A heat station heat supply management intelligent prediction method and system
By collecting and recombining heating forecast impact data in heating stations, an LSTM-attention mechanism model is constructed, which solves the problem of insufficient accuracy in secondary water supply temperature prediction in existing technologies, achieves higher precision heating management, and reduces energy waste.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TIANJIN ENERGY INTERNET OF THINGS TECH CO LTD
- Filing Date
- 2025-07-03
- Publication Date
- 2026-05-15
AI Technical Summary
Existing technologies rely too heavily on meteorological conditions for predicting the secondary water supply temperature of heating stations, neglecting user behavior and building functional requirements, resulting in decreased prediction accuracy, especially with larger errors when facing complex operating conditions.
By integrating time variables, collecting multiple sets of heating forecast impact data, reorganizing the data fluctuation degree, constructing a hybrid neural network model based on LSTM-attention mechanism, and combining sliding window correlation analysis, the feature subset is dynamically selected to achieve accurate prediction of secondary water supply temperature.
It improves the accuracy of secondary water supply temperature prediction, reduces energy waste, and ensures the intelligence and accuracy of heating management.
Smart Images
Figure CN120705818B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of heating forecasting technology, and more specifically, to an intelligent forecasting method and system for heating management of heating stations. Background Technology
[0002] The secondary water supply temperature at a heating station refers to the temperature of the hot water delivered to buildings through the secondary side pipe network (i.e., the user-side pipe network) after heat exchange at the heating station in a centralized heating system. It is a crucial parameter for the operation and control of the heating system, directly affecting the indoor heating effect for users and the system's energy efficiency.
[0003] Current technology for predicting water supply temperature in secondary heating networks of public buildings has limitations. The main issue is that prediction models rely excessively on correlation analysis of meteorological conditions, neglecting the impact of personalized control strategies during actual operation. Traditional prediction methods typically only consider the static relationship between meteorological parameters such as outdoor temperature and wind speed and water supply temperature, failing to effectively integrate time-of-day control strategies implemented by heating station operators based on user behavior characteristics and building functional requirements. This technical deficiency leads to decreased prediction accuracy when facing complex operating conditions involving manual temperature adjustments. Summary of the Invention
[0004] In view of this, the present invention proposes an intelligent prediction method and system for heating management of heating stations. The present invention improves the prediction accuracy of secondary water supply temperature, reduces prediction error, and reduces energy waste by integrating time variables.
[0005] This invention proposes an intelligent prediction method for heating management of heat stations, comprising:
[0006] Receive heating management instructions from the heating station, preset a fixed sampling time period, and collect multiple sets of heating prediction impact data of the heating station based on the fixed sampling time period, wherein the fixed sampling time period includes multiple sub-fixed sampling time segments;
[0007] Each set of heating forecast impact data is reorganized to obtain multiple heating forecast impact data groups of the same dimension, and the degree of data fluctuation of each heating forecast impact data in the same dimension heating forecast impact data group is determined.
[0008] The sub-secondary water supply temperature prediction factor of the heating station is calculated based on the degree of fluctuation of all data, and the comprehensive secondary water supply temperature prediction factor of the heating station is calculated based on all sub-secondary water supply temperature prediction factors.
[0009] Historical heating operation data is collected, time-related features are extracted from the historical heating operation data, and a secondary water supply prediction model for the heating station is constructed. The comprehensive secondary water supply temperature prediction factor is input into the secondary water supply prediction model of the heating station to obtain the predicted secondary water supply temperature of the heating station.
[0010] Furthermore, before receiving the heating management instructions from the heating station and setting a fixed sampling time period, it also includes:
[0011] Obtain the previous management instruction and determine the instruction category of the previous management instruction;
[0012] The heating station management instruction is matched with the instruction category of the previous management instruction, and it is determined whether the heating station management instruction is the same as the previous management instruction.
[0013] If the heating station management instruction matches the instruction category of the previous management instruction, then it is determined that the heating station management instruction is the same as the previous management instruction. The first time point of sending the heating station management instruction is collected, and the second time point of sending the previous management instruction is collected.
[0014] Calculate the time difference between the first time node and the second time node, and determine whether the heating station heating management instruction can be received based on the relationship between the time difference and the preset time difference.
