A big data-based clean heating system load prediction method

Abstract

The application relates to the technical field of data prediction, in particular to a clean heating system load prediction method based on big data, which comprises the following steps: obtaining gas supply data and various meteorological data in each of a plurality of historical days; grouping the gas supply data; analyzing the correlation between the various meteorological data and the gas supply data, and determining the influence degree of the various meteorological data on the gas supply demand; determining the theoretical gas supply data and the error of each gas supply data in each group based on the distribution of the gas supply data in all days with the same meteorological data and in combination with the influence degree; determining the confidence degree of the preset position data in each group being predicted by the remaining data and the feasibility of each group; and determining a weighted likelihood function by comprehensively considering the gas supply data in all groups and the feasibility, and constructing an AR model for predicting the gas supply demand. The application aims to improve the accuracy of the prediction of the gas supply demand.

Description

Technical Field

[0001] This application relates to the field of data prediction technology, specifically to a method for predicting the load of a clean heating system based on big data. Background Technology

[0002] Currently, cities across my country are actively promoting the transformation to clean heating. After the transformation, it is essential to rely on upstream companies to adjust the gas supply to match the demand of heat users. Insufficient supply will lead to the failure of the transformation, while oversupply poses a significant risk to gas storage. Therefore, accurate forecasting of gas supply and demand is crucial for effective gas allocation.

[0003] Currently, autoregressive models (AR models) are commonly used to predict future gas supply based on historical gas supply data. The construction of the AR model is crucial. Existing technologies generally calculate autoregressive coefficients based on historical gas supply data. However, due to the influence of factors such as natural disasters or public health emergencies, some historical gas supply data are not obtained from the combined influence of multiple weather factors. Since the core of gas supply demand changes is to cope with meteorological changes, this results in a large deviation in the autoregressive coefficients, which in turn leads to a large error in the prediction of gas supply demand. Summary of the Invention

[0004] In view of the above, it is necessary to provide a big data-based method for predicting the load of clean heating systems, which improves the accuracy of gas demand prediction compared to traditional methods.

[0005] The load forecasting method for clean heating systems based on big data proposed in this application adopts the following technical solution: One embodiment of this application provides a method for load forecasting of a clean heating system based on big data, the method comprising the following steps: Obtain daily gas supply data and various meteorological data over multiple historical days; Based on the autocorrelation of all acquired gas supply data, the order of the AR model is determined, and the gas supply data is grouped based on the order. Analyze the correlation between various meteorological data and gas supply data to determine the degree of influence of various meteorological data on gas supply demand; Based on the distribution of gas supply data for all days with the same meteorological data, and in combination with the aforementioned degree of influence, the theoretical gas supply data for each gas supply data in each group is determined; based on the difference between each gas supply data in each group and its theoretical gas supply data, the error of each gas supply data in each group is determined. Based on the error and order of the gas supply data in each group, the confidence level of the preset location data in each group can be predicted by the other data is determined; Based on the confidence level and the error in the gas supply data at the preset locations in each group, the feasibility of each group is determined. By combining the gas supply data from all groups with the aforementioned feasibility, a weighted likelihood function is determined; based on the likelihood function, an AR model for predicting gas supply demand is constructed.

[0006] In one embodiment, the method for grouping the gas supply data is as follows: the order p of the AR model is determined using the autocorrelation function graph, and each consecutive p+1 adjacent gas supply data points in time are grouped together.

[0007] In one embodiment, the process for determining the degree of influence is as follows: The collected meteorological data are arranged according to time sequence to form various meteorological sequences; All collected gas supply data are arranged in time sequence to form a gas supply sequence. Calculate the correlation coefficients between each meteorological sequence and the gas supply sequence; Based on the distribution of the correlation coefficients, the degree of influence of various meteorological data on gas supply demand is determined.

[0008] In one embodiment, the degree of influence is the proportion of each correlation coefficient in the sum of all correlation coefficients.

[0009] In one embodiment, the expression for the theoretical gas supply data is: In the formula, This represents the theoretical gas supply data for the v-th gas supply data in the u-th group; k represents the type of meteorological data. This indicates the degree of impact of the i-th type of meteorological data on gas supply demand; the i-th type of meteorological data on the same day as the v-th gas supply data in the u-th group is denoted as I. This represents the average gas supply data for all days that are numerically equal to I.

