A method for determining pressure fluctuation of a heating pipe network

Abstract

The application discloses a kind of for heating pipe network pressure fluctuation determination method, including the pressure data of heat exchange station in heating pipe network is collected;According to different time interval, time series data is generated by downsampling;The same amplitude of time series data is calculated;According to the same amplitude of a period of time before mutation point set, the same amplitude threshold value is calculated;According to the same amplitude of a period of time before mutation point set and the same amplitude of time series data corresponding time point, the absolute value of same amplitude difference value is calculated;If the absolute value of same amplitude difference value is greater than the same amplitude threshold value, then the pressure data of corresponding time point is judged as pressure abnormal fluctuation data;The application is according to the pressure data variation trend of heating pipe network, using time series mutation point detection method to carry out abnormal fluctuation detection and determination to the pressure data of heating pipe network, solves the problem that the fixed threshold is not accurate under multiple pressure ranges, leading to the problem that the false alarm rate and the missing alarm rate of pressure abnormal fluctuation of heating pipe network are relatively large.

Description

Technical Field

[0001] This invention belongs to the technical field of pressure fluctuation detection in heating system pipelines, and specifically relates to a method for determining pressure fluctuations in heating pipelines. Background Technology

[0002] With the rapid urbanization in recent years, my country has built the world's largest centralized heating system. The heating pipeline network is a fundamental component of this system, playing a crucial role in distributing heat to end users. This network comprises complex structures such as branching and looping pipes. Pressure fluctuations within the heating pipeline network can pose a serious challenge to the stable and reliable operation of the heating system. Therefore, real-time monitoring of the heating pipeline network pressure, timely detection and alarm of pressure fluctuations, and prevention of fluid oscillations or pipe bursts caused by abnormal pressure fluctuations are essential to ensure the stability and safety of the heating system.

[0003] Numerous factors can cause pressure fluctuations in heating networks, such as pipe material, laying method, construction method, operation and control, and pipe or connector malfunctions. Accurate detection of pressure fluctuations in heating networks is a highly challenging task. Current methods for detecting heating pressure fluctuations primarily rely on differential analysis of pressure time-series data and the application of fixed thresholds. However, different measuring points within the heating network have varying pressure ranges, leading to inaccurate threshold settings and misjudgments of pressure fluctuations. Summary of the Invention

[0004] The purpose of this invention is to address the aforementioned shortcomings in the prior art by providing a method for determining pressure fluctuations in heating pipe networks, thereby solving the problems of inaccurate threshold settings and misjudgments of pressure fluctuations in heating pipe network pressure detection.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A method for determining pressure fluctuations in a heating network, comprising the following steps:

[0007] S1. Collect pressure data of heat exchange stations within the heating network;

[0008] S2. Preprocess the collected pressure data and downsample the pressure data according to different time intervals to obtain time series data corresponding to different time intervals;

[0009] S3. Based on a time series data, perform abrupt change point detection to obtain a set of abrupt change points in the pressure data, and calculate the year-on-year amplitude of the time series data;

[0010] S4. Calculate the year-on-year amplitude threshold based on the year-on-year amplitude over a period of time prior to the set of mutation points;

[0011] S5. Calculate the absolute value of the difference in amplitude between the same period of time and the amplitude at the corresponding time point in the time series data, based on the amplitude of the same period of time in the period of time before the set of mutation points.

[0012] S6. If the absolute value of the year-on-year amplitude difference is greater than the year-on-year amplitude threshold, the pressure data at the corresponding time point is judged to be pressure abnormal fluctuation data; otherwise, the pressure data at the corresponding time point is normal data.

[0013] S7. Repeat steps S4 to S6 until all time points corresponding to the time series data are traversed to determine the abnormal pressure fluctuation data.

[0014] S8. Repeat steps S3 to S7 to determine abnormal pressure fluctuations in the remaining time series data at different time intervals.

