Heat supply pipe network load prediction method, system, equipment and medium

Abstract

The invention provides a heat supply pipe network load prediction method, system, equipment and medium, and the method comprises the following steps: taking collected operation load data of a heat supply system in a set time period and a preset optimal feature set as the input of a preset TCN-Attention heat supply pipe load prediction model, and obtaining a predicted load value of the heat supply system; wherein the preset optimal feature set is obtained by screening historical operation load data of the heat supply system and data such as weather and temperature corresponding to the historical operation load data; according to the method, the optimal feature set which has the maximum influence on the load of the heat supply pipe network but has the minimum correlation with the load of the heat supply pipe network is obtained from the two factors of the internal features and the meteorological conditions of the heat supply system, so that the input of the model is reduced, and the prediction efficiency is improved; meanwhile, the TCN network is optimized by using an attention mechanism so as to improve the attention to effective features, and thus the prediction accuracy and efficiency are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of heating, and relates to a method, system, device and medium for predicting the load of a heating pipe network. Background Art

[0002] Urban central heating is a basic service industry related to people's livelihood and well-being, and is also an important part of social energy consumption. As of the end of 2023, the length of urban central heating pipelines across the country was 5.237 million kilometers, a year-on-year increase of 6.13%; the central heating area was 11.549 billion square meters, a year-on-year increase of 3.81%. The urban steam central heating capacity across the country was 123,900 tons per hour, a year-on-year decrease of 1.30%; the hot water central heating capacity was 631,200 megawatts, a year-on-year increase of 5.17%. The heating system is a complex and large system with obvious time-delay characteristics. Therefore, in order to meet the actual heat demand of residents, the system needs to adjust heating in advance. Heat load prediction becomes particularly important. Accurate heat load prediction can provide a reference for the system to adjust heating in advance. In addition, relying on manual experience to adjust heating manually easily leads to a mismatch between the heat supply and the heat demand of users, resulting in energy waste. Heat load prediction can help the district heating system make correct decisions, ensuring both the heat demand of users and not generating excessive heat, thus contributing to energy conservation and emission reduction.

[0003] Existing district heating load prediction methods mainly rely on black-box models. Black-box models use historical data to predict the heating load. Therefore, black-box models are also called data-driven models. In recent years, with the explosive development of deep learning algorithms, deep neural networks specifically designed for processing time series, such as recurrent neural networks and long short-term memory networks, have shown excellent performance in predicting the heat load of heating systems. Cui M S et al. used a bidirectional long short-term memory network to predict the heat load of a heating system; Wang Jin proposed a hybrid model composed of empirical mode decomposition, imperialist competitive optimization algorithm and support vector machine; Eseye A T et al. proposed a similar-day method to meet the heating load under local climate conditions, and used a grasshopper optimization algorithm to optimize the hyperparameters of the support vector machine, greatly improving the prediction accuracy of the model. The above prediction methods are prone to gradient disappearance and gradient explosion, and do not consider the influence of the internal characteristics of the heating system, resulting in inaccurate heat load prediction results. Summary of the Invention

[0004] In order to further improve the prediction accuracy of the heat load, the present invention proposes a method, system, device and medium for predicting the load of a heating pipe network.

[0005] To achieve the above object, the technical solution adopted by the present invention is: A method for predicting the load of a heating pipe network provided by the present invention includes the following steps: Taking the collected operation load data of the heating system within a preset time period and the preset optimal feature set as the input of the preset TCN-Attention heating pipe load prediction model, the predicted load value of the heating system is obtained; Among them, the preset optimal feature set is screened from the historical operation load data of the heating system and the weather temperature data corresponding to the historical operation load data.

[0006] Preferably, the method for screening the optimal feature set from the historical operation load data of the heating system and the weather temperature data corresponding to the historical operation load data is as follows: Collect the historical operation load data of the heating system and the historical weather temperature data corresponding to the historical operation load data; Preprocess the historical operation load data and the historical weather temperature data corresponding to the historical operation load data to obtain the processed data; Use the processed data to construct an influence factor set, which includes supply temperature, return temperature, supply pressure, return pressure, instantaneous flow rate of the pipe network, cumulative flow rate of the pipe network, outlet pressure of the circulating pump group, cumulative makeup water flow rate, makeup water tank level, air temperature, outdoor temperature, relative humidity, wind force level, and air quality index; Select the optimal feature set from the influence factor set.

