Thermal load and water supply temperature prediction method and device based on indoor temperature analysis

Abstract

The invention discloses a thermal load and water supply temperature prediction method and device based on indoor temperature analysis. The method comprises the steps of obtaining historical operation data of a building and constructing a historical operation data set; constructing a heat consumer indoor average temperature sequence; constructing a heat supply load prediction model; constructing a two-network water supply temperature prediction model; the heat supply instantaneous load of the building is predicted based on the heat supply load prediction model, and the two-network water supply temperature prediction model predicts the two-network water supply temperature of the building based on the output of the heat supply load prediction model. The heat load and the water supply temperature can be reasonably predicted, and accurate energy supply is achieved on the premise that comfort is guaranteed.

Description

Technical Field

[0001] The present invention belongs to the field of energy technology, and in particular relates to a method and device for predicting heat load and water supply temperature based on indoor temperature analysis. Background Art

[0002] The central heating area in northern China is huge, and the energy consumption of the heating system accounts for about 20% of the building energy consumption. In actual operation, it is difficult to balance the energy saving of the system and the comfort of heat users. The heat supply of the thermal system is generally higher than the demand, so it is urgent to achieve accurate energy supply between the thermal station and the heat users.

[0003] Heat load prediction is the basis for achieving accurate thermal energy supply. Many scholars and institutions have conducted a lot of research, but they mainly focus on improving the accuracy of prediction models. For example, CN202410185684.0 improves the prediction accuracy of the model through the particle swarm optimization algorithm; CN202311268008.1 reduces the prediction error by establishing the correlation between date types and weather types. CN202310000030.1 improves the accuracy of heat load prediction through room temperature correction and CNNBiLSTM algorithm.

[0004] Through the analysis of Pearson correlation coefficient method, heat load prediction is mainly related to outdoor comprehensive meteorological data, indoor temperature, water supply temperature, supply and return water temperature difference and instantaneous heat. Most of the existing heat load prediction algorithms use outdoor comprehensive meteorological data as input data, ignoring the influence of indoor temperature. On this basis, a few heat load prediction algorithms add typical indoor temperature or average temperature data to improve prediction accuracy. However, the spatial positions of various rooms in a building are different. For example, the temperatures of rooms located in the middle floor, top floor, first floor, orientation, and corner vary greatly. The typical room temperature and average temperature cannot represent the actual temperature characteristics of the entire building.

[0005] In order to match the heat supply with the heat load demand, most of the heat stations in my country currently use the climate compensation method to establish the functional relationship between the secondary water supply temperature of the heat station and the outdoor temperature. For example, CN202410228380.8 uses outdoor comprehensive meteorological parameters as input and uses machine learning methods to establish a water supply temperature prediction model. However, this model ignores the important impact of heat load demand on water supply temperature. Therefore, there is a large deviation between the predicted water supply temperature and the actual demand, resulting in oversupply.

[0006] It can be seen that the existing methods have the following technical problems that need to be solved: (1) There is a lack of indoor temperature characteristic analysis and optimization methods for the entire building (multiple rooms), campus and region, resulting in the inability to reasonably use room temperature parameters for heating load prediction; (2) The water supply temperature prediction model does not consider the heating load and its optimization, resulting in the inability to achieve accurate energy supply while maintaining comfort. Summary of the invention

[0007] The purpose of the present invention is to provide a method and device for predicting heat load and water supply temperature based on indoor temperature analysis, which can reasonably predict heat load and water supply temperature and realize accurate energy supply under the premise of ensuring comfort.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] A method for predicting heat load and water supply temperature based on indoor temperature analysis, the method comprising:

[0010] Acquire the historical operation data of the building and construct a historical operation data set, wherein the historical operation data includes the secondary network water supply temperature, the secondary network return water temperature, the heating instantaneous load, the secondary network instantaneous flow, the outdoor comprehensive meteorological parameters, the heat user room temperature parameters and the time parameters;

[0011] Constructing a heat user indoor average temperature sequence, comprising: calculating the average indoor temperature of all heat users at each moment according to the historical operation data set, and forming the heat user indoor average temperature sequence with the time series and the corresponding average indoor temperatures of all heat users;

