A heat load prediction method and system considering heat utilization characteristics in alpine regions

Abstract

The application provides a heat load prediction method and system considering heat characteristics in high-cold regions, and belongs to the technical field of central heating. Specifically, the method comprises the following steps: extracting weather, building type, time and corresponding heat load data that affect the heat supply load; performing weighted processing to obtain weighted building types and weighted weather; performing correlation analysis based on a comprehensive correlation coefficient method to obtain a correlation coefficient, and performing further weighted processing to obtain a predicted building type, a predicted weather and a predicted time; inputting the predicted building type, the predicted weather, the predicted time and the corresponding heat load data as a training set into a prediction model to obtain a trained prediction model; based on the trained prediction model, inputting the current predicted building type, the predicted weather and the predicted time into the trained prediction model to obtain a heat load prediction result; and improving the heat load prediction factors, so that the prediction accuracy is further improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of central heating, and particularly relates to a heat load prediction method and system considering heat consumption characteristics in high-cold regions. BACKGROUND

[0002] With the rapid development of the national economy of China, people's requirements for the comfort of living environment are getting higher and higher. Central heating is an important basic undertaking in China, and is an important means to ensure the comfort of living environment in northern China in winter. At present, the energy and environmental protection situation in China is severe, and the extensive central heating mode does not meet the requirements of green development. Relying on advanced technical means and control strategies to ensure the efficient and energy-saving operation of the central heating system is the development trend of the building central heating system. With the development of metering and monitoring technology, network control technology and information processing technology, relying on these advanced technologies, researching energy-saving control strategies of the building central heating system and improving the energy efficiency of the central heating system have become an important content of research and attention in the relevant field.

[0003] The important degrees of the heating influencing factors in high-cold regions and the heating influencing factors in ordinary regions are not the same. If the traditional load prediction method is applied to the high-cold regions, not only the error is large, but also the load plan cannot be well arranged based on the prediction result. Therefore, how to combine the actual situation of the high-cold regions and transform the heat load prediction factors is an urgent technical problem to be solved.

[0004] In view of the above technical problems, the application provides a heat load prediction method and system considering heat consumption characteristics in high-cold regions. SUMMARY

[0005] To achieve the object of the application, the application adopts the following technical scheme:

[0006] According to one aspect of the application, a heat load prediction method considering heat consumption characteristics in high-cold regions is provided.

[0007] A heat load prediction method considering heat consumption characteristics in high-cold regions, characterized in that it specifically comprises:

[0008] S1 extracts weather, building type, time and corresponding heat load data influencing the heat load;

[0009] S2: the building type and the weather are weighted to obtain a weighted building type and a weighted weather;

[0010] S3, based on the comprehensive correlation coefficient method of the Pearson product-moment correlation coefficient method and the Spearman rank correlation coefficient method, performs correlation analysis between the weighted building type, the weighted weather, the time and the heat load data, and obtains a weighted building type correlation coefficient, a weighted weather correlation coefficient and a time correlation coefficient according to an analysis result;

[0011] S4, based on the weighted building type correlation coefficient, the weighted weather correlation coefficient and the time correlation coefficient, performs further weighting processing on the weighted building type, the weighted weather and the time, and obtains a predicted building type, a predicted weather and a predicted time;

[0012] S5 inputs the predicted building type, the predicted weather, the predicted time and corresponding heat load data as a training set into a prediction model based on the GRU algorithm and the LIBSVM algorithm, and obtains a trained prediction model;

[0013] S6, based on the trained prediction model, performs weighting processing on a current building type, weather and time, obtains a current predicted building type, a predicted weather and a predicted time, and inputs the current predicted building type, the predicted weather and the predicted time into the trained prediction model, and obtains a heat load prediction result.

[0014] By first performing weighting processing on the building type and the weather to obtain a weighted building type and a weighted weather, the influence degree of the building type and the weather on the final load prediction can be deepened, which is also combined with the actual situation of the high-cold region. The influence degree of the building type and the weather in the high-cold region is deeper. By obtaining the weighted building type correlation coefficient, the weighted weather correlation coefficient and the time correlation coefficient through the comprehensive correlation coefficient method, the actual situation of the high-cold region can be further combined to further improve the accuracy of the load prediction in the high-cold region. By inputting the predicted building type, the predicted weather, the predicted time and corresponding heat load data as a training set into a prediction model based on the GRU algorithm and the LIBSVM algorithm, a trained prediction model is obtained, and the final heat load result is obtained based on the prediction model, so that the original actual situation of the high-cold region is not combined, and the heat load prediction factors are not reformed, thereby further improving the accuracy of the load prediction in the high-cold region.