[0015] If the time node difference is greater than or equal to the preset time node difference, it is determined that the heating station heating management instruction can be received.
[0016] If the time node difference is less than the preset time node difference, it is determined that the heating station heating management instruction cannot be received, and a log reminder is generated and sent.
[0017] If the heating station management instruction does not match the instruction category of the previous management instruction, it is determined that the heating station management instruction is different from the previous management instruction, and the heating station management instruction is received.
[0018] Furthermore, in determining the degree of data fluctuation for each heating forecast impact data point in the same-dimensional heating forecast impact data group, the following steps are included:
[0019] Based on the sub-fixed sampling time segment, the heating prediction impact data group of the same dimension is sorted in time order, and the initial heating prediction impact data and the final heating prediction impact data are determined.
[0020] One heating prediction impact data point is randomly extracted from the same-dimensional heating prediction impact data group and used as the baseline heating prediction impact data.
[0021] Calculate the initial data difference between the baseline heating forecast impact data and the initial heating forecast impact data, wherein the initial data difference is the absolute value of the difference between the baseline heating forecast impact data and the initial heating forecast impact data;
[0022] Calculate the difference between the baseline heating forecast impact data and the final heating forecast impact data, wherein the difference between the final data is the absolute value of the difference between the baseline heating forecast impact data and the final heating forecast impact data.
[0023] The degree of data fluctuation in the baseline heating forecast impact data is calculated based on the initial data difference and the final data difference.
[0024] The remaining heating forecast impact data in the same dimension heating forecast impact data group are extracted and calculated to obtain the corresponding data fluctuation degree.
[0025] Furthermore, when calculating the degree of data fluctuation in the baseline heating forecast impact data based on the initial data difference and the final data difference, the calculation includes:
[0026] The degree of data fluctuation affecting the baseline heating forecast data is calculated using the following formula:
[0027]
[0028] Where q represents the data fluctuation degree of the baseline heating forecast impact data, w1 represents the initial data difference, w2 represents the final data difference, r represents the number of heating forecast impact data in the same dimension heating forecast impact data group, and y t+1 For the (t+1)th heating forecast impact data in the same dimension heating forecast impact data group, y t This is the t-th heating forecast impact data in the same dimension heating forecast impact data group.
[0029] Furthermore, when calculating the sub-secondary water supply temperature prediction factor for the heating station based on the degree of fluctuation of all data, the following is included:
[0030] Determine the first standard data fluctuation level corresponding to all data fluctuation levels, wherein the first standard data fluctuation level is the average of all data fluctuation levels;
[0031] Determine the maximum data fluctuation level corresponding to all data fluctuation levels, and calculate the first ratio of the first standard data fluctuation level to the maximum data fluctuation level as the first calculation factor;
[0032] Determine the second standard data volatility corresponding to all data volatility levels, where the second standard data volatility level is the variance of all data volatility levels;
[0033] A first data identifier is generated for all data fluctuations exceeding the second standard data fluctuation level;
[0034] Generate a second data identifier for all data fluctuation levels that are less than or equal to the second standard data fluctuation level;
[0035] Count the number of first identifiers of the first data identifier generated, and count the number of second identifiers of the second data identifier generated;
[0036] Determine the exponential function template, construct a first exponential function based on the first number of identifiers and the exponential function template, and calculate the value of the first exponential function. The constant of the first exponential function is the first number of identifiers, the base is the second standard data fluctuation degree, and the independent variable is the maximum data fluctuation degree corresponding to the first data identifier.
[0037] A second exponential function is constructed based on the second identifier quantity and the exponential function template, and the value of the second exponential function is calculated. The constant of the second exponential function is the second identifier quantity, the base is the second standard data fluctuation degree, and the independent variable is the maximum data fluctuation degree corresponding to the second data identifier.
[0038] Calculate a second ratio between the first exponential function value and the second exponential function value, and use it as a second calculation factor;
[0039] The sum of the first calculation factor and the second calculation factor is determined as the sub-secondary water supply temperature prediction factor of the heating station.
[0040] Furthermore, when calculating the comprehensive secondary water supply temperature prediction factor for the heating station based on all the sub-secondary water supply temperature prediction factors, it includes:
[0041] Extract the same sub-secondary water supply temperature prediction factor from all the sub-secondary water supply temperature prediction factors, and obtain multiple sets of sub-secondary water supply temperature prediction factors.