[0012] Constructing a heating load prediction model, including: taking the instantaneous flow of the second network, the water supply temperature of the second network, the water supply temperature of the second network, the outdoor comprehensive meteorological parameters, and the average indoor temperature sequence of the heat user as the input of the heating load prediction model, taking the instantaneous heating load as the output of the heat load prediction model, and using a machine learning algorithm to construct the heating load prediction model;

[0013] Constructing a second network water supply temperature prediction model, including: taking the heating instantaneous load, outdoor comprehensive meteorological parameters, and the second network return water temperature as inputs of the second network water supply temperature prediction model, taking the second network water supply temperature as output of the second network water supply temperature prediction model, and using a machine learning algorithm to construct the second network water supply temperature prediction model;

[0014] The instantaneous heating load of the building is predicted based on the heating load prediction model, and the secondary network water supply temperature prediction model predicts the secondary network water supply temperature of the building based on the output of the heating load prediction model.

[0015] Preferably, according to the indoor average temperature sequence of the heat users, the median of the average indoor temperatures of all the heat users per day is obtained and used as the overall feature corresponding to the day;

[0016] According to the indoor average temperature sequence of the heat users, the difference between the average values ​​of the indoor temperatures of all the heat users at adjacent time differences within each day is obtained, and used as the change feature corresponding to the day;

[0017] Clustering the plurality of overall characteristics and change characteristics to obtain a plurality of daily indoor average temperature fluctuation categories;

[0018] Counting the maximum overlapping interval of the average values ​​corresponding to each daily indoor average temperature fluctuation category;

[0019] Determine the upper limit and lower limit of the daily indoor average temperature based on the maximum overlapping interval corresponding to each of the daily indoor average temperature fluctuation categories;

[0020] Divide the heat user indoor average temperature sequence into multiple subsequences by day, and use the daily indoor average temperature upper limit and lower limit to replace the temperature data in the subsequences that exceed the daily indoor average temperature upper limit and lower limit, so as to optimize the heat user indoor average temperature sequence;

[0021] Updating the heating load prediction model using the optimized indoor average temperature sequence of the heat users;

[0022] Accordingly, the output of the updated heating load prediction model is used to predict the secondary network water supply temperature of the building.

[0023] Preferably, the step of "using the upper limit and lower limit of the daily indoor average temperature to replace the temperature data in the subsequence that exceeds the upper limit and lower limit of the daily indoor average temperature" includes:

[0024] In the heat user indoor average temperature sequence divided by day, the temperature data in the subsequence that is greater than the daily indoor average temperature upper limit is replaced by the daily indoor average temperature upper limit;

[0025] The temperature data in the subsequence that is less than the lower limit of the daily indoor average temperature is replaced by the lower limit of the daily indoor average temperature.

[0026] Preferably, the daily average indoor temperature fluctuation category includes N subsequences;

[0027] The step of “counting the maximum overlapping interval of the average values ​​corresponding to each of the daily average indoor temperature fluctuation categories” comprises:

[0028] Sort the average values ​​corresponding to each hour in the N subsequences from small to large and divide them into four equal parts;

[0029] The average values ​​corresponding to the 75% and 25% positions are respectively used as the upper and lower boundaries corresponding to the hour;

[0030] A data set consisting of the upper and lower boundaries of each hour corresponding to each of the daily indoor average temperature fluctuation categories;

[0031] The maximum overlapping interval of the average values ​​corresponding to each daily average indoor temperature fluctuation category is counted based on the data set.

[0032] Preferably, the step of "determining the upper limit and lower limit of the daily average indoor temperature based on the maximum overlapping interval corresponding to each of the daily average indoor temperature fluctuation categories" includes:

[0033] The upper and lower limits of the daily average indoor temperature are determined as follows:

[0034]

[0035] in, The upper or lower limit of the daily average indoor temperature, DAY r is the number of days in the rth daily average indoor temperature fluctuation category, DAY sum is the total number of days, t r It is the upper or lower limit of the maximum overlapping interval corresponding to the r-th daily indoor average temperature fluctuation category after clustering.

[0036] Preferably, after the step of "building a heating load prediction model", the method further comprises:

[0037] Using statistical indicators such as mean square error, mean absolute error or root mean square error to verify the accuracy of the heating load prediction model;

[0038] After the step of "building a second network water supply temperature prediction model", the method further includes:

[0039] The accuracy of the secondary network water supply temperature prediction model is verified by using statistical indicators such as mean square error, mean absolute error or root mean square error.