[0015] The building type and the weather are weighted first to obtain weighted building types and weighted weather, which is combined with the actual situation of the alpine region, so that the influence degree of the building type and the weather in the final load prediction can be deepened, the comprehensive correlation coefficient method based on the Pearson product difference correlation coefficient method and the Spearman rank correlation coefficient method is adopted, so that the correlation analysis result obtained is more accurate, and according to the correlation analysis result, the final weighted building type correlation coefficient, weighted weather correlation coefficient and time correlation coefficient are obtained, the correlation coefficient of the larger correlation is greater than 1, and the correlation coefficient of the smaller correlation is less than 1, the weighted building type, weighted weather and time are further weighted based on the weighted building type correlation coefficient, the weighted weather correlation coefficient and the time correlation coefficient, to obtain a predicted building type, a predicted weather and a predicted time, so as to further realize the modification of the heat load prediction factor, further improve the overall prediction accuracy, and the prediction model based on the GRU algorithm and the LIBSVM algorithm is adopted, which integrates the advantages of the GRU algorithm in processing time series data and the strong generalization ability of the LIBSVM algorithm, so as to further improve the prediction accuracy.

[0016] The further technical scheme is that the weather includes current temperature, maximum temperature of the day, minimum temperature of the day, snowfall, humidity, average temperature of the day, the time includes holiday type, month value, day value and hour value, and the building type is divided into office building influence factor, school influence factor, factory influence factor and residential influence factor.

[0017] The further technical scheme is that the office building influence factor, the school influence factor, the factory influence factor and the residential influence factor are determined by expert scoring.

[0018] The further technical scheme is that before the building type and the weather are weighted, the weather and the time need to be dimensionally reduced by the PCA principal component analysis method.

[0019] The PCA principal component analysis method is adopted for dimension reduction, so that the input quantity is further reduced, and the overall prediction efficiency is further improved.

[0020] The further technical scheme is that the calculation formula of the weighted building type and the weighted weather is:

[0021] T j =t j T jc

[0022] T t =t t Ttc

[0023] wherein T j , T t are weighted building type, weighted weather, t j , t t are weight value of building type, weight value of weather, T jc , T tc are building type, weather.

[0024] Further technical solutions are that the calculation formula of the predicted building type, the predicted weather and the predicted time is:

[0025] T jf = t jf T j

[0026] T tf = t tf T t

[0027] T sf = t s T s

[0028] wherein T jf , T tf , T sf are predicted building type, predicted weather, predicted time, t jf , t tf , t s are weight value of weighted building type, weight value of weighted weather, weight value of time, T jc , T tc , T s are weighted building type, weighted weather, time.

[0029] By using the correlation analysis result, the predicted building type, the predicted weather and the predicted time are further obtained, so that the heat load influencing factors are further modified, and the overall prediction accuracy becomes more accurate.

[0030] Further technical solutions are that the specific steps of the prediction model based on the GRU algorithm and the LIBSVM algorithm are:

[0031] S11 inputs the current predicted building type, the predicted weather and the predicted time into the GRU algorithm-based prediction model to obtain a GRU prediction result;

[0032] S12 inputs the current predicted building type, the predicted weather and the predicted time into the prediction model based on the LIBSVM algorithm to obtain a LIBSVM prediction result;

[0033] S13 obtains a thermal load prediction result based on the GRU prediction result and the LIBSVM prediction result.

[0034] By adopting the prediction model based on the GRU algorithm and the LIBSVM algorithm, not only the advantages of the original GRU algorithm in processing time series data are combined, but also the advantage of the LIBSVM algorithm that the generalization ability is strong, so that the overall prediction result becomes more accurate.

[0035] Further technical solutions are that a calculation formula of the thermal load prediction result is:

[0036] P = X1P g + X2P s + t

[0037] Wherein P is the thermal load prediction result, P g , P s respectively are the GRU prediction result and the LIBSVM prediction result, X1 and X2 respectively are the GRU prediction result weight and the LIBSVM prediction result weight, and t is a penalty term.

[0038] By setting the penalty term, the penalty term is set according to the mean square error between the thermal load data of the training set and the prediction result of the prediction model based on the GRU algorithm and the LIBSVM algorithm, so as to further improve the overall prediction accuracy, so that the prediction result becomes smoother.