[0042] The number of the first sub-set of secondary water supply temperature prediction factors in the statistical sub-set of secondary water supply temperature prediction factors.
[0043] Extract one sub-secondary water supply temperature prediction factor from each of the sets of sub-secondary water supply temperature prediction factors, and calculate the sum of the first sub-secondary water supply temperature prediction factors;
[0044] Obtain a preset sub-secondary water supply temperature prediction factor, remove all sub-secondary water supply temperature prediction factor sets that are smaller than the preset sub-secondary water supply temperature prediction factor, and count the number of second sub-secondary water supply temperature prediction factor sets in the remaining sub-secondary water supply temperature prediction factor sets.
[0045] Extract one sub-secondary water supply temperature prediction factor from the remaining set of sub-secondary water supply temperature prediction factors, and calculate the sum of the second sub-secondary water supply temperature prediction factors.
[0046] The comprehensive secondary water supply temperature prediction factor of the heating station is calculated based on the number of the first set of secondary water supply temperature prediction factors, the number of the second set of secondary water supply temperature prediction factors, the sum of the first and second sets of secondary water supply temperature prediction factors, and the sum of the second sets of secondary water supply temperature prediction factors.
[0047] Further, when calculating the comprehensive secondary water supply temperature prediction factor of the heating station based on the number of the first sub-secondary water supply temperature prediction factor set, the number of the second sub-secondary water supply temperature prediction factor set, the sum of the first sub-secondary water supply temperature prediction factors, and the sum of the second sub-secondary water supply temperature prediction factors, the calculation includes:
[0048] The comprehensive secondary water supply temperature prediction factor for the heating station is calculated using the following formula:
[0049]
[0050] Where u is the comprehensive secondary water supply temperature prediction factor of the heating station, i1 is the number of the first sub-secondary water supply temperature prediction factor set, i2 is the number of the second sub-secondary water supply temperature prediction factor set, p1 is the sum of the first sub-secondary water supply temperature prediction factors, and p2 is the sum of the second sub-secondary water supply temperature prediction factors.
[0051] Furthermore, in collecting historical heating operation data, extracting time-related features from the historical heating operation data, and constructing a prediction model for the secondary water supply of the heating station, the following steps are included:
[0052] Extract time-related features from historical heating operation data, wherein the time-related features include time period features, historical control mode features, and heating system response delay features;
[0053] The sliding window correlation analysis method was used to determine the time-varying correlation between the correlation features at each time and the secondary water supply temperature, and the feature subset was dynamically selected.
[0054] Construct a hybrid neural network structure based on LSTM-attention mechanism, where LSTM layers are used to capture temporal dependencies and the attention mechanism is used to identify key time nodes;
[0055] The feature subset is divided according to the heating regulation period. Sub-models are trained for different periods and hybrid neural network structures. The sub-models are dynamically combined through a gating mechanism to obtain the secondary water supply prediction model of the heating station.
[0056] On the other hand, this application also provides an intelligent predictive system for heating station management, comprising:
[0057] The data acquisition module is used to receive heating management instructions from the heating station, preset a fixed sampling time period, and collect multiple sets of heating prediction impact data of the heating station based on the fixed sampling time period, wherein the fixed sampling time period includes multiple sub-fixed sampling time segments;
[0058] The data analysis module is used to reorganize the data of each group of heating forecast impact data to obtain multiple heating forecast impact data groups of the same dimension, and to determine the degree of data fluctuation of each heating forecast impact data in the same dimension heating forecast impact data group.
[0059] The factor calculation module is used to calculate the sub-secondary water supply temperature prediction factor of the heating station based on the degree of fluctuation of all data, and to calculate the comprehensive secondary water supply temperature prediction factor of the heating station based on all sub-secondary water supply temperature prediction factors.
[0060] The temperature prediction module is used to collect historical heating operation data, extract time-related features from the historical heating operation data, and construct a secondary water supply prediction model for the heating station. The comprehensive secondary water supply temperature prediction factor is input into the secondary water supply prediction model of the heating station to obtain the predicted secondary water supply temperature of the heating station.
