A heating demand load evaluation method and system based on room temperature and IBA-LSTM

Abstract

The present invention provides a method and system for evaluating the heating demand load based on room temperature and IBA-LSTM, including: setting multiple indoor temperature measurement points for different heat users in a heating system, obtaining indoor temperature measurement values ​​at multiple times within a preset period; assigning different weight coefficients to multiple indoor temperature measurement points of different heat users, and calculating the indoor temperature fusion value of each indoor temperature measurement point; using a sensitivity analysis method to analyze the various influencing factors affecting the heating demand load evaluation, and screening out important influencing factors; using an improved bat algorithm IBA to optimize the parameters of the LSTM network algorithm; constructing a heating demand load evaluation model based on the optimized LSTM network algorithm, and evaluating the load demand of the heating system. The present invention determines the indoor temperature fusion value by determining the weight coefficient for each room temperature measurement point, and uses IBA-LSTM to construct a heating demand load evaluation model, which can improve the model training speed and has high model evaluation accuracy.

Description

Technical Field

[0001] The present invention belongs to the technical field of smart heating, and specifically relates to a heating demand load evaluation method based on room temperature and IBA-LSTM. Background Art

[0002] With the rapid development of information technology and smart heating, a large amount of heating data has been preserved. Mining effective information from historical data and building a reasonable and appropriate heating load forecasting model are of great significance for heating companies to improve heating quality and heating energy efficiency. Load forecasting is mainly achieved by analyzing the changing characteristics of heating heat load and combining the factors affecting load changes to accurately forecast the heat load. Timely and accurate heat load forecasting can not only guarantee the heat demand of users, but also guarantee the interests of heating companies and achieve a balance between supply and demand. Therefore, heat load forecasting is an important basis for heating companies to make operational decisions, and it is a necessary foundation for heating companies to reduce energy waste, achieve efficient management of centralized heating systems, and operate economically.

[0003] Since long-term stable indoor temperature data is difficult to obtain, most load assessments only consider the impact of factors such as outdoor temperature and historical heating load, but not the impact of indoor temperature. In practice, when the building type, human behavior, heating standards, etc. are determined, under certain meteorological conditions, the heating load theoretically has a fixed value, while the heating amount can be infinite, and different indoor temperatures can correspond to different heating amounts. If the indoor temperature is ignored, the accuracy of the prediction model cannot be guaranteed. In addition, the load assessment model has a large number of input variables. From the perspective of model construction, the more input variables there are, the longer the model training takes. In addition, there are many algorithms for load assessment. How to combine the actual engineering data, comprehensively consider the assessment accuracy of the model, and select an appropriate assessment method is also a problem currently faced. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the present invention provides a heating demand load assessment method and system based on room temperature and IBA-LSTM, by determining the weight coefficient of each room temperature measuring point, and finally determining the indoor temperature fusion value, the indoor temperature fusion value can effectively and accurately characterize the indoor temperature, which is beneficial to the accuracy of the construction of the heating demand load assessment model; and adopts an improved intelligent optimization algorithm to optimize the parameters of the LSTM network algorithm, and constructs a heating demand load assessment model based on the optimized LSTM network algorithm, which can improve the model training speed, reduce the model training error, the model evaluation accuracy is high, and the heating demand load can be better evaluated.

[0005] In order to solve the above technical problems, the technical solution of the present invention is:

[0006] The present invention provides a method for evaluating heating demand load based on room temperature and IBA-LSTM, comprising:

[0007] Step S1, setting multiple indoor temperature measurement points for different heat users in the heating system, and obtaining indoor temperature measurement values ​​at multiple times within a preset period;

[0008] Step S2, assigning different weight coefficients to multiple indoor temperature measurement points of different heat users, and calculating the indoor temperature fusion value of each indoor temperature measurement point;

[0009] Step S3: Analyze various factors affecting the heating demand load assessment using a sensitivity analysis method to screen out important factors;

[0010] Step S4: using an improved intelligent optimization algorithm to optimize the parameters of the LSTM network algorithm;

[0011] Step S5: construct a heating demand load assessment model based on the optimized LSTM network algorithm to assess the load demand of the heating system.