[0010] In one embodiment, the confidence level is directly proportional to the order of the remaining data in each group and inversely proportional to the error of the remaining data.

[0011] In one embodiment, the feasibility is directly proportional to the confidence level and inversely proportional to the error of the last gas supply data in each group.

[0012] In one embodiment, the feasibility is expressed as: In the formula, The parameter represents the feasibility of the u-th group; norm[] represents the normalization function; This represents the confidence level that the last data point in the u-th group can be predicted by the remaining data. This represents the error of the gas supply data in the (p+1)th group of the u-th group; This indicates a value that is greater than 0 by default.

[0013] In one embodiment, the process of determining the weighted likelihood function is as follows: In the formula, represents the weighted likelihood function; n represents the total number of gas supply data; p represents the order of the AR model; express The probability density function is obtained through the formula for predicting the dependent variable; Indicates the feasibility of the u-th group; The expression for the formula predicting the dependent variable is as follows: In the formula, Let θ represent the row vector consisting of all data except the last data in the u-th group; θ represents the row vector consisting of p unknown autoregressive coefficients; T represents the transpose operation; Represents an unknown constant term; This represents the gas supply data for the (p+1)th group in the u-th group.

[0014] In one embodiment, the method for constructing the AR model for predicting gas supply demand is as follows: The optimal autoregressive coefficients and constant term are obtained by using the maximum likelihood estimation method based on the weighted likelihood function; The expression for the AR model is: ; This represents the gas supply data for day t. This represents the v-th optimal autoregressive coefficient; This represents the gas supply data for TV day; This represents the value of white noise on day t. This represents a constant term.

[0015] This application has at least the following beneficial effects: This application analyzes the correlation between various meteorological data and gas supply data to determine the degree of influence of various meteorological data on gas supply demand. By comparing the correlation between different types of meteorological data and gas supply, it highlights the degree of influence of various meteorological data on gas supply demand. Furthermore, by combining the combined effects of multiple meteorological factors on gas supply, it calculates the difference between the gas supply truly caused by weather changes and the actual gas supply, avoiding interference from historical gas supply data due to factors such as natural disasters or public health emergencies when calculating the autocorrelation coefficient using historical gas supply data. It also analyzes the reliability of gas supply data in each group for prediction, weighting it according to the reliability when constructing the likelihood function, and using the weighted likelihood function to obtain the optimal autoregression parameters, improving the accuracy of the calculated autoregression coefficients. Based on this, an AR model is constructed to predict future gas supply demand, improving the accuracy of gas supply demand prediction and reducing the problems of transition failure due to insufficient gas supply or the high risk of gas storage due to oversupply. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating the steps of a big data-based method for predicting the load of a clean heating system provided in this application; Figure 2 This is a schematic diagram illustrating the process for determining the degree of impact. Detailed Implementation

[0018] In the description of the embodiments in this application, the words "exemplary," "or," and "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary," "or," and "for example" is intended to present the relevant concepts in a specific manner.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. It should be understood that, unless otherwise stated, " / " in this application means "or".

[0020] It should also be noted that the terms "first" and "second" in this application are used to distinguish similar objects, rather than to describe a specific order or sequence.

[0021] The following description, in conjunction with the accompanying drawings, details a specific scheme for a big data-based method for predicting the load of a clean heating system provided in this application.

[0022] This application provides an embodiment of a big data-based method for predicting the load of a clean heating system. Specifically, it provides the following method for predicting the load of a clean heating system based on big data. Please refer to [link to relevant documentation]. Figure 1 The method includes the following steps: Step 1: Obtain daily gas supply data and various meteorological data for multiple historical days.

[0023] Assuming a sufficient gas supply, the gas supply volume for each of the following n consecutive days (including the current day) is collected, with units of... The system obtains various meteorological data for each day of the current day (n consecutive days) from the weather logs of the meteorological station, including temperature, humidity, wind speed, atmospheric pressure, and precipitation, in units of ℃, %, km / h, hPa, and mm, respectively. The temperature is the average of all temperatures occurring within a day, the humidity is the average of all humidity levels occurring within a day, the wind speed is the average of all wind speeds occurring within a day, and the atmospheric pressure is the average of all atmospheric pressures occurring within a day.