[0015] Furthermore, step S2 involves preprocessing the collected pressure data, including:

[0016] Pressure data is preprocessed to remove outliers and generate pressure time-series data:

[0017] P = {p1, ... p} i ...p n}

[0018] Where n is the total number of samples, p i Let be the pressure value at time i.

[0019] Further, in step S2, the pressure data is downsampled at different time intervals to obtain time-series data corresponding to different time intervals, including:

[0020] The time series data P is downsampled according to the time interval t1 to generate the time series data E = {e1, ... e1}. i ...e n}, where e i Let i be the pressure value at time i;

[0021] The time series data P is downsampled according to the time interval t2 to generate the time series data F = {f1, ... f2}. i ...f n}, where f i Let i be the pressure value at time i;

[0022] The time series data P is downsampled according to the time interval t3 to generate the time series data G = {g1, ... g3}. i ...g n}, where g i Let be the pressure value at time i.

[0023] Furthermore, in step S3, mutation point detection is performed based on time-series data to obtain a set of mutation points in the stress data, including:

[0024] A time window is formed by a fixed time interval w, and a loss function C is constructed. (a，a+w) :

[0025]

[0026] Among them, C (a，a+w) Let g be the loss function from time a to time a+w; w is the fixed time interval of the window; g avg The average pressure from time a to time a+w;

[0027] A time window is formed with a fixed time interval w, and two adjacent windows W are... t W t+1 Slide along the timeline to calculate the difference d(C) between the two time windows. (a，a+w) C (a+w，a+2w) ):

[0028] d(C (a，a+w) C (a+w，a+2w) ) = C (a，a+2w) -C (a，a+w) -C (a+w，a+2w)

[0029] Wherein d(C (a，a+w) C (a+w，a+2w) ) represents the difference in pressure data between time periods [a, a+w] and [a+w, a+2w]; Ca, a+2w represents the loss function from time a to time a+2w; Ca, a+w represents the loss function from time a to time a+w; C (a+w，a+2w) Let w be the loss function from time a+w to time a+2w; w is the fixed time interval of the window.

[0030] Calculate d(C) using a sliding window method (a，a+w) C (a+w，a+2w) The set of abrupt change points M = {m1, ..., m2} that correspond to the first k maximum values ​​and generate pressure data at those times is M = {m1, ..., m3}. i ,...m k}, where k is the total number of mutation points, m i Let i be the pressure value at time i within a certain time interval.

[0031] Further, step S3 calculates the year-on-year amplitude of the time series data, including:

[0032] Calculate the year-on-year amplitude CA = {ca1, ..., ca1} at the corresponding time points in the time series data. i , ..., ca n}, the same amplitude ca at time point ii :

[0033]

[0034] Among them, ca i Let e ​​be the year-on-year amplitude of the pressure at time i; i Let e ​​be the pressure value at time i; i-1 Let be the pressure value at time i-1.

[0035] Furthermore, step S4 specifically includes:

[0036] Obtain the set of all time points L = {l1, ..., li, ... l1} from the set of mutation points M. k}, where k is the total number of mutation points, l i Let i be the moment of the mutation point.

[0037] Obtain a certain moment l from set L of the set of the same amplitude CA. i The first X sets of the same amplitude are CAX = {cax1, ..., cax} i ,...cax x}, where x is the total number of samples, cax i Let l be the year-on-year amplitude at time i. i Let i be the moment of the mutation point numbered i in set L;

[0038] Calculate the year-on-year amplitude threshold θ based on the first X sets of year-on-year amplitudes, CAX:

[0039] θ = max(cax) max -cax avg cax avg -cax min )

[0040] Where θ is the year-on-year amplitude threshold; cax max The maximum year-on-year amplitude in the CAX set; cax min The minimum year-on-year amplitude in the CAX set; cax avg This represents the mean of the year-on-year amplitudes in the CAX set.