[0012] Furthermore, step S1 includes: setting the measurement point position of each user floor according to the area, orientation, layout and wall and window position relationship of each user floor; setting multiple indoor temperature measurement points in the indoor rooms of the bottom floor users, top floor users and middle floor users of a unit.

[0013] Furthermore, the step S2 comprises:

[0014] Obtain historical heating load data, preset indoor temperature target values ​​for heating, use AHP hierarchical analysis method to analyze heating users, and obtain weight coefficients for each indoor temperature monitoring point;

[0015] The indoor temperature fusion value is calculated based on the indoor temperature measurement value of each indoor temperature monitoring point and the corresponding weight coefficient. The calculation formula is:

[0016]

[0017] n is the number of indoor temperature monitoring points; Where i = 1, 2, 3, ..., n is the indoor temperature measurement value of the i-th indoor temperature monitoring point at time k; c i where i=1,2,3,…,n,c1+c2+c3+…+c n =1 is the weight coefficient of each indoor temperature monitoring point;

[0018] The weight coefficient calculation formula is expressed as:

[0019]

[0020] is the arithmetic mean of the indoor temperature measurements, and r is the adjustment coefficient.

[0021] Furthermore, the step S3 comprises:

[0022] Set the output function Y=f(x1,x2,…,x z ), including z influencing parameters; the influencing factors of the parameters include: indoor temperature, heating history parameters, building thermal characteristic parameters, meteorological parameters and social parameters;

[0023] The heating history parameters include return water temperature, flow rate, pressure and heating load;

[0024] The building thermal characteristic parameters include the type, age, geographical location and orientation of the building;

[0025] The meteorological parameters include outdoor temperature, wind speed and solar radiation; the social parameters include holidays and rest days.

[0026] According to the probability distribution of each parameter, the value range of each parameter is mapped to [0,1], discretized by a preset sampling level p, and each parameter is randomly sampled once at the p sampling levels to generate a vector X.

[0027] Preferably, the important influencing factors screened out in step S3 include:

[0028] In the process of generating samples, a base sample point X is generated. 1 , increase the disturbance variable, and generate other sample points X in turn i , after generating different sample points, we obtain z+1 sample points and calculate the basic effect value of each parameter:

[0029]

[0030] By calculating the mean μ of the base effect values ​​of each parameter i and standard deviation σ i , to judge the importance or sensitivity of the parameter, the mean μ i and standard deviation σ i The calculation is expressed as:

[0031]

[0032]

[0033] Among them, d i,e (X) is X i The e-th basic effect value, N is the sample size;

[0034] The mean represents the sensitivity of the parameter. The larger the value, the more important the parameter is, which in turn determines the sensitivity ranking of the parameters. The standard deviation represents the degree of interaction between parameters. If the standard deviation is large, it means that the interaction between the parameter and other parameters is large. If the standard deviation is small, it means that the interaction between the parameter and other parameters is small.

[0035] Furthermore, the step S4 comprises the following steps:

[0036] S401. Set the basic parameters of the bat population, including the total number of individuals m, the dimension d, and the maximum value f of the pulse emission frequency. max , minimum value f min , pulse wave emission rate r i , pulse rate increase / decrease coefficient γ, loudness A i , loudness reduction coefficient α and maximum number of iterations N max ;

[0037] S402, set the speed and position coordinates of the individual bat, including the position x of the individual bat i (h,η,n), flight speed v i and the pulse emission frequency f i ; h is the number of hidden layer neurons, η is the learning rate of LSTM, and n is the number of iterations when the LSTM loss function reaches the requirement;

[0038] S403, modifying the calculation formula of the position and flight speed of individual bats;

[0039] S404, check the random number rand obtained by random flight, when rand>r i When , a new solution is obtained by performing non-directional perturbations on the position of the current optimal solution;

[0040] S405, in the process of optimizing LSTM with the bat algorithm, the root mean square error is used as the fitness function;