[0024] In this embodiment, the value of n is 50. The value of n is preset by the user and can be set by the implementer. This application does not impose any special restrictions.

[0025] Step 2: Based on the autocorrelation of all the acquired gas supply data, determine the order of the AR model, and group the gas supply data according to the order.

[0026] Since this application requires building an AR model based on historical gas supply data and identifying trend changes in gas supply demand through the AR model, the gas supply data needs to be processed as follows before building the AR model: (1) The order p of the AR model is determined by using the autocorrelation function (ACF) plot based on the collected historical gas supply data; wherein, the use of the ACF plot to determine the order of the AR model is a well-known technique and will not be described in detail in this application. (2) Group the gas supply data. Since the order p of the AR model has been determined, the gas supply data for each p days can predict the gas supply demand for the p+1th day. Therefore, the gas supply data for each consecutive p+1 days in time will be grouped as a gas supply data set, as shown in the example below: There are np groups in total. Each is a group. These are the gas supply data for the 1st, 2nd, 3rd, pth, p+1th, p+2th, npth, n-p+1th, n-1th, and nth gas supply data, respectively.

[0027] Step 3: Analyze the correlation between various meteorological data and gas supply data to determine the degree of influence of various meteorological data on gas supply demand.

[0028] Since gas demand is closely related to outdoor weather conditions, temperature, humidity, wind speed, atmospheric pressure and precipitation affect people's daily activities and comfort in different ways, and therefore have different degrees of impact on gas demand.

[0029] Since the collected meteorological data are obtained under the condition of maintaining a sufficient gas supply, the stronger the positive correlation between one type of meteorological data and the gas supply data, the greater the impact of that meteorological data on the gas supply demand.

[0030] Based on the above analysis, the correlation between various meteorological data and gas supply data is analyzed to determine the degree of influence of various meteorological data on gas supply demand. The expression is as follows: In the formula, This indicates the degree of impact of the i-th type of meteorological data on gas supply demand; This indicates the operation of calculating the correlation coefficient; k represents the type of meteorological data, and in this embodiment, k=5; E represents the meteorological sequence composed of all the collected i-th meteorological data arranged in chronological order; E represents the gas supply sequence composed of all the collected gas supply data arranged in chronological order; ω represents a preset constant greater than 0, the purpose of which is to avoid the denominator being 0. The value of ω is preset by the user and can be set by the implementer. In this embodiment, the value of ω is 0.001.

[0031] In this embodiment, the correlation coefficient between the meteorological sequence and the gas supply sequence is the Pearson correlation coefficient. As another implementation method, based on the ability to measure the correlation between the meteorological sequence and the gas supply sequence, the implementer may use other existing technologies to measure the correlation between the meteorological sequence and the gas supply sequence, such as Spearman correlation coefficient, Kendall rank correlation coefficient, etc. This application does not impose any special restrictions.

[0032] It should be noted that the larger the calculated impact value, the greater the impact of the i-th meteorological data on gas supply demand compared to other types of meteorological data. A flowchart illustrating the process of determining the impact level is shown below. Figure 2 As shown.

[0033] Step 4: Based on the distribution of gas supply data for all days with the same meteorological data, and in conjunction with the degree of influence, determine the theoretical gas supply data for each gas supply data in each group; based on the difference between each gas supply data in each group and its theoretical gas supply data, determine the error of each gas supply data in each group.

[0034] The core of the change in gas supply demand is to respond to meteorological changes. For each group, each gas supply data in each group may be formed due to the influence of factors such as natural disasters or public health emergencies, and is not the result of multiple meteorological data. Therefore, it is necessary to calculate the error of each gas supply data relative to the gas supply data that is actually caused by weather factors.