[0041] Furthermore, step S5 specifically includes:

[0042] Based on set L at a certain moment l i The same amplitude ca in the CA set i The amplitude of the CAX set at the last moment is the same as that of the CAX set. x Calculate the year-on-year amplitude ca i With cax x absolute value of the difference

[0043] The method for determining pressure fluctuations in heating pipe networks provided by this invention has the following beneficial effects:

[0044] This invention combines the advantages of inflection point detection and short-term comparison. First, it uses inflection point detection to identify the moment of pressure abrupt change. Then, it uses short-term comparison to filter and obtain pressure anomaly fluctuation data for a specific time interval. By detecting anomalies at multiple time intervals, it obtains pressure anomalies across different time dimensions, reducing the false negative rate by approximately 5% compared to methods with fixed thresholds. This provides more accurate data support for the stable operation of heating networks. Attached Figure Description

[0045] Figure 1 A flowchart for determining pressure fluctuations in heating pipe networks. Detailed Implementation

[0046] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0047] Example 1

[0048] This embodiment describes a method for determining pressure fluctuations in a heating network. Based on the changing trends of the heating network's pressure data, this embodiment utilizes a time-series abrupt change point detection method to detect and determine abnormal fluctuations in the heating network's pressure data. (Refer to...) Figure 1 Specifically, it includes the following steps:

[0049] Step S1: Collect pressure data from heat exchange stations within the heating network;

[0050] Step S2: Preprocess the collected pressure data and downsample the pressure data according to different time intervals to obtain time series data corresponding to different time intervals;

[0051] Step S3: Based on a time series data, perform abrupt change point detection to obtain a set of abrupt change points in the pressure data, and calculate the year-on-year amplitude of the time series data;

[0052] Step S4: Calculate the year-on-year amplitude threshold based on the year-on-year amplitude over a period of time prior to the set of mutation points;

[0053] Step S5: Calculate the absolute value of the difference in amplitude between the same period of time and the amplitude at the corresponding time point in the time series data, based on the amplitude of the same period of time in the period of time before the set of mutation points.

[0054] Step S6: If the absolute value of the year-on-year amplitude difference is greater than the year-on-year amplitude threshold, then the pressure data at the corresponding time point is determined to be pressure abnormal fluctuation data; otherwise, the pressure data at the corresponding time point is normal data.

[0055] Step S7: Repeat steps S4 to S6 until all time points corresponding to the time series data are traversed to determine abnormal pressure fluctuation data.

[0056] Step S8: Repeat steps S3 to S7 to determine abnormal pressure fluctuations in the remaining time series data at different time intervals.

[0057] Example 2

[0058] This embodiment, as a preferred embodiment of embodiment 1, provides a specific implementation method based on the method steps of embodiment 1. Since the heating network includes multiple heat exchange stations, and each heat exchange station contains multiple pressure sensors such as primary supply pressure and primary return pressure sensors, the following description uses data collected from one pressure sensor at one of the heat exchange stations to describe the pressure anomaly fluctuation detection method, which specifically includes the following steps:

[0059] Step S1: Acquisition of pressure data;

[0060] Data from a pressure sensor at a heat exchange station within the heating network is collected at 10-second intervals and stored in a database.

[0061] Step S2: Processing pressure data and generating time series data;

[0062] The pressure values ​​are preprocessed to remove outliers, generating pressure time-series data P = {p1, ... p2}. i ...p n}, where n is the total number of samples, p i Let i be the pressure value at time i;

[0063] The time series data P is downsampled at 1-minute intervals to generate time series data E = {e1, ... e2}. i ...e n}, where n is the total number of samples, e i Let i be the pressure value at time i;

[0064] The time series data P is downsampled at 10-minute intervals to generate time series data F = {f1, ... f2}. i ...f n}, where n is the total number of samples, f i Let i be the pressure value at time i;

[0065] The time series data P is downsampled at 30-minute intervals to generate time series data G = {g1, ... g2}.i ...g n}, where n is the total number of samples, g i Let i be the pressure value at time i;

[0066] This embodiment uses time-series data E as an example for the following description;

[0067] Step S3: Calculate the set of abrupt change points and the year-on-year amplitude of the pressure data;

[0068] A time window is formed by a fixed time interval w, and a loss function C is constructed. (a，a+w) :