[0039] In another aspect, the present application provides a heat load prediction system considering heat consumption characteristics in alpine regions, which adopts the heat load prediction method considering heat consumption characteristics in alpine regions described above, and comprises a data acquisition module, a data processing module, a model training module, and a result output module; the data acquisition module is responsible for extracting weather, building type, time, and corresponding heat load data affecting the heat load; the data processing module is responsible for weighted processing of the building type and the weather to obtain weighted building type and weighted weather, performing correlation analysis between the weighted building type, the weighted weather, the time, and the heat load data based on a comprehensive correlation coefficient method of Pearson sum-difference correlation coefficient method and Spearman rank correlation coefficient method, obtaining a weighted building type correlation coefficient, a weighted weather correlation coefficient, and a time correlation coefficient according to the analysis result, and performing further weighted processing on the weighted building type, the weighted weather, and the time based on the weighted building type correlation coefficient, the weighted weather correlation coefficient, and the time correlation coefficient to obtain a predicted building type, a predicted weather, and a predicted time; the model training module is responsible for inputting the predicted building type, the predicted weather, the predicted time, and the corresponding heat load data into a prediction model based on a GRU algorithm and a LIBSVM algorithm as a training set to obtain a trained prediction model; and the result output module is responsible for obtaining a current predicted building type, a current predicted weather, and a current predicted time by performing weighted processing on a current building type, a current weather, and a current time based on the trained prediction model, and inputting the current predicted building type, the current predicted weather, and the current predicted time into the trained prediction model to obtain a heat load prediction result. BRIEF DESCRIPTION OF DRAWINGS

[0040] The above and other features and advantages of the present application will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings in which:

[0041] Figure 1 is a flowchart of a heat load prediction method considering heat consumption characteristics in alpine regions according to embodiment 1.

[0042] Figure 2 is a flowchart of specific steps of a prediction model based on a GRU algorithm and a LIBSVM algorithm according to embodiment 1.

[0043] Figure 3 is a framework diagram of a heat load prediction system considering heat consumption characteristics in alpine regions according to embodiment 2. DETAILED DESCRIPTION

[0044] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations can be implemented in any numerous ways, and should not be construed as limited to any implementation described herein; rather, these described implementations should be considered as well as numerous other alternatives that will be apparent to those skilled in the art. Identical reference numerals have been used, where possible, to designate identical elements in the figures.

[0045] The use of the terms "one", "a", "an", "the" and "said" are used generically for the purpose of convenience and are intended to include one or more of the elements / components / etc.

[0046] With the rapid development of China's national economy, people's comfort requirements for living environment are getting higher and higher. Central heating is an important basic undertaking in China, and is an important means to ensure the comfort of people's living environment in northern China in winter. The current energy and environmental protection situation in China is severe, and the extensive central heating mode does not meet the requirements of green development. Relying on advanced technical means and control strategies to ensure the efficient and energy-saving operation of the central heating system is the development trend of building central heating systems. With the development of metering and monitoring technology, network control technology and information processing technology, relying on these advanced technologies, researching energy-saving control strategies for building central heating systems and improving the energy efficiency of central heating systems have become an important content of research and attention in related fields.

[0047] The importance of the factors affecting heating in high-cold regions is different from that in ordinary regions. If the traditional load prediction method is applied to high-cold regions, not only the error is large, but also the load plan cannot be well arranged based on the prediction results. Therefore, how to combine the actual situation of high-cold regions to transform the heat load prediction factors is an urgent technical problem to be solved.

[0048] Example 1

[0049] To solve the above problems, according to one aspect of the present application, as shown in Figure 1 A heat load prediction method considering heat consumption characteristics in high-cold regions is provided, characterized in that it specifically comprises:

[0050] S1 extracts weather, building type, time and corresponding heat load data affecting the heat load;

[0051] S2: The building type and the weather are weighted to obtain a weighted building type and a weighted weather;

[0052] S3 uses a comprehensive correlation coefficient method based on Pearson's product-moment correlation coefficient method and Spearman's rank correlation coefficient method to perform correlation analysis between the weighted building type, the weighted weather, the time and the heat load data, and obtains the weighted building type correlation coefficient, the weighted weather correlation coefficient and the time correlation coefficient based on the analysis results;

[0053] S4 further weights the weighted building type, weighted weather, and time based on the weighted building type correlation coefficient, the weighted weather correlation coefficient, and the time correlation coefficient to obtain the predicted building type, predicted weather, and predicted time.

