data points in the status data sequence will simultaneously be in the same cluster. Thus, setting weights solely based on time decay during smoothing may increase errors or be affected by noise. Since data within the smoothing window are temporally adjacent, each data point within the window should exhibit a consistent behavior with the overall data changes within that window. The consistency of changes among adjacent data points can be obtained based on the discrete characteristics of the data within the adaptive smoothing window. Preferably, in this embodiment of the invention, the step of obtaining the consistency of changes includes: In the formula, This represents the consistency of change among the m-th adjacent data points within the adaptive smoothing window. This represents an exponential function with the natural constant as its base. Indicates the first The values of the adjacent data points, where H represents the slope of the fitted data within the adaptive smoothing window. This represents the value of the m-th adjacent data point. When... and The more similar they are, the more they mean The more consistent the data point is with the overall data change trend within the adaptive smoothing window, the greater the consistency of change. A smaller consistency of change means that the m-th adjacent data point is less consistent with the local data change trend, and its weight should be smaller during smoothing. Therefore, the smoothing weight coefficients of adjacent data points can be obtained based on the distance characteristics and consistency of change between the data point and its neighbors. Preferably, in this embodiment of the invention, the step of obtaining the smoothing weight coefficients of adjacent data points includes: In the formula, This represents the smoothing weight coefficient of the m-th adjacent data point. This indicates the consistency of change among the m-th adjacent data points. This represents an exponential function with the natural constant as its base. This indicates the maximum position of the adaptive smoothing window. This represents the position of the m-th adjacent data point. A larger position means that the adjacent data point is closer to the smoothing object of the adaptive smoothing window, while a smaller position means that the adjacent data point is from a more distant historical point. Therefore, when the position of a data point within the adaptive smoothing window is larger and the consistency of its changes is greater, the smoothing weight coefficient of that adjacent data point is larger, and its contribution to smoothing the object is greater.
[0029] The prediction and control module S4 is used to smooth the data based on the smoothing weight coefficients of adjacent data points within the adaptive smoothing window of the data points to obtain a smoothed data sequence; and to predict and control the heating status based on the smoothed data sequence.
[0030] After obtaining the smoothing weight coefficient of each adjacent data point within the adaptive smoothing window of the data point, smoothing can be performed according to the smoothing weight coefficient of the adjacent data points within the adaptive smoothing window of the data point to obtain a smoothed data sequence; preferably, in this embodiment of the invention, the step of obtaining the smoothed data sequence includes: In the formula This represents the final smoothing weight of the m-th adjacent data point. This represents the smoothing weight coefficient of the m-th adjacent data point. This represents the sum of the smoothing weight coefficients of all adjacent data points within the adaptive smoothing window; this formula ensures that the sum of the smoothing weights of all adjacent data points within the window is a constant of 1. The smoothed value of a data point is obtained by calculating the sum of the products of all adjacent data points and their corresponding smoothing weight coefficients within the adaptive smoothing window. A smoothed data sequence is constructed based on the smoothed values of different data points, and the order of the smoothed data sequence is consistent with the state data sequence. By customizing the smoothing window and smoothing weights during smoothing, the smoothing effect of data points can be improved, preserving local variations and long-term trend characteristics of the data while removing noise, thereby improving the accuracy of heating state prediction based on the smoothed data sequence. Finally, heating state prediction and regulation can be performed based on the smoothed data sequence. The ARIMA algorithm is used to predict the smoothed data sequence to obtain the state prediction results; heating is then regulated based on the state prediction results, improving the accuracy of the adjustment. It should be noted that the ARIMA prediction algorithm is existing technology, and the specific steps will not be elaborated here. For example, if the state prediction result indicates that the heating temperature will drop in the future, the control system can automatically adjust the waste heat supply status in advance to keep the heating stable. Implementers can determine the control method based on the state prediction result according to the implementation scenario, and there are no restrictions here.
[0031] In summary, this invention provides an automatic heating control system based on power plant waste heat. It obtains local fluctuation coefficients based on the recent historical data variation characteristics of data points in a state data sequence; obtains data clusters based on data points and local fluctuation coefficients; obtains average fluctuation trend coefficients based on the fluctuation and trend characteristics of continuous data segments within the data clusters; obtains an adaptive smoothing window based on the data smoothing characteristics of continuous data segments and the average fluctuation trend coefficient; obtains the consistency of change among adjacent data points based on the data dispersion characteristics within the adaptive smoothing window; and obtains smoothing weight coefficients based on the distance characteristics and consistency of change between data points and adjacent data points. This invention smooths the data by using the smoothing weight coefficients of adjacent data points within the adaptive smoothing window to obtain a smoothed data sequence; and performs heating state prediction and control based on the smoothed data sequence, thus improving the accuracy of prediction and control.
[0032] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0033] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
Claims
1. An automatic heating control system based on waste heat from power plants, characterized in that, The system includes the following modules: The data acquisition module is used to acquire the status data sequence for monitoring the heating status; The window analysis module is used to obtain the local fluctuation coefficient of the data points based on the recent historical data change characteristics of the data points in the state data sequence; to obtain different data clusters in the state data sequence based on the data points and the local fluctuation coefficients; to obtain the average fluctuation trend coefficient based on the fluctuation and trend characteristics of continuous data segments in the data clusters; and to obtain the adaptive smoothing window of the data points based on the data smoothing characteristics of the continuous data segments and the average fluctuation trend coefficient. The weight analysis module is used to obtain the consistency of change of adjacent data points based on the data discrete characteristics within the adaptive smoothing window of the data points; and to obtain the smoothing weight coefficient of adjacent data points based on the distance characteristics and consistency of change between the data points and the adjacent data points. The prediction and control module is used to smooth the data based on the smoothing weight coefficients of adjacent data points within the adaptive smoothing window of the data points to obtain a smoothed data sequence; and to predict and control the heating status based on the smoothed data sequence.