[0069]

[0070] Among them, C (a，a+w) Let e ​​be the loss function from time a to time a+w; w is the fixed time interval of the window; e i e is the pressure value at time i; avg The average pressure from time a to time a+w;

[0071] A time window is formed with a fixed time interval w, and two adjacent windows W are... t W t+1 Slide along the timeline to calculate the difference d(C) between the two time windows. (a，a+w) C (a+w，a+2w) ):

[0072] d(C (a，a+w) C (a+w，a+2w) ) = C (a，a+2w) -C (a，a+w) -C (a+w，a+2w) (2)

[0073] Wherein d(C (a，a+w) C (a+w，a+2w) ) represents the difference in pressure data between time periods [a, a+w] and [a+w, a+2w]; Ca, a+2w represents the loss function from time a to time a+2w; Ca, a+w represents the loss function from time a to time a+w; C (a+w，a+2w) Let w be the loss function from time a+w to time a+2w; w is the fixed time interval of the window.

[0074] Calculate d(C) using a sliding window method (a，a+w) C (a+w，a+2w) The set of abrupt change points M = {m1, ..., m2} that correspond to the first k maximum values ​​and generate pressure data at those times is M = {m1, ..., m3}. i ,...m k}, where k is the total number of mutation points, m i Let i be the pressure value at time i within a certain time interval.

[0075] Calculate the year-on-year amplitude CA = {ca1, ..., ca1} at the corresponding time point in the time series data E. i , ..., ca n}, the same amplitude ca at time point i i Specifically:

[0076]

[0077] Among them, ca i Let e ​​be the year-on-year amplitude of the pressure at time i; i Let e ​​be the pressure value at time i; i-1 Let be the pressure value at time i-1.

[0078] Step S4: Calculate the same-year amplitude threshold;

[0079] Obtain the set of all time points L = {l1, ..., l1} from the set of mutation points M. i , ...l k}, where k is the total number of mutation points, l i Let i be the mutation point time.

[0080] Obtain a specific moment l from set L in set CA. i The first X (default 60) asymmetric amplitude sets CAX = {cax1, ..., cax1} i ,...cax x}, where x is the total number of samples, cax i Let l be the year-on-year amplitude at time i. i Let i be the mutation point time in set L with number i.

[0081] The year-on-year amplitude threshold θ is calculated based on the CAX set, as follows:

[0082] θ = max(cax) max -cax avg cax avg -cax min (4)

[0083] Where θ is the year-on-year amplitude threshold; cax max The maximum year-on-year amplitude in the CAX set; cax min The minimum year-on-year amplitude in the CAX set; cax avg This represents the mean of the year-on-year amplitudes in the CAX set.

[0084] Step S5: Calculate the absolute value of the year-on-year amplitude difference;

[0085] Get the value at a certain moment l in set L i The same amplitude ca in the CA seti and the last moment of the CAX set cax x And calculate the year-on-year amplitude ca i Compared with the same period amplitude cax x absolute value of difference

[0086] Step S6: If the absolute value of the year-on-year amplitude difference is greater than the year-on-year amplitude threshold, then the pressure data at the corresponding time point is determined to be pressure abnormal fluctuation data; otherwise, the pressure data at the corresponding time point is normal data.

[0087] That is, if The pressure data is then determined to be an abnormal fluctuation.

[0088] Step S7: Repeat steps S4 to S6 until all time points corresponding to the time series data are traversed to determine abnormal pressure fluctuation data.

[0089] That is, the judgment is based on the abnormal pressure fluctuations in steps S4 to S6 of the set L loop.

[0090] Step S8: Repeat steps S3 to S7 to determine the pressure abnormal fluctuations in the remaining time series data at different time intervals, that is, determine the pressure abnormal fluctuations of the three sequences E, F, and G.

[0091] Although specific embodiments of the invention have been described in detail with reference to the accompanying drawings, this should not be construed as limiting the scope of protection of this patent. Various modifications and variations that can be made by a person skilled in the art without inventive effort within the scope described in the claims still fall within the scope of protection of this patent.