[0035] Based on the above analysis, assuming that the gas supply data of the vth group in the uth group is only affected by the ith type of meteorological data, and temporarily ignoring the influence of other types of meteorological data on the gas supply, the day on which the vth gas supply data in the uth group is collected is denoted as day a, the ith type of meteorological data on day a is denoted as I, and the average value of the gas supply data of all days that are numerically equal to I is denoted as... Then the gas supply on day a should be the same as... While the gas supply is equal, it is influenced by multiple meteorological data. Therefore, by combining the influence of various meteorological data on gas supply demand, a weighted average is calculated to obtain the theoretical gas supply caused by meteorological data. The expression is as follows: In the formula, This represents the theoretical gas supply data for the v-th gas supply data in the u-th group; k represents the type of meteorological data. This indicates the degree of impact of the i-th type of meteorological data on gas supply demand; This represents the average gas supply data for all days that are numerically equal to I.

[0036] Furthermore, the difference between the gas supply data of each group and its theoretical gas supply data is taken as the error of the gas supply data of each group.

[0037] In this embodiment, the difference between each gas supply data and its theoretical gas supply data is the absolute value of the difference. As another implementation method, based on the ability to measure the difference between each gas supply data and its theoretical gas supply data, the implementer may use other calculation methods for measurement, such as ratio, square of difference, etc. This application does not impose any special restrictions.

[0038] Step 5: Based on the error and order of the gas supply data in each group, determine the confidence level that the preset location data in each group can be predicted by the other data.

[0039] For each group of p+1 gas supply data points, in order for the first p gas supply data points to accurately predict the p+1th gas supply data point, the error of the first p gas supply data points should be as small as possible. Furthermore, since there is autocorrelation among the gas supply data points, that is, the gas supply data point of any day is related to the gas supply data point of the previous day or multiple days, the gas supply data point of the first p gas supply data points that is closer to the p+1th gas supply data point in time is more correlated with the p+1th gas supply data point.

[0040] Based on the above analysis, and considering the order and error of the gas supply data in each group, the confidence level that the last data point in each group can be predicted by the remaining data is determined by the following expression: ; represents the confidence level that the last data point in the u-th group can be predicted by the remaining data; p represents the order of the AR model; v represents the order of the gas supply data in the group; The error of the gas supply data of the vth group in the uth group is represented by norm[], which represents the normalization function. This indicates a preset value greater than 0, the purpose of which is to avoid a denominator of 0. The value is preset by a person, and the implementer can set it himself. In this embodiment... The value is 0.001.

[0041] In this embodiment, the Min-Max normalization method is used to... Normalization is performed as another implementation method to achieve the desired result. Based on the normalization process, implementers may use other existing technologies, such as Z-Score normalization method, decimal scaling normalization method, etc., and this application does not impose any special restrictions.

[0042] It should be noted that when calculating the confidence level, v is used as a weight, and the gas supply data that is closer to the p+1th gas supply data in time is assigned a higher weight.

[0043] Step 6: Based on the confidence level and the error of the gas supply data at the preset location in each group, determine the feasibility of each group.

[0044] This application calculates the autoregressive coefficients and constant terms of the AR model using all groupings, and analyzes each group. The first p gas supply data are used to predict the (p+1)th gas supply data. If the (p+1)th gas supply data is caused by weather factors, the error of the (p+1)th gas supply data is relatively small. Since the (p+1)th gas supply data can be regarded as the prediction result of the first p gas supply data, if this group of gas supply data is feasible in data prediction, then on the basis of small prediction error, the confidence that the (p+1)th gas supply data can be predicted by the first p gas supply data is also high.

[0045] Based on the above analysis, and considering the confidence level that the last data point in each group can be predicted by the remaining data, as well as the error of the last gas supply data, the feasibility of each group is determined using the following expression: In the formula, The parameter represents the feasibility of the u-th group; norm[] represents the normalization function; This represents the confidence level that the last data point in the u-th group can be predicted by the remaining data. This represents the error of the gas supply data in the (p+1)th group of the u-th group; This indicates a preset value greater than 0, the purpose of which is to avoid a denominator of 0. The value is preset by a person, and the implementer can set it himself. In this embodiment... The value is 0.001.

[0046] In this embodiment, the Min-Max normalization method is used to... Normalization is performed as another implementation method to achieve the desired result. Based on the normalization process, implementers may use other existing technologies, such as Z-Score normalization method, decimal scaling normalization method, etc., and this application does not impose any special restrictions.