2. The automatic heating control system based on power plant waste heat according to claim 1, characterized in that, The step of obtaining the local fluctuation coefficient of a data point based on the change characteristics of recent historical data of the data point in the state data sequence includes: In the formula, R represents the local fluctuation coefficient of the data point, and T represents the amount of data within the preset recent historical range of the data point. This represents the value of the t-th data point. Indicates the first The value of each data point. Represents the average difference characteristic value. This represents the slope between the t-th data point and the previous data point. Indicates the first The slope of each data point relative to the previous data point This represents the characteristic value of the average change difference.
3. The automatic heating control system based on power plant waste heat according to claim 1, characterized in that, The step of obtaining different data clusters in the state data sequence based on the data points and the local fluctuation coefficient includes: A planar coordinate system is constructed for the data points, with the horizontal axis representing the numerical value of the data points and the vertical axis representing the local fluctuation coefficient of the data points. The data points are then clustered in the planar coordinate system using the K-nearest neighbor clustering algorithm to obtain different data clusters.
4. The automatic heating control system based on power plant waste heat according to claim 1, characterized in that, The step of obtaining the average volatility trend coefficient based on the volatility and trend characteristics of continuous data segments in the data cluster includes: The range of data points in the data cluster with a number of consecutive points exceeding a constant 3 is taken as a continuous data segment; the coefficient of variation of the continuous data segment is calculated to obtain the first fluctuation characteristic value; the absolute value of the fitting slope of the continuous data segment is calculated to obtain the trend characteristic value; the sum of the absolute values of the differences between each data point and the previous data point in the continuous data segment is calculated and positively correlated to obtain the second fluctuation characteristic value; the product of the first fluctuation characteristic value, the trend characteristic value and the second fluctuation characteristic value is calculated and normalized to obtain the average fluctuation trend coefficient of the continuous data segment.
5. The automatic heating control system based on power plant waste heat according to claim 1, characterized in that, The step of obtaining the adaptive smoothing window for data points based on the data smoothing characteristics of the continuous data segment and the average fluctuation trend coefficient includes: The continuous data segment is fitted, and the absolute value of the difference between the last term of the continuous data segment and the corresponding fitted value is calculated to obtain the smoothness difference of the continuous data segment; the average fluctuation trend coefficient corresponding to the continuous data segment with the smallest smoothness difference is used as the judgment threshold; the length of the continuous data segment with the smallest smoothness difference is used as the benchmark length. The adjustment degree is obtained by calculating the difference between the average fluctuation trend coefficient of any continuous data segment in the data cluster and the judgment threshold; the adjustment amount is obtained by multiplying the adjustment degree by the baseline length; the adaptive window length of any data point in the continuous data segment is obtained by calculating the sum of the baseline length and the adjustment amount and rounding it up; for other data points in the data cluster that are not in continuous data segments, the adaptive smoothing window of the other data points is the average of the adaptive window lengths of the preceding and following continuous data segments and rounded up; the adaptive smoothing window of the data points is constructed based on the adaptive window length of the data points.
6. The automatic heating control system based on power plant waste heat according to claim 1, characterized in that, The step of obtaining the consistency of change of adjacent data points based on the data discrete features within the adaptive smoothing window of the data points includes: In the formula This represents the consistency of change of the m-th adjacent data point within the adaptive smoothing window. This represents an exponential function with the natural constant as its base. Indicates the first The values of the adjacent data points, where H represents the fitting slope of the data within the adaptive smoothing window. This represents the value of the m-th adjacent data point.
7. The automatic heating control system based on power plant waste heat according to claim 1, characterized in that, The step of obtaining the smoothing weight coefficient of adjacent data points based on the distance characteristics and change consistency between the data points and their adjacent data points includes: In the formula This represents the smoothing weight coefficient of the m-th adjacent data point. This indicates the consistency of change among the m-th adjacent data points. This represents an exponential function with the natural constant as its base. This indicates the maximum position of the adaptive smoothing window. This indicates the position of the m-th adjacent data point.
8. The automatic heating control system based on power plant waste heat according to claim 1, characterized in that, The step of smoothing the data according to the smoothing weight coefficients of adjacent data points within the adaptive smoothing window to obtain a smoothed data sequence includes: In the formula This represents the final smoothing weight of the m-th adjacent data point. This represents the smoothing weight coefficient of the m-th adjacent data point. This represents the sum of the smoothing weight coefficients of all adjacent data points within the adaptive smoothing window; the sum of the products of all adjacent data points and their corresponding smoothing weight coefficients within the adaptive smoothing window is calculated to obtain the smoothing value of the data point, and the smoothed data sequence is constructed based on the smoothing values of different data points.
9. The automatic heating control system based on power plant waste heat according to claim 1, characterized in that, The steps of predicting and regulating the heating status based on the smoothed data sequence include: The smoothed data sequence is predicted using the ARIMA algorithm to obtain state prediction results; heating is then regulated based on these state prediction results.