[0041] S406: Check the obtained random number and the new solution obtained in S403. If the condition rand<A is met, i And f(x i )<f(x * ), the f(x i ) is at position x i The pulse emission frequency; f(x * ) is the corrected optimal position x * The pulse emission frequency. * is the optimal position calculated after comparing all bat positions; choose to accept the new solution and at the same time increase r i The value of A is reduced iThe value of

[0042] S407, comparing the obtained new solution with the global optimal solution, if it is better than the global optimal solution, then correcting the optimal solution and the optimal position of the individual bat;

[0043] S408, judging whether the number of iterations meets the maximum number of iterations requirement, if not, using the gradient descent method to find a new solution in the neighborhood of the current optimal solution, when the new solution is better than the current optimal solution, the optimal solution is replaced by the newly calculated solution; if satisfied, the optimal solution remains unchanged, and the process goes to S402 to repeat;

[0044] S409. If the number of iterations is met, the optimal solution found is output, that is, the optimal solution of the number of hidden layer neurons, LSTM learning rate, and number of iterations is obtained.

[0045] Preferably, in step S403, the calculation formula for correcting the position and flight speed of the individual bat is expressed as:

[0046] f i =f min +(f max -f min )β;

[0047]

[0048]

[0049] Among them, β∈[0,1] satisfies uniform distribution, x * is the optimal position calculated after comparing all bat positions; the maximum value of the pulse emission frequency f max , minimum value f min and the pulse emission frequency f i ; are the flight speeds at time t and time t+1 respectively; They are the positions at time t and time t+1 respectively.

[0050] Preferably, in step S406, increasing r i The value of A is reduced i The value of is:

[0051] Bats will constantly adjust the loudness of sound waves and the pulse wave emission rate according to the position of prey to improve their hunting efficiency; the calculation of the loudness of sound waves and the pulse wave emission rate of bats is expressed as:

[0052]

[0053]

[0054] is the initial pulse wave emission rate of the bat; is the pulse wave emission rate of the bat at time t; are the sound wave loudness of the bat at time t and time t+1 respectively.

[0055] Furthermore, the step S5 comprises:

[0056] Collect historical data on important influencing factors of the heating system as sample data, and perform data preprocessing on the sample data;

[0057] Divide the preprocessed data into training set and test set according to the preset ratio;

[0058] The data of the training set is input into the LSTM network algorithm after parameter optimization for training, a heating demand load evaluation model is generated, and then the trained model is evaluated on the test set;

[0059] The data to be evaluated is input into the trained heating demand load evaluation model to obtain the heating demand load evaluation value.

[0060] A heating demand load evaluation system based on room temperature and IBA-LSTM, comprising:

[0061] Room temperature measurement point setting module: used to set multiple indoor temperature measurement points for different heat users in the heating system, and obtain indoor temperature measurement values ​​at multiple times within a preset period;

[0062] Measurement point weight calculation module: used to assign different weight coefficients to multiple indoor temperature measurement points of different heat users, and calculate the indoor temperature fusion value of each indoor temperature measurement point;

[0063] Sensitivity analysis module: used to analyze various factors affecting the evaluation of heating demand load using sensitivity analysis methods and screen out important factors;

[0064] LSTM algorithm optimization module: used to optimize the parameters of the LSTM network algorithm using an improved intelligent optimization algorithm;

[0065] Evaluation model building module: used to build a heating demand load evaluation model based on the optimized LSTM network algorithm to evaluate the load demand of the heating system.