Claims

1. A method for determining pressure fluctuations in a heating network, characterized in that, Includes the following steps: S1. Collect pressure data of heat exchange stations within the heating network; S2. Preprocess the collected pressure data and downsample the pressure data according to different time intervals to obtain time series data corresponding to different time intervals; S3. Based on a time series data, perform abrupt change point detection to obtain a set of abrupt change points in the pressure data, and calculate the year-on-year amplitude of the time series data; S4. Calculate the year-on-year amplitude threshold based on the year-on-year amplitude over a period of time prior to the set of mutation points; S5. Calculate the absolute value of the difference in amplitude between the same period of time and the amplitude at the corresponding time point in the time series data, based on the amplitude of the same period of time in the period of time before the set of mutation points. S6. If the absolute value of the difference in amplitude between the same period last year is greater than the threshold of amplitude between the same period last year, then the pressure data at the corresponding time point is judged to be pressure abnormal fluctuation data. Conversely, the pressure data at the corresponding time point is normal. S7. Repeat steps S4 to S6 until all time points corresponding to the time series data are traversed to determine the abnormal pressure fluctuation data. S8. Repeat steps S3 to S7 to determine abnormal pressure fluctuations in the remaining time series data at different time intervals. In step S2, the pressure data is downsampled at different time intervals to obtain time-series data corresponding to different time intervals, including: Time series data according to t Time series data is generated by downsampling at time intervals of 1. ,in, In order to be in The pressure value at any given moment; Time series data according to t Time series data is generated by downsampling at a time interval of 2. ,in, In order to be in The pressure value at any given moment; Time series data according to t Time series data is generated by downsampling at time intervals of 3. ,in, In order to be in The pressure value at any given moment; In step S3, mutation point detection is performed based on time-series data to obtain a set of mutation points in the stress data, including: At fixed time intervals Forming time windows and constructing loss functions : in, For a moment At the time The loss function; This is a fixed time interval for the window; For a moment At the time Mean internal pressure; At fixed time intervals Create a time window, connecting two adjacent windows. , Slide along the timeline to calculate the difference between the two time windows. : in, From a time period With time period Differences in pressure data; For a moment At the time The loss function; For a moment At the time The loss function; From time At the time The loss function; This is a fixed time interval for the window; Calculate using a sliding window forward The set of abrupt change points that generate pressure data at the time corresponding to the maximum value. ,in, This represents the total number of mutation points. For a certain time interval The pressure value at any given moment; The year-on-year amplitude of the time series data is calculated in step S3, including: Calculate the year-on-year amplitude at the corresponding time point in the time series data. At time point Year-on-year amplitude : in, In order to be in The year-on-year amplitude of pressure at any given moment; In order to be in Constant pressure value; In order to be in The pressure value at any given moment; Step S4 specifically includes: Obtain the set of mutation points Set of all time points in the middle , in, The total number of mutation points. For the number The moment of change; In the same period of the amplitude Retrieving a collection from a collection At a certain moment The former A set of year-on-year amplitudes ,in The total number of samples, For at any time The year-on-year amplitude, For set The Chinese number is The moment of change; According to the previous A set of year-on-year amplitudes Calculate the year-on-year amplitude threshold : = in, The threshold for year-on-year amplitude; for The maximum amplitude of the same period in the set; for The minimum year-on-year amplitude in the set; for The average amplitude of the same period in the set.

2. The method for determining pressure fluctuations in a heating network according to claim 1, characterized in that, The preprocessing of the collected pressure data in step S2 includes: Pressure data is preprocessed to remove outliers and generate pressure time-series data: Where n is the total number of samples, In order to be in The pressure level at any given moment.

3. The method for determining pressure fluctuations in a heating network according to claim 2, characterized in that, Step S5 specifically includes: According to the set At a certain moment of Year-on-year amplitude in the set and The year-on-year amplitude at the last moment of the set Calculate the year-on-year amplitude and absolute value of the difference .

Images