[0047] Step 7: Combine the gas supply data from all groups with the feasibility to determine the weighted likelihood function; construct an AR model for predicting gas supply demand based on the likelihood function.

[0048] For each group, the first p gas supply data points can be used as independent variables for a prediction sample, and the (p+1)th gas supply data point can be used as the dependent variable for a prediction sample. The independent variables are associated with p unknown autoregressive coefficients. The unknown constant term μ, together with the unknown constant term μ, can form a formula for predicting the dependent variable, expressed as: In the formula, This represents the row vector consisting of all data in the u-th group except for the last data, and its specific expression is: ,in Let represent the 1st, 2nd, and pth gas supply data in the u-th group, respectively; θ represents a row vector composed of p unknown autoregressive coefficients, specifically expressed as: ,in These represent the 1st, 2nd, and pth autoregressive coefficients, respectively; T represents the transpose operation. This represents the gas supply data for the (p+1)th group in the u-th group; represents an unknown constant term; p represents the order of the AR model.

[0049] Existing technologies generally adopt The likelihood function is constructed using the method described in the equation, where . represents the likelihood function; n represents the total number of gas supply data; p represents the order of the AR model; express The probability density function is obtained through the formula for predicting the dependent variable; however, this method of constructing the likelihood function essentially treats the feasibility of different groups as the same, but the feasibility of different groups is different after calculation. In order to make the group with higher feasibility contribute more and the group with lower feasibility contribute less, the likelihood function can be constructed by weighting in the process of determining multiple unknown parameters using the maximum likelihood estimation method to achieve parameter optimization.

[0050] Based on the above analysis, and considering the feasibility of each group and the formula for predicting the dependent variable, the weighted likelihood function is determined as follows: In the formula, represents the weighted likelihood function; n represents the total number of gas supply data; p represents the order of the AR model; express The probability density function is obtained through the formula for predicting the dependent variable; This indicates the feasibility of the u-th group. Obtaining the probability density function through the formula for predicting the dependent variable is a well-known technique, and will not be elaborated upon in this application.

[0051] The optimal autoregressive coefficients and constant term can be obtained by calculating the partial derivatives of the weighted likelihood function and solving the system of equations using the gradient descent method. The calculation of the partial derivatives and the solution of the system of equations using the gradient descent method are well-known techniques and will not be elaborated upon in this application.

[0052] Based on the optimal autoregressive coefficients, constant term, and random disturbance term, an AR model is constructed, expressed as follows: ; This represents the gas supply data for day t. represents the constant term; p represents the order of the AR model; This represents the v-th optimal autoregressive coefficient; This represents the gas supply data for TV day; Let represent the value of white noise on day t, where white noise is a random disturbance term.

[0053] AR models are used to predict future gas supply based on historical gas supply data. By using these predictions, gas companies can dynamically adjust their supply strategies based on real-time changes, plan and manage the supply chain in advance, and reduce the risks of transition failures due to insufficient supply or significant gas storage risks due to oversupply. This allows cities to successfully achieve their clean heating transition goals.

[0054] In summary, this application analyzes the correlation between various meteorological data and gas supply data, determines the degree of influence of various meteorological data on gas supply demand, and highlights the degree of influence of various meteorological data on gas supply demand by comparing the correlation between different types of meteorological data and gas supply. Furthermore, by combining the combined effects of multiple meteorological factors on gas supply, it calculates the difference between the gas supply truly caused by weather changes and the actual gas supply, avoiding interference from historical gas supply data due to factors such as natural disasters or public health emergencies when calculating the autocorrelation coefficient using historical gas supply data. It also analyzes the reliability of gas supply data in each group for prediction, weights it according to the reliability when constructing the likelihood function, and uses the weighted likelihood function to obtain the optimal autoregression parameters, improving the accuracy of the calculated autoregression coefficient. Based on this, an AR model is constructed to predict future gas supply demand, improving the accuracy of gas supply demand prediction and reducing the problems of transition failure due to insufficient gas supply or the high risk of gas storage due to oversupply.