[0066] The beneficial effects of the present invention are:

[0067] (1) The present invention sets multiple indoor temperature measurement points for different heat users in the heating system, and obtains indoor temperature measurement values ​​at multiple times within a preset period; assigns different weight coefficients to multiple indoor temperature measurement points of different heat users, and calculates indoor temperature fusion values ​​of each indoor temperature measurement point; and the weight coefficient can be dynamically adjusted, and the weight coefficient can be determined for each room temperature measurement point, and finally the indoor temperature fusion value is determined. The indoor temperature fusion value can effectively and accurately characterize the indoor temperature, which is beneficial to the accuracy of the construction of the heating demand load assessment model;

[0068] (2) The present invention adopts a sensitivity analysis method to analyze the various influencing factors that affect the evaluation of heating demand load, which can screen out important influencing factors, ignore the less sensitive factors, and improve the efficiency of subsequent evaluation model construction;

[0069] (3) The present invention optimizes the parameters of the LSTM network algorithm through an improved bat algorithm IBA, uses the improved bat algorithm to optimize the number of neurons in the hidden layer of the LSTM model, the learning rate, and the number of iterations, and uses the descent gradient optimization as interference to assist the bat algorithm to jump out of the extreme value, so as to achieve a more efficient and rapid search for the optimal solution; and uses the improved bat algorithm to optimize the LSTM parameters, and constructs a heating demand load evaluation model based on the optimized LSTM network algorithm, which can improve the model training speed, reduce the model training error, and the model evaluation accuracy is high, which can better evaluate the heating demand load.

[0070] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0072] Figure 1 This is a flow chart of a method for evaluating heating demand load based on room temperature and IBA-LSTM according to the present invention;

[0073] Figure 2 This is a structural schematic diagram of a heating demand load evaluation system based on room temperature and IBA-LSTM in the present invention. DETAILED DESCRIPTION

[0074] The present invention is further described below in conjunction with the accompanying drawings and embodiments: In order to clearly illustrate the technical features of this solution, the present invention is described in detail below through specific implementation methods and in conjunction with its accompanying drawings. The disclosure below provides many different embodiments or examples for realizing different structures of the present invention. In order to simplify the disclosure of the present invention, the components and settings of specific examples are described below. In addition, the present invention may repeat reference numbers and / or letters in different examples. This repetition is for the purpose of simplification and clarity, and does not itself indicate the relationship between the various embodiments and / or settings discussed. It should be noted that the components illustrated in the accompanying drawings are not necessarily drawn to scale. The present invention omits the description of known components and processing techniques and processes to avoid unnecessary limitations on the present invention.

[0075] In one embodiment of the present invention, Figure 1 As shown, this embodiment 1 provides a method for evaluating heating demand load based on room temperature and IBA-LSTM, including:

[0076] Step S1, setting multiple indoor temperature measurement points for different heat users in the heating system, and obtaining indoor temperature measurement values ​​at multiple times within a preset period;

[0077] Step S2, assigning different weight coefficients to multiple indoor temperature measurement points of different heat users, and calculating the indoor temperature fusion value of each indoor temperature measurement point;

[0078] Step S3: Analyze various factors affecting the heating demand load assessment using a sensitivity analysis method to screen out important factors;

[0079] Step S4: using an improved intelligent optimization algorithm to optimize the parameters of the LSTM network algorithm;

[0080] Step S5: construct a heating demand load assessment model based on the optimized LSTM network algorithm to assess the load demand of the heating system.

[0081] In this embodiment, in step S1, multiple indoor temperature measuring points are set for different heat users in the heating system, and indoor temperature measurement values ​​at multiple times are obtained within a preset period, including: multiple indoor temperature measuring points are set in the indoor rooms of the ground floor users, top floor users and middle floor users of a unit; wherein, the house types of users on each floor are the same, and the measurement point positions of users on each floor are set according to the area, orientation, layout and wall and window position relationship of the house types of users on each floor.

[0082] In this embodiment, in step S2, different weight coefficients are assigned to multiple indoor temperature measurement points of different heat users, and indoor temperature fusion values ​​of each indoor temperature measurement point are calculated, including:

[0083] According to the multiple indoor temperature measurement points of different heat users, the corresponding indoor temperature measurement values ​​at multiple times are collected and the corresponding historical heating load data is obtained, the indoor temperature target value of heating is preset, and the AHP hierarchical analysis method is used to perform heating analysis on heat users to obtain the weight coefficient of each indoor temperature monitoring point;

[0084] The indoor temperature fusion value is calculated based on the indoor temperature measurement values ​​of each indoor temperature monitoring point and the corresponding weight coefficient, which is expressed as:

[0085]

[0086] n is the number of indoor temperature monitoring points; Where i = 1, 2, 3, ..., n is the indoor temperature measurement value of the i-th indoor temperature monitoring point at time k; c i where i=1,2,3,…,n,c1+c2+c3+…+c n =1 is the weight coefficient of each indoor temperature monitoring point;

[0087] The weight coefficient is dynamically adjusted over time and is a dynamic weighting factor. The calculation formula is expressed as follows:

[0088]

[0089] is the arithmetic mean of the indoor temperature measurement values, and r is the adjustment coefficient, which is related to the error of the sensor measuring the indoor temperature itself and the average and fusion values ​​of the indoor temperature at different indoor temperature monitoring points at different times.

[0090] In this embodiment, in step S3, a sensitivity analysis method is used to analyze various influencing factors affecting the heating demand load assessment, and important influencing factors are screened out, specifically including:

[0091] Set the output function Y=f(x1,x2,…,x z ), including z influencing parameters, and then according to the probability distribution of each parameter, the value range of each parameter is mapped to [0,1], and discretized by the preset sampling level p, so that the value of each parameter is Randomly select values ​​from the p sampling points, and make each parameter randomly select values ​​at p sampling points, generating a vector X = (x1, x2, ..., x z ); Among them, the influencing parameters at least include indoor temperature, heating history parameters, building thermal characteristic parameters, meteorological parameters and social parameters; heating history parameters include supply and return water temperature, flow, pressure and heating load; building thermal characteristic parameters include building type, age, geographical location and orientation; meteorological parameters include outdoor temperature, wind speed and solar radiation; social parameters include holidays and rest days;

[0092] In the process of generating samples, a base sample point X is first generated. 1 , and then increase the disturbance variable to generate other sample points X in turn i , after generating different sample points, we obtain z+1 sample points and calculate the basic effect value of each parameter:

[0093]

[0094] By calculating the mean μ of the base effect values ​​of each parameter i and standard deviation σ i , to judge the importance or sensitivity of the parameter, the mean μ i and standard deviation σ i The calculation is expressed as:

[0095]

[0096]

[0097] Among them, d i,e (X) is X i The e-th basic effect value, N is the sample size; the mean represents the sensitivity of the parameter, the larger the value, the more important the parameter, and then determines the sensitivity ranking of the parameter, specifically: the more important the more sensitive, the larger the value, the higher the sensitivity, indicating that the parameter is more important; the standard deviation represents the degree of interaction between parameters, if the standard deviation is large, it means that the interaction between the parameter and other parameters is large, if the standard deviation is small, it means that the interaction between the parameter and other parameters is small.

[0098] It should be noted that sensitivity analysis is to study the impact of different changes (or variability) in model input parameters on the variability of model output. Generally speaking, sensitivity analysis methods mainly include the following steps:

[0099] (1) Select model parameters and define independent and interdependent parameters of the model;

[0100] (2) Set the parameter range and the parameter probability density function (or prior distribution);

[0101] (3) Select appropriate parameter sampling design methods, such as MC sampling, LHS sampling, and M-OAT sampling, to generate parameter samples;

[0102] (4) Apply the parameter sample to run the model, calculate the corresponding output objective function value, and form an objective function set;

[0103] (5) Select certain methods to evaluate the degree of correlation between model parameters and outputs as well as the internal interactions of parameters.

[0104] In this embodiment, in step S4, an improved intelligent optimization algorithm is used to optimize the parameters of the LSTM network algorithm, specifically including:

[0105] S401. Set the basic parameters of the bat population, including the total number of individuals m, the dimension d, and the maximum value f of the pulse emission frequency. max , minimum value f min , pulse wave emission rate r i , pulse rate increase / decrease coefficient γ, loudness A i , loudness reduction coefficient α and maximum number of iterations N max ;

[0106] S402, set the speed and position coordinates of the individual bat, including the position x of the individual bat i (h,η,n), flight speed v i and the pulse emission frequency f i ; Where h is the number of hidden layer neurons, η is the learning rate of LSTM, and n is the number of iterations when the LSTM loss function reaches the requirement;

[0107] S403, correcting the position and flight speed of individual bats;

[0108] S404, check the random number rand obtained by random flight, when rand>r i When , a new solution is obtained by performing non-directional perturbations on the position of the current optimal solution;

[0109] S405. In the process of optimizing LSTM by the bat algorithm, the root mean square error is used as the fitness function. Selecting a more reasonable fitness function can directly affect the prediction accuracy of the optimized model.