[0040] Preferably, the step of "building a historical operation data set" includes:

[0041] The historical operation data is timestamped, and missing, duplicated and abnormal data are cleaned and normalized.

[0042] Preferably, the average indoor temperature of all heat users is calculated as shown in the following formula:

[0043]

[0044] in, is the average indoor temperature of all heat users corresponding to the i-th moment; t (a,b ) j is the j-th indoor temperature value corresponding to the i-th moment, where a is the building number, b is the household number of the building; and n is the total number of households in the building corresponding to the i-th moment.

[0045] An electronic device includes a processor and a storage device, wherein the storage device is suitable for storing a plurality of program codes, and the program codes are suitable for being loaded and run by the processor to execute the above-mentioned heat load and water supply temperature prediction method based on indoor temperature analysis.

[0046] A computer-readable storage medium stores a plurality of program codes, wherein the program codes are suitable for being loaded and run by a processor to execute the above-mentioned heat load and water supply temperature prediction method based on indoor temperature analysis.

[0047] The advantages of the present invention are:

[0048] The heat load and water supply temperature prediction method and device based on indoor temperature analysis provided by the present invention can reasonably predict the heat load and water supply temperature, and realize accurate energy supply under the premise of ensuring comfort.

[0049] Furthermore, the present invention analyzes the characteristics of indoor temperature, optimizes the indoor average temperature sequence of heat users and then optimizes the heating load prediction model, cascades the heating load prediction model with the secondary network water supply temperature prediction model, and reasonably predicts the instantaneous heating load of the building and the secondary network water supply temperature, thereby reducing energy consumption while ensuring comfort. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a main flow diagram of a heat load and water supply temperature prediction method based on indoor temperature analysis of the present invention;

[0051] Figure 2 It is a flow chart of an optimized heating load prediction model of the present invention;

[0052] Figure 3 It is a flow chart of the present invention for calculating the maximum overlapping interval of the average value corresponding to each daily indoor average temperature fluctuation category. DETAILED DESCRIPTION

[0053] See attached Figure 1 , Figure 1 The main process of the heat load and water supply temperature prediction method based on indoor temperature analysis of the present invention. The heat load and water supply temperature prediction method based on indoor temperature analysis provided by the present invention includes:

[0054] Step S1: Obtain the historical operation data of the building and construct a historical operation data set.

[0055] Specifically, the building contains multiple rooms, and the historical operation data includes the water supply temperature of the second network, the return water temperature of the second network, the instantaneous load of the heating, the instantaneous flow of the second network, the outdoor comprehensive meteorological parameters, the room temperature parameters of the heat users, and the time parameters. Among them, the outdoor comprehensive meteorological parameters include parameters such as temperature, humidity, wind speed, and solar radiation. The room temperature parameters of the heat users include parameters such as building number, household number, and indoor temperature. The time parameter can be a timestamp. After obtaining the historical operation data, the timestamps of these historical operation data are aligned, the missing, repeated, and abnormal data are cleaned, and the data is normalized, and finally these historical operation data are constructed into a historical operation data set.

[0056] Step S2: Construct the average indoor temperature series of thermal users.

[0057] Specifically, according to the historical operation data set, the average indoor temperature of all heat users at each moment is calculated, and the time series and the average indoor temperatures of all heat users corresponding to it form a heat user indoor average temperature sequence.

[0058] The average indoor temperature of all heat users can be calculated according to formula (1):

[0059]

[0060] is the average indoor temperature of all heat users corresponding to the i-th moment; t (a,b ) j is the j-th indoor temperature value corresponding to the i-th moment, where a is the building number, b is the household number of the building; and n is the total number of households in the building corresponding to the i-th moment.

[0061] Step S3: Construct a heating load prediction model.

[0062] Specifically, the instantaneous flow of the second network, the water supply temperature of the second network, the water supply temperature of the second network, the outdoor comprehensive meteorological parameters, and the average indoor temperature sequence of the heat users in the data set are used as the input of the heating load prediction model, and the heating instantaneous load is used as the output of the heating load prediction model, and a machine learning algorithm (such as BI-LSTM bidirectional long short-term memory network) is used to build a heating load prediction model. In addition, the statistical indicators of mean square error (MSE) or mean absolute error (MAE) or root mean square error (RMSE) are used to verify the accuracy of the heating load prediction model, so that the model meets the accuracy requirements.