[0007] Preferably, use the linear interpolation method to process the historical operation load data and the historical weather temperature data corresponding to the historical operation load data, and then use DBSCAN for clustering to obtain the processed data; Preferably, use the maximum correlation and minimum redundancy method to select the optimal feature set from the influence factor set.

[0008] Preferably, the construction method of the preset TCN-Attention heating pipe load prediction model is as follows: Introduce Attention into the hidden layer of the TCN network to obtain the TCN-Attention model; Use the historical operation load data of the heating system and the preset optimal feature set to train the TCN-Attention model to obtain the TCN-Attention heating pipe load prediction model.

[0009] A heating pipe network load prediction system includes: A data acquisition unit for collecting the operation load data of the heating system within a preset time period; A model construction unit for constructing a TCN-Attention heating pipe load prediction model; A prediction unit, which uses the collected operation load data of the heating system in a set time period and a preset optimal feature set as the input of the constructed TCN-Attention heating pipe load prediction model to obtain the predicted load value of the heating system; Among them, the preset optimal feature set is screened from the historical operation load data of the heating system and the weather temperature data corresponding to the historical operation load data.

[0010] An electronic device, including a processor and a memory, where computer instructions are stored on the memory. When the computer instructions are executed by the processor, the electronic device executes the method described above.

[0011] A computing device cluster, including at least one computing device, and each computing device includes a processor and a memory; The processor of the at least one computing device is used to execute the instructions stored in the memory of the at least one computing device, so that the computing device cluster executes the method described above.

[0012] A computer program product, which contains computer executable instructions. When the computer executable instructions are executed, the method described above is implemented.

[0013] A computer-readable storage medium stores computer executable instructions. When the computer executable instructions are executed by a processor, the method described above is implemented. Compared with the prior art, the beneficial effects of the present invention are: A heating pipe network load prediction method provided by the present invention obtains a set of features that have the greatest impact on the heating pipe network load but have the smallest correlation with each other from two types of factors, namely the internal characteristics of the heating system and meteorological conditions, as the optimal feature set, reducing the input of the model and improving the prediction efficiency; at the same time, the attention mechanism is used to optimize the TCN network to improve the attention to effective features, thereby improving the prediction accuracy and efficiency.

[0014] Furthermore, when performing data preprocessing, the linear interpolation method is used to supplement the empty data, and then DBSCAN is used to cluster the abnormal data to find and remove the outliers; in the feature selection process, mRMR is used for feature selection, which ensures high correlation while taking into account the redundancy between various influencing factors, reduces the data dimension, and improves the generalization ability of the model.

[0015] Furthermore, in the process of establishing the influencing factor set, only outdoor temperature is not selected as the influencing factor. Because when the temperature remains constant, the lower the relative humidity, the faster the sweat evaporates from the human body. The evaporation of sweat will absorb part of the heat of the human body, resulting in a feeling of coldness in the human body, and the heat supply should be appropriately increased. Strong wind will increase the heat loss of the building and the sense of coldness of the human body. When the wind force level is relatively high, the heating load is relatively large. The air quality index affects the heating load by influencing solar radiation. Therefore, the present invention establishes an influencing factor set including the internal characteristics of the heating system, outdoor temperature, relative humidity, wind force level, and air quality index.

[0016] Furthermore, in the prediction model, the Attention mechanism is introduced to strengthen the feature weights of the hidden layer in the TCN, so as to process and analyze the data input to the TCN layer, and then the prediction result is output through the output layer. By weighting different features, the key features are highlighted, effectively improving the model training speed and prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is the flowchart of the prediction process of the present invention; Figure 2 is the flowchart of TCN-Attention. DETAILED DESCRIPTION OF THE INVENTION

[0018] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0019] It should be understood that when used in the specification and appended claims of the present application, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0020] It should also be understood that the term " / and" as used in the specification and appended claims of the present application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0021] As used in the specification of this application and the appended claims, the term "if" may be construed, depending on the context, as "when", "once", "in response to determining", or "in response to detecting". Similarly, the phrases "if determined" or "if [the described condition or event] is detected" may be construed, depending on the context, to mean "once determined", "in response to determining", "once [the described condition or event] is detected", or "in response to detecting [the described condition or event]".

[0022] In addition, in the description of the specification of this application and the appended claims, the terms "first", "second", "third", etc. are used only for differentiating descriptions and cannot be construed as indicating or implying relative importance.