[0054] S5 inputs the predicted building type, predicted weather, predicted time, and corresponding heat load data as a training set into the prediction model based on the GRU algorithm and the LIBSVM algorithm to obtain the prediction model after training.

[0055] S6, based on the prediction model after training, weights the current building type, weather, and time to obtain the current predicted building type, predicted weather, and predicted time, and then feeds them into the prediction model after training to obtain the heat load prediction result.

[0056] By first weighting the building type and the weather to obtain weighted building type and weighted weather, the influence of building type and weather on the final load forecast can be deepened. This also combines with the actual situation in high-altitude and cold regions, where the influence of building type and weather is more profound. By using the comprehensive correlation coefficient method to obtain the weighted building type correlation coefficient, weighted weather correlation coefficient, and time correlation coefficient, the actual situation in high-altitude and cold regions can be further combined to improve the accuracy of load forecasting in these regions. The predicted building type, predicted weather, predicted time, and corresponding heat load data are used as the training set and input into the prediction model based on the GRU algorithm and LIBSVM algorithm to obtain the trained prediction model. The final heat load result is obtained based on the prediction model, thus solving the problem that the original method did not take into account the actual situation in high-altitude and cold regions and did not modify the heat load forecasting factors, thereby further improving the accuracy of load forecasting in high-altitude and cold regions.

[0057] The building type and the weather are weighted first to obtain weighted building types and weighted weather, combined with the actual situation of the alpine region, so that the influence degree of the building type and the weather in the final load prediction can be deepened, the comprehensive correlation coefficient method based on the Pearson product difference correlation coefficient method and the Spearman rank correlation coefficient method is adopted, so that the correlation analysis result obtained is more accurate, and according to the correlation analysis result, the final weighted building type correlation coefficient, weighted weather correlation coefficient and time correlation coefficient are obtained, the correlation coefficient of the larger correlation is greater than 1, and the correlation coefficient of the smaller correlation is less than 1, the weighted building type, weighted weather and time are further weighted based on the weighted building type correlation coefficient, the weighted weather correlation coefficient and the time correlation coefficient, to obtain a predicted building type, a predicted weather and a predicted time, so as to further realize the modification of the heat load prediction factor, further improve the overall prediction accuracy, and the prediction model based on the GRU algorithm and the LIBSVM algorithm is adopted, which integrates the advantages of the GRU algorithm in processing time series data and the strong generalization ability of the LIBSVM algorithm, so as to further improve the prediction accuracy.

[0058] The further technical scheme is that the weather includes current temperature, maximum temperature of the day, minimum temperature of the day, snowfall, humidity, average temperature of the day, the time includes holiday type, month value, day value and hour value, and the building type is divided into office building influence factor, school influence factor, factory influence factor and residential influence factor.

[0059] The further technical scheme is that the office building influence factor, the school influence factor, the factory influence factor and the residential influence factor are determined by expert scoring.

[0060] The further technical scheme is that before the building type and the weather are weighted, the weather and the time need to be reduced in dimension by the PCA principal component analysis method.

[0061] The PCA principal component analysis method is adopted to reduce the dimension, so that the input quantity is further reduced, and the overall prediction efficiency is further improved.

[0062] The further technical scheme is that the calculation formula of the weighted building type and the weighted weather is:

[0063] T j =t j T jc

[0064] T t =t t Ttc

[0065] wherein T j , T t are weighted building type, weighted weather, t j , t t are weight value of building type, weight value of weather, T jc , T tc are building type, weather.

[0066] Further technical solutions are that the calculation formula of the predicted building type, the predicted weather and the predicted time is:

[0067] T jf = t jf T j

[0068] T tf = t tf T t

[0069] T sf = t s T s

[0070] wherein T jf , T tf , T sf are predicted building type, predicted weather, predicted time, t jf , t tf , t s are weight value of weighted building type, weight value of weighted weather, weight value of time, T jc , T tc , T s are weighted building type, weighted weather, time.

[0071] By using the correlation analysis result, the predicted building type, the predicted weather and the predicted time are further obtained, so that the heat load influencing factors are further modified, and the overall prediction accuracy becomes more accurate.

[0072] Further technical solutions are that the specific steps of the prediction model based on the GRU algorithm and the LIBSVM algorithm are:

[0073] S11 inputs the current predicted building type, the predicted weather and the predicted time into the GRU algorithm-based prediction model to obtain a GRU prediction result;

[0074] S12 inputs the current predicted building type, the predicted weather and the predicted time into the prediction model based on the LIBSVM algorithm to obtain a LIBSVM prediction result;

[0075] S13 obtains a thermal load prediction result based on the GRU prediction result and the LIBSVM prediction result.