[0063] Step S4: Construct a secondary network water supply temperature prediction model.

[0064] Specifically, the instantaneous heating load, outdoor comprehensive meteorological parameters, and the return water temperature of the second network in the data set are used as the input of the second network water supply temperature prediction model, and the second network water supply temperature is used as the output of the second network water supply temperature prediction model, and a machine learning algorithm (such as a BI-LSTM bidirectional long short-term memory network) is used to construct the second network water supply temperature prediction model. The statistical indicators of mean square error (MSE) or mean absolute error (MAE) or root mean square error (RMSE) are used to verify the accuracy of the second network water supply temperature prediction model.

[0065] Step S5: The instantaneous heating load of the building is predicted based on the heating load prediction model, and the secondary network water supply temperature prediction model predicts the secondary network water supply temperature of the building based on the output of the heating load prediction model.

[0066] Specifically, the instantaneous heating load of the building is predicted based on the instantaneous flow of the second network, the water supply temperature of the second network, the water supply temperature of the second network, the comprehensive outdoor meteorological parameters, and the average indoor temperature sequence of the heat users on that day as the input of the heating load prediction model. The prediction results, as well as the current instantaneous heating load, comprehensive outdoor meteorological parameters, and the return water temperature of the second network are used as the input of the second network water supply temperature prediction model to predict the second network water supply temperature of the building.

[0067] See attached Figure 2 , the embodiment of the present invention further includes the step of optimizing the heating load prediction model:

[0068] Step SA1: According to the average indoor temperature sequence of heat users, the median of the average indoor temperatures of all heat users per day is obtained and used as the overall feature corresponding to that day.

[0069] Specifically, the overall feature is the temperature corresponding to the middle position (50%) of the indoor average temperature sequence of heat users after it is divided by day and the indoor average temperature is sorted by size. For example, according to the frequency of data collection, there are multiple average indoor temperatures per day. For example, if the data is collected once per hour, there will be the average indoor temperature of all 24 heat users. The median is the corresponding temperature value (the average indoor temperature of all heat users) in the middle position (the 12th) after the average indoor temperature of all 24 heat users is sorted, which represents the typical temperature and is used as the overall feature of the day.

[0070] Step SA2: According to the indoor average temperature sequence of heat users, the difference between the average values ​​of the indoor temperatures of all heat users at adjacent time differences within each day is obtained, and used as the corresponding change feature of the day.

[0071] Taking the kth day as an example, one data is collected every hour, and one day is the average value of the indoor temperature of all 24 heat users. Then the difference expression of the average value of the indoor temperature of all heat users with adjacent time differences is: T k，0 -Tk，23 , T k，1 -T k，0 , T k，2 -T k，1 .............T k，23 -T k，22 ; T k，0 is the average indoor temperature of all heat users at time 0 on the kth day, T k，23 is the average indoor temperature of all heat users at 23:00 on the kth day, T k，1 is the average indoor temperature of all heat users at 1 o'clock on the kth day, T k，2 is the average indoor temperature of all heat users at 2 on the kth day, T k，22 is the average indoor temperature of all heat users at 22:00 on the kth day.

[0072] Step SA3: Cluster the multiple overall characteristics and change characteristics to obtain multiple daily indoor average temperature fluctuation categories.

[0073] For example, MD(T k ), (T k，0 -T k，23 T k，1 -T k，0 T k，2 -T k，1 .............T k，23 -T k，22 ), where MD(T k ) is the overall characteristic of the kth day; (T k，0 -T k1，23 T k，1 -T k，0 T k，2 -T k，1 .............T k，23 -T k，22 ) is the change feature of the kth day. By date range, multiple feature data sets of overall features and change features can be obtained.

[0074] Cluster analysis is performed on the feature data set to obtain the daily average indoor temperature fluctuation category. The daily average indoor temperature fluctuation category generated by cluster analysis generally generates more than 3 categories. Each category can display the number of days belonging to the category and the temperature curve (24 hours) of each day in the number of days belonging to the category. The data clustering algorithm can be K-means, DBSCAN, FCM, spectral clustering, etc.

[0075] Step SA4: Count the maximum overlapping intervals of the average values ​​corresponding to each daily indoor average temperature fluctuation category.