[0023] Reference to "one embodiment" or "some embodiments" or the like described in the specification of this application means that a specific feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of this application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "comprising", "including", "having", and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0024] Embodiment 1 As Figure 1 , Figure 2 shown, a method for predicting the load of a heat supply pipe network provided by the present invention specifically includes the following steps: Step 1, collect historical operation load data of the heat supply system and data such as weather temperature corresponding to the historical operation load data; Step 2, preprocess the collected historical data, use linear interpolation method to supplement missing data, use DBSCAN for clustering, remove outliers, and normalize the data, specifically as follows: 2.1) For a single empty data in the actual operation data of the heat supply pipe network, outdoor temperature, and relative humidity, use the linear interpolation method to fill the missing value. For the th missing data that appears, the method of filling with the linear interpolation method is as follows:

[0025] Where: , is the relative position of the missing data, is the time interval of the missing data.

[0026] For a single missing data in the wind force level and air quality index, the mode method is used to fill in the missing values.

[0027] 2.2) The steps of the DBSCAN algorithm are roughly described as follows: For a given neighborhood distance and the minimum number of samples in the neighborhood : 2.2.1) Traverse all samples to find the set of all core objects that satisfy the neighborhood distance ; 2.2.2) Arbitrarily select a core object and find all samples that are density-reachable from it to generate a clustering cluster; 2.2.3) Remove the samples that are density-reachable found in 2.2.2) from the remaining core objects; 2.2.4) Repeat steps 2.2.2)-2.2.3) from the updated set of core objects until all core objects have been traversed or removed.

[0028] 2.3) Data normalization: When the input numerical feature values are in different ranges, the data is normalized and denormalized according to the following formula.

[0029]

[0030]

[0031] Among them, represents the normalized data; is the original data; is the minimum value in the dataset; is the maximum value in the dataset; is the denormalized data.

[0032] Step 3: Establish an influence factor set including heating supply temperature, heating return temperature, supply pressure, return pressure, instantaneous pipeline network flow rate, cumulative pipeline network flow rate, outlet pressure of the circulating pump group, cumulative makeup water flow rate, makeup water tank liquid level, air temperature, outdoor temperature, relative humidity, wind force level, and air quality index; Step 4: Use the Max-Relevance and Min-Redundancy (mRMR) method to select the optimal feature set from the influence factor set. The specific calculation steps are as follows: 4.1) Solving the maximum relevance. The maximum relevance is represented by the average of the mutual information between the feature and the target variable as:

[0033] Among them, represents and the maximum correlation between; representing factors; representing the target variable; includes all ; representing the number of elements in; features and the target variable the mutual information between, and its expression is as follows:

[0034] where, representing and the joint probability density of; and respectively represent and the probability density of.

[0035] 4.2) Solving the minimum redundancy. The minimum redundancy requires that the dependence between each feature reaches the minimum, which can be expressed by the following formula:

[0036] where, represents the minimum redundancy between.

[0037] 4.3) mRMR can be expressed as:

[0038] where, represents and the difference between.

[0039] Step 5, introduce Attention to strengthen the feature weights of the hidden layer in the TCN network, so as to process and analyze the data input into the TCN network, and then output the prediction result through the output layer. By weighting different features, key features are highlighted, effectively improving the model training speed and prediction accuracy. The attention mechanism can better establish the dependence between the states of different time points of the model. Specifically: The output of the hidden layer of the TCN network can be expressed as:

[0040] where, is the hidden layer feature at the th time step, is the length of the time series.

[0041] At the output of the hidden layer of the TCN introduce the Attention mechanism to calculate the importance weights of the features at each time step: (1) Calculate the attention scores Use a fully connected layer (or other learnable mapping) to calculate the attention scores at each time step:

[0042] where and are learnable parameters, is the th attention score at the

[0043] (2) Normalize the attention weights Use the Softmax function to normalize the attention scores to obtain the attention weights:

[0044] where represents the importance weight of the feature at the th time step.

[0045] (3) Weighted summation Use the attention weights to perform weighted summation on the hidden layer features to obtain the enhanced feature representation:

[0046] where is the enhanced feature vector.

[0047] Input the enhanced feature into the subsequent layers (such as the fully connected layer or the output layer) of the TCN network for the final prediction task.