[0110] S406: Check the obtained random number and the new solution obtained in S403. If the condition rand<A is met, i And f(x i )<f(x * ), the f(x i ) is at position x i The pulse emission frequency; f(x * ) is the corrected optimal position x * The pulse emission frequency. * is the optimal position calculated after comparing all bat positions; choose to accept the new solution and at the same time increase r i The value of A is reduced i The value of

[0111] S407, comparing the obtained new solution with the global optimal solution, if it is better than the global optimal solution, then correcting the optimal solution and the optimal position of the individual bat;

[0112] S408, judging whether the number of iterations meets the maximum number of iterations requirement, if not, using the gradient descent method to find a new solution in the neighborhood of the current optimal solution, when the new solution is better than the current optimal solution, the optimal solution is replaced by the newly calculated solution, otherwise, the optimal solution remains unchanged, and the process goes to S402 to repeat;

[0113] S409. If the number of iterations is met, the optimal solution found is output, that is, the optimal solution of the number of hidden layer neurons, LSTM learning rate, and number of iterations is obtained.

[0114] It should be noted that when using LSTM to evaluate the heating demand load, there are the following difficulties: first, the determination of the number of hidden layer neurons; second, the determination of the LSTM learning rate; third, the determination of the number of iterations. The number of hidden layer neurons will affect the fitting effect of the entire LSTM model on the heating demand load. Too few will lead to poor fitting effect, and too many will affect the prediction efficiency and the speed of training and learning; the setting of learning rate and number of iterations will affect the learning rate of the model and the final prediction effect. It is difficult to obtain the optimal solution by subjectively selecting these model parameters, and the subjective trial method takes a lot of time. Therefore, the improved bat algorithm is used to find the optimal solution of the LSTM model parameters.

[0115] The improved bat algorithm is based on the bat algorithm and uses the gradient descent algorithm as a noise, thereby improving the speed of the bat algorithm in finding the optimal solution within the regional range and reducing the probability of the bat algorithm falling into the extreme value within the regional range. The improved bat algorithm can be considered as a combination of the bat algorithm in finding the optimal solution globally and the gradient descent algorithm in finding the optimal solution in the neighborhood. Each time the global optimization is performed, the positions of all individual bats in the population are updated. After completing an update, the gradient descent method is used to calculate the new position of x*, x * =x * -μf′(x * ), f′(x * ) is the derivative of the objective function at the current best position, μ is the correction coefficient of the gradient descent algorithm, if the new position is the optimal solution, then the new solution is retained, otherwise the original optimal solution is retained.

[0116] In this embodiment, in S403, the calculation formula for correcting the position and flight speed of the individual bat is expressed as:

[0117] f i =f min +(f max -f min )β;

[0118]

[0119]

[0120] Among them, β∈[0,1] satisfies uniform distribution, x * is the optimal position calculated after comparing all bat positions. The maximum value of the pulse transmission frequency f max , minimum value f min and the pulse emission frequency f i ; are the flight speeds at time t and time t+1 respectively; They are the positions at time t and time t+1 respectively.

[0121] Preferably, in step S406, increasing r i The value of A is reduced i The value of is:

[0122] Bats will constantly adjust the loudness of sound waves and the pulse wave emission rate according to the position of prey to improve their hunting efficiency; the calculation of the loudness of sound waves and the pulse wave emission rate of bats is expressed as:

[0123]

[0124]

[0125] is the initial pulse wave emission rate of the bat; is the pulse wave emission rate of the bat at time t; are the sound wave loudness of the bat at time t and time t+1 respectively.