[0076] Specifically, after the indoor average temperature series of the heat user is divided by day, multiple subsequences can be obtained. Each subsequence is classified by the above clustering. Assume that the daily indoor average temperature fluctuation category contains N subsequences, and a total of 3 categories are generated by clustering. Figure 3 The step of “calculating the maximum overlapping interval of the average values ​​corresponding to each daily indoor average temperature fluctuation category” includes:

[0077] Step SA41: sort the average values ​​corresponding to each hour in the N subsequences from small to large and divide them into four equal parts.

[0078] Step SA42: Use the average values ​​corresponding to the 75% and 25% positions as the upper and lower boundaries corresponding to the hour, respectively.

[0079] Step SA43: The upper and lower boundaries of each hour corresponding to each daily indoor average temperature fluctuation category are combined into a data set.

[0080] Step SA44: Count the maximum overlapping intervals of the average values ​​corresponding to each daily indoor average temperature fluctuation category based on the data set.

[0081] Specifically, each category contains subsequences with similar changes over N days; sort the subsequences with similar changes over N days, and the average indoor temperatures of all heat users in each hour (0-23) from small to large, and then divide them into four equal parts, and use the temperatures corresponding to the 75% and 25% positions as the upper and lower boundaries to obtain the upper and lower boundaries of 0-23 hours.

[0082] Calculation method example:

[0083] Suppose the average values ​​of the indoor temperatures of all heat users in a subsequence are sorted as follows: 12, 15, 17, 19, 20, 23, 25, 28, 30, 33, 34, 35, 36, 37.

[0084] but:

[0085] Q1 represents the location of the lower quartile = (14+1) / 4 = 3.75

[0086] Q2 represents the location of the median = 2*(14+1) / 4 = 7.5

[0087] Q3 represents the location of the upper quartile = 3*(14+1) / 4 = 11.25

[0088] The 3.75th, 7.5th, and 11.25th items in the variable are the lower quartile, median, and upper quartile, respectively. That is:

[0089] Q1=0.25*third term+0.75*fourth term=0.25*17+0.75*19=18.5

[0090] Q2 = 0.5 * seventh term + 0.5 * eighth term = 0.5 * 25 + 0.5 * 28 = 26.5

[0091] Q3 = 0.75 * 11th item + 0.25 * 12th item = 0.75 * 34 + 0.25 * 35 = 34.25

[0092] Through the above method, we can get the temperature corresponding to the upper quartile and the temperature corresponding to the lower quartile of each hour (0-23) of the N subsequences in each category:

[0093] Category 1: (T1 0sy ,T1 0xy )(T1 1sy ,T1 1xy ).............(T1 23sy ,T1 23xy );

[0094] Category 2: (T2 0sy ,T2 0xy )(T2 1sy ,T2 1xy ).............(T2 23sy ,T2 23xy );

[0095] Category 3: (T3 0sy ,T3 0xy )(T3 1sy ,T3 1xy ).............(T3 23sy ,T3 23xy );

[0096] Find the maximum overlap interval T of the upper and lower quartile intervals (upper and lower boundaries) for each hour (0-23 hours) in the three categories r, ,Right now:

[0097] Category 1: (T1 max ,T1 min );

[0098] Category 2: (T2 max ,T2 min );

[0099] Category 3: (T3 max ,T3 min )

[0100] Example of the maximum overlap interval algorithm for each category

[0101] 1. Operation steps:

[0102] (1) Classify the upper and lower boundaries of all moments into 1:

[0103] (T1 0sy ,T1 0xy )(T1 1sy ,T1 1xy ).............(T1 23sy ,T1 23xy )

[0104] Sort by starting value from smallest to largest.

[0105] (2) Use a counter to record how many intervals overlap currently:

[0106] Event points (interval start and end points) traverse all intervals:

[0107] When the starting point of an interval is within the current interval, the counter count+1;

[0108] When the end point of an interval is within the current interval, the counter count-1.

[0109] (3) The boundary of the maximum overlap interval is:

[0110] Start point: The earliest starting point when the counter reaches its maximum value.

[0111] End point: The latest end point at which the counter decreases from the maximum value.

[0112] Specific calculation example:

[0113] Time 1: interval [20,22]

[0114] Time 2: interval [21,23]

[0115] Time 3: interval [19,21]

[0116] Time 4: interval [22,24]

[0117] (1) Arrange the interval endpoints: Arrange the upper and lower limits (starting points and ending points) of all intervals into event points: {20, 22, 21, 23, 19, 21, 22, 24}.