[0048] Step 6, Use the historical heat load data and the optimal feature set as the input of the prediction model to establish a TCN-Attention heat pipe load prediction model. The specific process is as follows: 6.1) Construct a TCN network and set parameters; 6.2) Add the Attention mechanism to the TCN network; 6.3) Input the training set data into the TCN-Attention model; 6.4) Use the data to train the model; 6.5) Repeat 6.4) until the set number of training times is reached; 6.6) Input the test set data into the trained TCN-Attention; 6.7) Output predicted data.

[0049] Step 7, select three main error evaluation indicators, namely mean absolute error, root mean square error, and symmetric mean absolute percentage error, to evaluate the prediction model. Adjust the model parameters according to the evaluation results to optimize the model. The calculation of each evaluation indicator is as follows:

[0050]

[0051]

[0052] Step 8, collect the operation load data of the heating system in the set time period, and use the operation load data and the optimal feature set as the input of the TCN-Attention heating pipe load prediction model to obtain the predicted load value of the heating system.

[0053] Embodiment 2 A heating pipe network load prediction system provided in this embodiment includes: A data collection unit for collecting the operation load data of the heating system in the set time period; A model construction unit for constructing a TCN-Attention heating pipe load prediction model; A prediction unit for using the operation load data of the heating system in the set time period collected and the preset optimal feature set as the input of the constructed TCN-Attention heating pipe load prediction model to obtain the predicted load value of the heating system; Among them, the preset optimal feature set is selected from the historical operation load data of the heating system and data such as the corresponding weather temperature of the historical operation load data.

[0054] Embodiment 3 This embodiment also provides a computing device. The computing device includes: a bus, a processor, a memory, and a communication interface. The processor, the memory, and the communication interface communicate through the bus. The computing device can be a server or a terminal device. It should be understood that the number of processors and memories in the computing device of the present application is not limited.

[0055] The bus can be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus can include a path for transmitting information between various components of the computing device (e.g., memory, processor, communication interface).

[0056] The processor can include any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a Tensor Processing Unit (TPU), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), a microprocessor (MP), or a digital signal processor (DSP).

[0057] The memory can include volatile memory, such as random access memory (RAM). The processor can also include non-volatile memory, such as read-only memory (ROM), flash memory, a hard disk drive (HDD), or a solid state drive (SSD).

[0058] The memory stores executable program code, and the processor executes the executable program code to respectively implement the functions of the foregoing first generation module, second generation module, and adjustment module, thereby implementing, for example, *methods, etc. That is, instructions for the methods and functions of the computing device involved in any of the foregoing embodiments can be stored on the memory.

[0059] The communication interface uses a transceiver module such as, but not limited to, a network interface card or a transceiver to implement communication between the computing device and other devices or a communication network.

[0060] Embodiment 4 This embodiment also provides a computing device cluster. The computing device cluster includes at least one computing device. The computing device can be a server, such as a central server, an edge server, or a local server in a local data center. In some embodiments, the computing device can also be a terminal device such as a desktop computer, a laptop computer, or a smart phone.

[0061] The computing device cluster includes at least one computing device. Instructions for performing the methods and functions of the computing device involved in any of the above embodiments can be stored in the memories of one or more of the computing devices in the computing device cluster.

[0062] In some possible implementation manners, partial instructions for performing the methods and functions of the computing device involved in any of the above embodiments can also be stored separately in the memories of one or more of the computing devices in the computing device cluster. In other words, a combination of one or more computing devices can jointly execute the instructions for performing the methods and functions of the computing device.

[0063] It should be noted that different computing devices in the computing device cluster can store different instructions for performing partial functions of the device respectively.

[0064] In some possible implementation manners, one or more of the computing devices in the computing device cluster can be connected through a network. Among them, the network can be a wide area network or a local area network, etc. Two computing devices are connected through the network. Specifically, they are connected to the network through the communication interfaces in each computing device.

[0065] The embodiments of the present disclosure also provide a computer program product including instructions, which when running on a computer, causes the computer to execute the methods and functions of the computing device involved in any of the above embodiments.

[0066] Embodiment 5 This embodiment also provides a computer-readable storage medium, on which computer instructions are stored. When a processor runs the instructions, the processor is caused to execute the methods and functions of the computing device involved in any of the above embodiments.

[0067] Generally, the various embodiments of the present disclosure may be implemented in hardware or specific circuits, software, logic, or any combination thereof. Some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software, which may be executed by a controller, a microprocessor, or other computing devices. Although the various aspects of the embodiments of the present disclosure are shown and described as block diagrams, flowcharts, or using some other graphical representation, it should be understood that the blocks, devices, systems, techniques, or methods described herein may be implemented as, by way of non-limiting example, hardware, software, firmware, specific circuits or logic, general hardware or a controller or other computing devices, or some combination thereof.