[0126] In this embodiment, in step S5, a heating demand load evaluation model is constructed based on the optimized LSTM network algorithm to evaluate the load demand of the heating system, specifically including:

[0127] Collect historical data on important influencing factors of the heating system as sample data, and perform data preprocessing on the sample data;

[0128] Divide the preprocessed data into training set and test set according to the preset ratio;

[0129] The data of the training set is input into the LSTM network algorithm after parameter optimization for training, a heating demand load evaluation model is generated, and then the trained model is evaluated on the test set;

[0130] The data to be evaluated is input into the trained heating demand load evaluation model to obtain the heating demand load evaluation value.

[0131] In this embodiment, the preprocessing of the sample data includes:

[0132] The 3σ principle of Raida criterion was used to remove outliers from the data;

[0133] Use one of the mean method, previous moment value method, cumulative value method, and similarity algorithm to fill in missing values ​​in the data;

[0134] The data was normalized using data standardization methods.

[0135] In this embodiment, after obtaining the heating demand load evaluation value, it also includes: using performance evaluation indicators to measure the fit between the evaluation value and the true value; the performance evaluation indicators include mean square error MSE, root mean square error RMSE, mean absolute error MAE and mean relative error MAPE.

[0136] In one embodiment of the present invention, Figure 2 As shown, this embodiment 2 proposes a heating demand load evaluation system based on room temperature and IBA-LSTM, and the heating demand load evaluation system includes:

[0137] Room temperature measurement point setting module: used to set multiple indoor temperature measurement points for different heat users in the heating system, and obtain indoor temperature measurement values ​​at multiple times within a preset period;

[0138] Measurement point weight calculation module: used to assign different weight coefficients to multiple indoor temperature measurement points of different heat users, and calculate the indoor temperature fusion value of each indoor temperature measurement point;

[0139] Sensitivity analysis module: used to analyze various factors affecting the evaluation of heating demand load using sensitivity analysis methods and screen out important factors;

[0140] LSTM algorithm optimization module: used to optimize the parameters of the LSTM network algorithm using an improved intelligent optimization algorithm;

[0141] Evaluation model building module: used to build a heating demand load evaluation model based on the optimized LSTM network algorithm to evaluate the load demand of the heating system.

[0142] In the embodiments provided in the present application, it should be understood that the disclosed systems and methods can also be implemented in other ways. The system embodiments described above are merely schematic, for example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and a module, a program segment or a part of a code contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0143] In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.

[0144] If the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0145] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A heating demand load evaluation method based on indoor temperature and GDBA-LSTM, characterized in that: include: Step S1, setting multiple indoor temperature measurement points for different heat users in the heating system, and obtaining indoor temperature measurement values ​​at multiple times within a preset period; Step S2, assigning different weight coefficients to multiple indoor temperature measurement points of different heat users, and calculating the indoor temperature fusion value of each indoor temperature measurement point; the indoor temperature fusion value can effectively and accurately represent the indoor temperature; Step S3: Analyze various factors affecting the heating demand load assessment using a sensitivity analysis method to screen out important factors; Step S4, using the improved bat algorithm GDBA to optimize the parameters of the LSTM network algorithm; Step S5: construct a heating demand load evaluation model based on the optimized LSTM network algorithm to evaluate the load demand of the heating system; The step S2 comprises: Obtain historical heating load data, preset indoor temperature target values ​​for heating, use AHP hierarchical analysis method to conduct heating analysis on heat users, and obtain the weight coefficient of each indoor temperature measurement point; The indoor temperature fusion value is calculated based on the indoor temperature measurement values ​​of each indoor temperature measurement point and the corresponding weight coefficient. The calculation formula is: n is the number of indoor temperature measurement points; Where i = 1, 2, 3, ..., n is the indoor temperature measurement value of the i-th indoor temperature measurement point at time k; c i where i=1,2,3,…,n,c1+c2+c3+…+c n =1 is the weight coefficient of each indoor temperature measurement point; The step S3 comprises: Set the output function Y=f(x1,x2,…,x z ), including z influencing parameters; the influencing parameters include: indoor temperature, heating history parameters, building thermal characteristic parameters, meteorological parameters and social parameters; The heating history parameters include return water temperature, flow, pressure and heating load; The building thermal characteristic parameters include the type, age, geographical location and orientation of the building; The meteorological parameters include outdoor temperature, wind speed and solar radiation; the social parameters include holidays and rest days; According to the probability distribution of each parameter, the value range of each parameter is mapped to [0,1], and discretized by the preset sampling level p, so that the value of each parameter is Randomly select values ​​from the p sampling points, and make each parameter randomly select values ​​at p sampling points, generating a vector X = (x1, x2, ..., x z ).