[0118] (2) Sort by starting point: Sort the intervals from the smallest to the largest starting point, and mark each event as the "starting point" or "end point":

[0119] 19 (starting point, +1)

[0120] 20 (starting point, +1)

[0121] 21 (starting point, +1)

[0122] 21 (end point, -1)

[0123] 22 (starting point, +1)

[0124] 22 (end point, -1)

[0125] 23 (end point, -1)

[0126] 24 (end point, -1)

[0127] (3) Dynamic counting: Use a counter to record how many intervals overlap at present:

[0128] Event points (interval start and end points) traverse all intervals:

[0129] When the starting point of an interval is within the current interval, the counter count+1;

[0130] When the end point of an interval is within the current interval, the counter count-1.

[0131] Counting process:

[0132] Initial count = 0

[0133] Event point 19 starts, count+1=1 (occurs once at [19-21]);

[0134] Event point 20 starts, count+2=3 (20 is in both [19,21] and [20,22], so it is 2);

[0135] Event point 21 starts, count+3=6 (21 appears 3 times in [19,21], [20,22], [21,23], 3 is the maximum overlap);

[0136] Event point 21 ends, count-3=3 (21 appears 3 times in [19,21], [20,22], [21,23]);

[0137] Event point 22 starts, count+3=6 (22 appears 3 times in [20,22], [21,23], [22,24], and the counter is maximum again);

[0138] Event point 22 ends, count-3=3 (22 appears 3 times in [20,22], [21,23], [22,24]);

[0139] Event point 23 ends, count-2=1 (23 is in [21,23], [22,24], which is 2);

[0140] Event point 24 ends, count-1=-0 (24 is 1 in [22,24]).

[0141] Find the maximum overlapping interval:

[0142] At the moment when the counter value is maximum (=6), find the start and end of the interval:

[0143] Starting point: 21 (the moment when the counter changes from 3 to 6)

[0144] End point: 22 (the moment when the counter changes from 6 to 3)

[0145] result

[0146] The maximum overlap interval is [21,22]

[0147] Based on this, the maximum overlapping interval corresponding to each daily indoor average temperature fluctuation category is obtained;

[0148] Step SA5: Determine the upper and lower limits of the daily average indoor temperature based on the maximum overlapping interval corresponding to each daily average indoor temperature fluctuation category.

[0149] Specifically, the upper and lower limits of the daily average indoor temperature are determined according to formula (2):

[0150]

[0151] in, The upper or lower limit of the daily average indoor temperature, DAY r is the number of days in the rth daily average indoor temperature fluctuation category, DAY sum is the total number of days, t r is the upper or lower limit of the maximum overlap interval corresponding to the rth daily indoor average temperature fluctuation category after clustering. When t is the upper limit of daily indoor average temperature, r is the upper limit of the maximum overlap interval corresponding to the rth daily indoor average temperature fluctuation category after clustering. When t is the lower limit of daily average indoor temperature, r It is the lower limit of the maximum overlapping interval corresponding to the rth daily indoor average temperature fluctuation category after clustering.

[0152] Step SA6: Divide the heat user indoor average temperature sequence into multiple subsequences by day, and use the daily indoor average temperature upper and lower limits to replace the temperature data in the subsequence that exceeds the daily indoor average temperature upper and lower limits, so as to optimize the heat user indoor average temperature sequence.

[0153] Specifically, in the indoor average temperature sequence of the heat users divided by day, the upper limit of the daily indoor average temperature is used to replace the temperature data in the subsequence that is greater than the upper limit of the daily indoor average temperature;

[0154] Use the lower limit of the daily average indoor temperature to replace the temperature data in the subsequence that is less than the lower limit of the daily average indoor temperature.

[0155] Step SA7: Update the heating load prediction model using the optimized average indoor temperature sequence of heat users. That is, retrain the heating load prediction model using the optimized average indoor temperature sequence of heat users.

[0156] Step SA8: Use the output of the updated heating load prediction model to predict the building's secondary network water supply temperature. That is, use the output of the trained heating load prediction model as the input of the secondary network water supply temperature prediction model to predict the secondary network water supply temperature.