[0055] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0056] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from its essential characteristics. Therefore, the embodiments described above should be considered exemplary and non-limiting in all respects.

Claims

1. A method for load forecasting of clean heating systems based on big data, characterized in that, The method includes the following steps: Obtain daily gas supply data and various meteorological data over multiple historical days; Based on the autocorrelation of all acquired gas supply data, the order of the AR model is determined, and the gas supply data is grouped based on the order. Analyze the correlation between various meteorological data and gas supply data to determine the degree of influence of various meteorological data on gas supply demand; Based on the distribution of gas supply data for all days with the same meteorological data, and in combination with the aforementioned degree of influence, the theoretical gas supply data for each gas supply data in each group is determined; based on the difference between each gas supply data in each group and its theoretical gas supply data, the error of each gas supply data in each group is determined. Based on the error and order of the gas supply data in each group, the confidence level that the preset location data in each group can be predicted by the other data is determined; the confidence level is directly proportional to the order of the other data in each group and inversely proportional to the error of the other data. Based on the confidence level and the error of the gas supply data at the preset position in each group, the feasibility of each group is determined; the feasibility is directly proportional to the confidence level and inversely proportional to the error of the last gas supply data in each group. By combining the gas supply data from all groups with the aforementioned feasibility, a weighted likelihood function is determined; an AR model for predicting gas supply demand is then constructed based on the likelihood function. The process for determining the weighted likelihood function is as follows: In the formula, represents the weighted likelihood function; n represents the total number of gas supply data; p represents the order of the AR model; express The probability density function is obtained through the formula for predicting the dependent variable; Indicates the feasibility of the u-th group; The expression for the formula predicting the dependent variable is as follows: In the formula, Let θ represent the row vector consisting of all data except the last data in the u-th group; θ represents the row vector consisting of p unknown autoregressive coefficients; T represents the transpose operation; μ represents the unknown constant term. This represents the gas supply data for the (p+1)th group in the u-th group.

2. The method for load forecasting of a clean heating system based on big data as described in claim 1, characterized in that, The method for grouping the gas supply data is as follows: the order p of the AR model is determined using the autocorrelation function graph, and each consecutive p+1 adjacent gas supply data points in time are grouped together.

3. The method for load forecasting of a clean heating system based on big data as described in claim 1, characterized in that, The process for determining the degree of influence is as follows: The collected meteorological data are arranged according to time sequence to form various meteorological sequences; All collected gas supply data are arranged in time sequence to form a gas supply sequence. Calculate the correlation coefficients between each meteorological sequence and the gas supply sequence; Based on the distribution of the correlation coefficients, the degree of influence of various meteorological data on gas supply demand is determined.

4. The method for load forecasting of a clean heating system based on big data as described in claim 3, characterized in that, The degree of influence refers to the proportion of each correlation coefficient in the sum of all correlation coefficients.

5. The method for load forecasting of a clean heating system based on big data as described in claim 1, characterized in that, The expression for the theoretical gas supply data is: In the formula, This represents the theoretical gas supply data for the v-th gas supply data in the u-th group; k represents the type of meteorological data. This indicates the degree of impact of the i-th type of meteorological data on gas supply demand; the i-th type of meteorological data on the same day as the v-th gas supply data in the u-th group is denoted as I. This represents the average gas supply data for all days that are numerically equal to I.

6. The method for load forecasting of a clean heating system based on big data as described in claim 1, characterized in that, The expression for feasibility is: In the formula, The parameter represents the feasibility of the u-th group; norm[] represents the normalization function; This represents the confidence level that the last data point in the u-th group can be predicted by the remaining data. This represents the error of the gas supply data for the (p+1)th group in the u-th group; τ represents a preset value greater than 0.

7. The method for load forecasting of a clean heating system based on big data as described in claim 6, characterized in that, The method for constructing the AR model used to predict gas supply demand is as follows: The optimal autoregressive coefficients and constant term are obtained by using the maximum likelihood estimation method based on the weighted likelihood function; The expression for the AR model is: ; This represents the gas supply data for day t. This represents the v-th optimal autoregressive coefficient; This represents the gas supply data for TV day; This represents the value of white noise on day t. This represents a constant term.

Images