2. The method for evaluating heating demand load based on indoor temperature and GDBA-LSTM according to claim 1, characterized in that: The step S1 comprises: Set the measurement point location for each floor based on the area, orientation, layout, wall, and window positions of the user's apartment on each floor; Multiple indoor temperature measuring points are set up in the rooms of the ground floor users, top floor users and middle floor users of a unit.

3. The method for evaluating heating demand load based on indoor temperature and GDBA-LSTM according to claim 1, characterized in that: The step S5 comprises: Collect historical data on important influencing factors of the heating system as sample data, and perform data preprocessing on the sample data; Divide the preprocessed data into training set and test set according to the preset ratio; The data of the training set is input into the LSTM network algorithm after parameter optimization for training, a heating demand load evaluation model is generated, and then the trained model is evaluated on the test set; The data to be evaluated is input into the trained heating demand load evaluation model to obtain the heating demand load evaluation value.

4. A heating demand load evaluation system based on indoor temperature and GDBA-LSTM, characterized in that: The heating demand load assessment system comprises: Room temperature measurement point setting module: used to set multiple indoor temperature measurement points for different heat users in the heating system, and obtain indoor temperature measurement values ​​at multiple times within a preset period; Measurement point weight calculation module: used to assign different weight coefficients to multiple indoor temperature measurement points of different heat users, and calculate the indoor temperature fusion value of each indoor temperature measurement point; the indoor temperature fusion value can effectively and accurately represent the indoor temperature; Specifically include: Obtain historical heating load data, preset indoor temperature target values ​​for heating, use AHP hierarchical analysis method to conduct heating analysis on heat users, and obtain the weight coefficient of each indoor temperature measurement point; The indoor temperature fusion value is calculated based on the indoor temperature measurement values ​​of each indoor temperature measurement point and the corresponding weight coefficient. The calculation formula is: n is the number of indoor temperature measurement points; Where i = 1, 2, 3, ..., n is the indoor temperature measurement value of the i-th indoor temperature measurement point at time k; c i where i=1,2,3,…,n,c1+c2+c3+…+c n =1 is the weight coefficient of each indoor temperature measurement point; Sensitivity analysis module: used to analyze various factors affecting the evaluation of heating demand load using sensitivity analysis methods and screen out important factors; Specifically include: Set the output function Y=f(x1,x2,…,x z ), including z influencing parameters; the influencing parameters include: indoor temperature, heating history parameters, building thermal characteristic parameters, meteorological parameters and social parameters; The heating history parameters include return water temperature, flow, pressure and heating load; The building thermal characteristic parameters include the type, age, geographical location and orientation of the building; The meteorological parameters include outdoor temperature, wind speed and solar radiation; the social parameters include holidays and rest days; According to the probability distribution of each parameter, the value range of each parameter is mapped to [0,1], and discretized by the preset sampling level p, so that the value of each parameter is Randomly select values ​​from the p sampling points, and make each parameter randomly select values ​​at p sampling points, generating a vector X = (x1, x2, ..., x z ); LSTM algorithm optimization module: used to optimize the parameters of the LSTM network algorithm using the improved bat algorithm GDBA; Evaluation model building module: used to build a heating demand load evaluation model based on the optimized LSTM network algorithm to evaluate the load demand of the heating system.

Images