[0157] In summary, the heat load and water supply temperature prediction method and device based on indoor temperature analysis provided by the present invention can reasonably predict the heat load and water supply temperature, and realize accurate energy supply while ensuring comfort.

[0158] Furthermore, the present invention analyzes the characteristics of indoor temperature, optimizes the indoor average temperature sequence of heat users and then optimizes the heating load prediction model, cascades the heating load prediction model with the secondary network water supply temperature prediction model, and reasonably predicts the instantaneous heating load of the building and the secondary network water supply temperature, thereby reducing energy consumption while ensuring comfort.

[0159] It is understood by those skilled in the art that the present invention implements all or part of the processes in the method of the above embodiment, and can also be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable storage medium may include: any entity or device, medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory, random access memory, electric carrier signal, telecommunication signal and software distribution medium that can carry the computer program code. It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable storage media do not include electric carrier signals and telecommunication signals.

[0160] Furthermore, the present invention also provides an electronic device. In an electronic device embodiment according to the present invention, the electronic device includes a processor and a storage device, the storage device can be configured to store a program for executing the heat load and water supply temperature prediction method based on indoor temperature analysis of the above method embodiment, and the processor can be configured to execute the program in the storage device, which includes but is not limited to the program for executing the heat load and water supply temperature prediction method based on indoor temperature analysis of the above method embodiment. For ease of explanation, only the parts related to the embodiment of the present invention are shown. For specific technical details not disclosed, please refer to the method part of the embodiment of the present invention.

[0161] Furthermore, the present invention also provides a computer-readable storage medium. In a computer-readable storage medium embodiment according to the present invention, the computer-readable storage medium can be configured to store a program for executing the heat load and water supply temperature prediction method based on indoor temperature analysis of the above-mentioned method embodiment, and the program can be loaded and run by the processor to implement the above-mentioned heat load and water supply temperature prediction method based on indoor temperature analysis. For ease of explanation, only the parts related to the embodiment of the present invention are shown. For specific technical details not disclosed, please refer to the method part of the embodiment of the present invention. The computer-readable storage medium can be a storage device formed by various electronic devices. Optionally, the computer-readable storage medium in the embodiment of the present invention is a non-temporary computer-readable storage medium.

[0162] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims

1. A method for predicting heat load and water supply temperature based on indoor temperature analysis, characterized in that: The method comprises: Acquire the historical operation data of the building and construct a historical operation data set, wherein the historical operation data includes the secondary network water supply temperature, the secondary network return water temperature, the heating instantaneous load, the secondary network instantaneous flow, the outdoor comprehensive meteorological parameters, the heat user room temperature parameters and the time parameters; Constructing a heat user indoor average temperature sequence, comprising: calculating the average indoor temperature of all heat users at each moment according to the historical operation data set, and forming the heat user indoor average temperature sequence with the time series and the corresponding average indoor temperatures of all heat users; Constructing a heating load prediction model, including: taking the instantaneous flow of the second network, the water supply temperature of the second network, the water supply temperature of the second network, the outdoor comprehensive meteorological parameters, and the average indoor temperature sequence of the heat user as the input of the heating load prediction model, taking the instantaneous heating load as the output of the heat load prediction model, and using a machine learning algorithm to construct the heating load prediction model; Constructing a second network water supply temperature prediction model, including: taking the heating instantaneous load, outdoor comprehensive meteorological parameters, and the second network return water temperature as inputs of the second network water supply temperature prediction model, taking the second network water supply temperature as output of the second network water supply temperature prediction model, and using a machine learning algorithm to construct the second network water supply temperature prediction model; The instantaneous heating load of the building is predicted based on the heating load prediction model, and the secondary network water supply temperature prediction model predicts the secondary network water supply temperature of the building based on the output of the heating load prediction model.