[0076] By adopting the prediction model based on the GRU algorithm and the LIBSVM algorithm, not only the advantages of the original GRU algorithm in processing time series data are combined, but also the advantage of the LIBSVM algorithm that the generalization ability is strong, so that the overall prediction result becomes more accurate.

[0077] Further technical solutions are that a calculation formula of the thermal load prediction result is:

[0078] P=X1P g +X2P s +t

[0079] Wherein P is the thermal load prediction result, P g , P s respectively are the GRU prediction result and the LIBSVM prediction result, X1, X2 respectively are the GRU prediction result weight and the LIBSVM prediction result weight, and t is a penalty term.

[0080] By setting the penalty term, the penalty term is set according to the mean square error between the thermal load data of the training set and the prediction result of the prediction model based on the GRU algorithm and the LIBSVM algorithm, so as to further improve the overall prediction accuracy, so that the prediction result becomes smoother.

[0081] Embodiment 2

[0082] As Figure 3As shown, the present application provides a heat load prediction system considering heat consumption characteristics in alpine regions, which adopts the heat load prediction method considering heat consumption characteristics in alpine regions, and comprises a data acquisition module, a data processing module, a model training module and a result output module; the data acquisition module is responsible for extracting weather, building type, time and corresponding heat load data affecting the heat load; the data processing module is responsible for weighted processing of the building type and the weather to obtain weighted building type and weighted weather, performing correlation analysis between the weighted building type, the weighted weather, the time and the heat load data based on a comprehensive correlation coefficient method of Pearson sum-difference correlation coefficient method and Spearman rank correlation coefficient method, obtaining a weighted building type correlation coefficient, a weighted weather correlation coefficient and a time correlation coefficient according to the analysis result, and performing further weighted processing on the weighted building type, the weighted weather and the time based on the weighted building type correlation coefficient, the weighted weather correlation coefficient and the time correlation coefficient to obtain a predicted building type, a predicted weather and a predicted time; the model training module is responsible for inputting the predicted building type, the predicted weather, the predicted time and corresponding heat load data into a prediction model based on a GRU algorithm and a LIBSVM algorithm as a training set to obtain a trained prediction model; and the result output module is responsible for obtaining a current predicted building type, a current predicted weather and a current predicted time by performing weighted processing on a current building type, weather and time based on the trained prediction model, and inputting the current predicted building type, the current predicted weather and the current predicted time into the trained prediction model to obtain a heat load prediction result.

[0083] In the embodiments of the present application, the term "a plurality of" refers to two or more, unless otherwise explicitly limited. The terms "mounting", "connecting", "fixing" and the like should be understood in a broad sense, for example, "connecting" can be fixed connection, can also be detachable connection, or integral connection. For those skilled in the art, the specific meaning of the above terms in the embodiments of the present application can be understood according to the specific circumstances.

[0084] In the description of the embodiments of the present application, it should be understood that the positions or location relationships indicated by the terms "upper", "lower" and the like are based on the positions or location relationships shown in the drawings, and are only for the convenience of describing the embodiments of the present application and simplifying the description, and do not indicate or imply that the devices or units referred to must have a particular direction, be constructed and operated in a particular position, therefore, it cannot be understood as a limitation on the embodiments of the present application.

[0085] In the description of the specification, the description of the terms "one embodiment", "one preferred embodiment", and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are contained in at least one embodiment or example of the embodiments of the present application. In the specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0086] The above merely describes the preferred embodiments of the embodiments of the present application and is not intended to limit the embodiments of the present application. The embodiments of the present application can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the embodiments of the present application shall be included in the protection scope of the embodiments of the present application.