[0068] Embodiment 6 This embodiment provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions included in program modules, which are executed in a device on a target real or virtual processor to perform the processes / methods as referenced above with respect to the accompanying drawings. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functions of program modules may be combined or divided as needed. The machine-executable instructions for program modules may be executed within a local or distributed device. In a distributed device, program modules may be located in local and remote storage media.

[0069] The computer program code for implementing the methods of the present disclosure may be written in one or more programming languages. This computer program code may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the program code is executed by the computer or other programmable data processing device, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code may be executed entirely on the computer, partially on the computer, as a stand-alone software package, partially on the computer and partially on a remote computer, or entirely on a remote computer or server.

[0070] In the context of the present disclosure, the computer program code or related data may be carried by any suitable carrier such that a device, apparatus, or processor can perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and the like. Examples of signals may include electrical, optical, radio, acoustic, or other forms of propagated signals, such as carrier waves, infrared signals, etc.

[0071] A computer-readable medium can be any tangible medium that contains or stores a program for or related to an instruction execution system, apparatus, or device, or a data storage device such as a data center that contains one or more available media. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any suitable combination thereof. More specific examples of computer-readable storage media include electrical connections with one or more wires, portable computer disks, hard disks, random access memories (RAMs), read-only memories (ROMs), erasable programmable read-only memories (EPROMs or flash memories), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0072] The above-described embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application and should all be included within the protection scope of the present application.

Claims

1. A method for predicting heat supply network load, characterized in that: The following steps are involved: The collected operating load data of the heating system in the set time period and the preset optimal feature set are used as inputs of the preset TCN-Attention heating pipe load prediction model to obtain the predicted load value of the heating system; The preset optimal feature set is obtained by screening the historical operating load data of the heating system and the weather temperature data corresponding to the historical operating load data.

2. A method for predicting heat supply network load according to claim 1, characterized in that: The optimal feature set is obtained by screening the historical operating load data of the heating system and the weather temperature data corresponding to the historical operating load data. The specific method is: Preprocessing the historical operating load data and the historical weather temperature data corresponding to the historical operating load data to obtain processed data; The processed data is used to construct an influencing factor set, which includes supply temperature, return temperature, supply pressure, return pressure, pipe network instantaneous flow, pipe network cumulative flow, circulation pump group outlet pressure, water supply flow cumulative, water supply tank liquid level, air temperature, outdoor temperature, relative humidity, wind force level and air quality index; The optimal feature set is selected from the influencing factor set.

3. A method for predicting heat supply network load according to claim 2, characterized in that: The historical operating load data and the historical weather temperature data corresponding to the historical operating load data are processed by linear interpolation method, and then DBSCAN is used for clustering to obtain the processed data.

4. A method for predicting heat supply network load according to claim 2, characterized in that: The maximum relevance minimum redundancy method is used to select the optimal feature set from the influencing factor set.

5. A method for predicting heat supply network load according to claim 1, characterized in that: The construction method of the preset TCN-Attention heating pipe load prediction model is: Introduce Attention into the hidden layer of the TCN network to obtain the TCN-Attention model; The TCN-Attention model is trained using the historical operating load data of the heating system and the preset optimal feature set to obtain the TCN-Attention heating pipe load prediction model.

6. A heating network load prediction system, characterized in that: include: A data collection unit, used to collect the operating load data of the heating system during a set time period; A model building unit, used to build a TCN-Attention heating pipe load prediction model; A prediction unit, which is used to use the collected operating load data of the heating system in a set time period and a preset optimal feature set as inputs to the constructed TCN-Attention heating pipe load prediction model to obtain a predicted load value of the heating system; The preset optimal feature set is obtained by screening the historical operating load data of the heating system and the weather temperature data corresponding to the historical operating load data.

7. An electronic device, characterized in that: The electronic device comprises a processor and a memory, wherein the memory stores computer instructions, and when the computer instructions are executed by the processor, the electronic device executes the method according to any one of claims 1 to 5.

8. A computing device cluster, characterized in that: comprising at least one computing device, each computing device comprising a processor and a memory; The processor of the at least one computing device is configured to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster executes the method according to any one of claims 1 to 5.

9. A computer program product, characterized in that The computer program product contains computer executable instructions, which implement the method according to any one of claims 1 to 5 when executed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images