2. The method for predicting heat load and water supply temperature based on indoor temperature analysis according to claim 1, characterized in that: According to the indoor average temperature sequence of the heat users, the median of the average indoor temperatures of all the heat users on a daily basis is obtained and used as the overall feature corresponding to the day; According to the indoor average temperature sequence of the heat users, the difference between the average values ​​of the indoor temperatures of all the heat users at adjacent time differences within each day is obtained, and used as the change feature corresponding to the day; Clustering the plurality of overall characteristics and change characteristics to obtain a plurality of daily indoor average temperature fluctuation categories; Counting the maximum overlapping interval of the average values ​​corresponding to each daily indoor average temperature fluctuation category; Determine the upper limit and lower limit of the daily indoor average temperature based on the maximum overlapping interval corresponding to each of the daily indoor average temperature fluctuation categories; Divide the heat user indoor average temperature sequence into multiple subsequences by day, and use the daily indoor average temperature upper limit and lower limit to replace the temperature data in the subsequences that exceed the daily indoor average temperature upper limit and lower limit, so as to optimize the heat user indoor average temperature sequence; Updating the heating load prediction model using the optimized indoor average temperature sequence of the heat users; Accordingly, the output of the updated heating load prediction model is used to predict the secondary network water supply temperature of the building.

3. The method for predicting heat load and water supply temperature based on indoor temperature analysis according to claim 2, characterized in that: The step of "using the upper and lower limits of the daily indoor average temperature to replace the temperature data in the subsequence that exceeds the upper and lower limits of the daily indoor average temperature" includes: In the heat user indoor average temperature sequence divided by day, the temperature data in the subsequence that is greater than the daily indoor average temperature upper limit is replaced by the daily indoor average temperature upper limit; The temperature data in the subsequence that is less than the lower limit of the daily indoor average temperature is replaced by the lower limit of the daily indoor average temperature.

4. The method for predicting heat load and water supply temperature based on indoor temperature analysis according to claim 3, characterized in that: The daily indoor average temperature fluctuation category includes N subsequences; The step of "counting the maximum overlapping interval of the average values ​​corresponding to each of the daily indoor average temperature fluctuation categories" includes: Sort the average values ​​corresponding to each hour in the N subsequences from small to large and divide them into four equal parts; The average values ​​corresponding to the 75% and 25% positions are respectively used as the upper and lower boundaries corresponding to the hour; A data set consisting of the upper and lower boundaries of each hour corresponding to each of the daily indoor average temperature fluctuation categories; The maximum overlapping interval of the average values ​​corresponding to each daily average indoor temperature fluctuation category is counted based on the data set.

5. The method for predicting heat load and water supply temperature based on indoor temperature analysis according to claim 4, characterized in that: The step of "determining the upper limit and lower limit of the daily indoor average temperature based on the maximum overlapping interval corresponding to each of the daily indoor average temperature fluctuation categories" includes: Determine the upper and lower limits of the daily average indoor temperature by the following formula: Among them, t is the upper or lower limit of the daily average indoor temperature, DAY r is the number of days in the rth daily average indoor temperature fluctuation category, DAY sum is the total number of days, t r It is the upper or lower limit of the maximum overlapping interval corresponding to the r-th daily indoor average temperature fluctuation category after clustering.

6. The method for predicting heat load and water supply temperature based on indoor temperature analysis according to claim 1, characterized in that: After the step of "building a heating load prediction model", the method further includes: Using statistical indicators such as mean square error, mean absolute error or root mean square error to verify the accuracy of the heating load prediction model; After the step of "building a second network water supply temperature prediction model", the method further includes: The accuracy of the secondary network water supply temperature prediction model is verified by using statistical indicators such as mean square error, mean absolute error or root mean square error.

7. The method for predicting heat load and water supply temperature based on indoor temperature analysis according to claim 1, characterized in that: The steps of "Building a Historical Run Dataset" include: The historical operation data is timestamped, and missing, duplicated and abnormal data are cleaned and normalized.

8. The method for predicting heat load and water supply temperature based on indoor temperature analysis according to claim 1, characterized in that: The average indoor temperature of all heat users is calculated as follows: in, is the average indoor temperature of all heat users corresponding to the i-th moment; t (a,b ) j is the j-th indoor temperature value corresponding to the i-th moment, where a is the building number, b is the household number of the building; and n is the total number of households in the building corresponding to the i-th moment.

9. An electronic device comprising a processor and a storage device, wherein the storage device is suitable for storing a plurality of program codes, characterized in that: The program code is suitable for being loaded and run by the processor to execute the heat load and water supply temperature prediction method based on indoor temperature analysis according to any one of claims 1 to 8.

10. A computer-readable storage medium storing a plurality of program codes, characterized in that: The program code is suitable for being loaded and run by a processor to execute the heat load and water supply temperature prediction method based on indoor temperature analysis according to any one of claims 1 to 8.

Images