Claims

1. A method for predicting heat load considering the heat usage characteristics of high-altitude and cold regions, characterized in that, Specifically, it includes: S1 extracts weather, building type, time, and corresponding heat load data that affect heating load; S2; Weight the building type and the weather to obtain weighted building type and weighted weather; S3 uses a comprehensive correlation coefficient method based on Pearson's product-moment correlation coefficient method and Spearman's rank correlation coefficient method to perform correlation analysis between the weighted building type, the weighted weather, the time and the heat load data, and obtains the weighted building type correlation coefficient, the weighted weather correlation coefficient and the time correlation coefficient based on the analysis results; S4 further weights the weighted building type, weighted weather, and time based on the weighted building type correlation coefficient, the weighted weather correlation coefficient, and the time correlation coefficient to obtain the predicted building type, predicted weather, and predicted time. S5 inputs the predicted building type, predicted weather, predicted time, and corresponding heat load data as a training set into the prediction model based on the GRU algorithm and the LIBSVM algorithm to obtain the prediction model after training. S6, based on the prediction model after training, weights the current building type, weather, and time to obtain the current predicted building type, predicted weather, and predicted time, and then feeds them into the prediction model after training to obtain the heat load prediction result. The formulas for calculating the weighted building type and weighted weather are as follows: T j =t j T jc T t =t t T tc Where T j T t These are weighted building types and weighted weather, respectively. j t t These are the weights for building type and weather, respectively, T. jc T tc These are respectively: building type and weather; The formulas for predicting building type, weather, and time are as follows: T jf =t jf T j T tf =t tf T t T sf =t s T s Where T jf T tf T sf These are respectively predicting building type, weather, and time, t jf t tf ts and ts are the weighted building type correlation coefficient, the weighted weather correlation coefficient, and the time correlation coefficient, respectively. j T t T s These are weighted building type, weighted weather, and time, respectively. The specific steps of the prediction model based on the GRU algorithm and the LIBSVM algorithm are as follows: S11 inputs the current predicted building type, predicted weather, and predicted time into the GRU-based algorithm to obtain the GRU prediction result; S12 inputs the current predicted building type, predicted weather, and predicted time into the prediction model based on the LIBSVM algorithm to obtain the LIBSVM prediction result; S13 obtains the heat load prediction result based on the GRU prediction result and the LIBSVM prediction result.

2. The heat load prediction method considering the heat usage characteristics of high-altitude and cold regions as described in claim 1, characterized in that, The weather data includes the current temperature, the highest temperature of the day, the lowest temperature of the day, snowfall, humidity, and the average temperature of the day. The time data includes the type of holiday, month, day, and hour. The building types are categorized into office building impact factors, school impact factors, factory impact factors, and residential impact factors.

3. The heat load prediction method considering the heat consumption characteristics of high-altitude and cold regions as described in claim 2, characterized in that, The impact factors for office buildings, schools, factories, and residences were determined using expert scoring.

4. The heat load prediction method considering the heat usage characteristics of high-altitude and cold regions as described in claim 1, characterized in that, Before weighting the building type and the weather, the weather and time need to be reduced in dimensionality using PCA principal component analysis.

5. The heat load prediction method considering the heat usage characteristics of high-altitude and cold regions as described in claim 1, characterized in that, The formula for calculating the heat load prediction result is as follows: P=X1P g +X2P s +t Where P is the predicted heat load, P g P s X1 and X2 are the GRU prediction results and the LIBSVM prediction results, respectively, and X1 and X2 are the weights of the GRU prediction results and the LIBSVM prediction results, respectively. t is a penalty term, which is set based on the mean square error between the training set heat load data and the prediction results of the prediction models based on the GRU and LIBSVM algorithms.

6. A heat load prediction system considering the heating characteristics of high-altitude and cold regions, employing the heat load prediction method considering the heating characteristics of high-altitude and cold regions as described in any one of claims 1-5, comprising a data acquisition module, a data processing module, a model training module, and a result output module; the data acquisition module is responsible for extracting weather, building type, time, and corresponding heat load data affecting the heating load; the data processing module is responsible for weighting the building type and the weather to obtain weighted building type and weighted weather, and performing correlation analysis between the weighted building type, the weighted weather, the time, and the heat load data based on a comprehensive correlation coefficient method using the Pearson product-moment correlation coefficient method and the Spearman rank correlation coefficient method, and obtaining the weighted building type correlation coefficient and the weighted weather correlation coefficient based on the analysis results. The system uses coefficients and time correlation coefficients to further weight the weighted building type, weighted weather, and time based on the weighted building type correlation coefficient, the weighted weather correlation coefficient, and the time correlation coefficient to obtain the predicted building type, predicted weather, and predicted time. The model training module is responsible for inputting the predicted building type, predicted weather, predicted time, and corresponding heat load data as a training set into the prediction model based on the GRU algorithm and the LIBSVM algorithm to obtain the prediction model after training. The result output module is responsible for weighting the current building type, weather, and time based on the prediction model after training to obtain the current predicted building type, predicted weather, and predicted time, and then inputting them into the prediction model after training to obtain the heat load prediction result.

Images