[0061] Furthermore, it also includes:
[0062] The instruction processing module is used to obtain the previous management instruction and determine the instruction category of the previous management instruction;
[0063] The heating station management instruction is matched with the instruction category of the previous management instruction, and it is determined whether the heating station management instruction is the same as the previous management instruction.
[0064] If the heating station management instruction matches the instruction category of the previous management instruction, then it is determined that the heating station management instruction is the same as the previous management instruction. The first time point of sending the heating station management instruction is collected, and the second time point of sending the previous management instruction is collected.
[0065] Calculate the time difference between the first time node and the second time node, and determine whether the heating station heating management instruction can be received based on the relationship between the time difference and the preset time difference.
[0066] If the time node difference is greater than or equal to the preset time node difference, it is determined that the heating station heating management instruction can be received.
[0067] If the time node difference is less than the preset time node difference, it is determined that the heating station heating management instruction cannot be received, and a log reminder is generated and sent.
[0068] If the heating station management instruction does not match the instruction category of the previous management instruction, it is determined that the heating station management instruction is different from the previous management instruction, and the heating station management instruction is received.
[0069] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0070] This invention collects multiple sets of heating prediction impact data from heating stations based on a fixed sampling time period; it reorganizes each set of heating prediction impact data to obtain multiple sets of heating prediction impact data in the same dimension, and determines the data fluctuation degree of each heating prediction impact data set; it calculates a sub-secondary water supply temperature prediction factor based on the data fluctuation degree, and calculates a comprehensive secondary water supply temperature prediction factor; it extracts time-related features from historical heating operation data and constructs a secondary water supply prediction model for the heating station; it inputs the comprehensive secondary water supply temperature prediction factor into the secondary water supply prediction model for the heating station to obtain the predicted secondary water supply temperature of the heating station; by integrating time variables, it improves the prediction accuracy of secondary water supply temperature, reduces prediction errors, and reduces energy waste. Attached Figure Description
[0071] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:
[0072] Figure 1 A flowchart illustrating the intelligent prediction method for heating station management provided in an embodiment of the present invention;
[0073] Figure 2 This is a schematic diagram of the structure of the intelligent prediction system for heating station management provided in an embodiment of the present invention. Detailed Implementation
[0074] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0075] like Figure 1As shown in some embodiments of this application, this embodiment provides an intelligent prediction method for heating station management, including:
[0076] S110: Receive heating management instructions from the heating station, preset a fixed sampling time period, and collect multiple sets of heating prediction impact data of the heating station based on the fixed sampling time period, wherein the fixed sampling time period includes multiple sub-fixed sampling time segments;
[0077] In this embodiment, the fixed sampling time period is preset, such as 1 hour or 2 hours. Each fixed sampling time period includes multiple sub-fixed sampling time segments. The number of sub-fixed sampling time segments is preferably 15, such as the 1st minute, the 4th minute, the 6th minute, etc. Specifically, 15 segments are sufficient.
[0078] In this embodiment, the heating forecast impact data includes secondary return water pressure, secondary supply water pressure, secondary return water pressure, secondary heat load, indoor temperature, outdoor temperature, solar radiation equivalent value, and wind speed, etc., which are not shown one by one here.
[0079] In this embodiment, as described above, each sub-fixed sampling time segment corresponds to multiple heating prediction impact data.
[0080] In some embodiments of this application, before receiving the heating management instruction from the heating station and preset a fixed sampling time period, the following steps are also included:
[0081] Obtain the previous management instruction and determine the instruction category of the previous management instruction;
[0082] The heating station management instruction is matched with the instruction category of the previous management instruction, and it is determined whether the heating station management instruction is the same as the previous management instruction.
[0083] If the heating station management instruction matches the instruction category of the previous management instruction, then it is determined that the heating station management instruction is the same as the previous management instruction. The first time point of sending the heating station management instruction is collected, and the second time point of sending the previous management instruction is collected.
[0084] Calculate the time difference between the first time node and the second time node, and determine whether the heating station heating management instruction can be received based on the relationship between the time difference and the preset time difference.
[0085] If the time node difference is greater than or equal to the preset time node difference, it is determined that the heating station heating management instruction can be received.
[0086] If the time node difference is less than the preset time node difference, it is determined that the heating station heating management instruction cannot be received, and a log reminder is generated and sent.
[0087] If the heating station management instruction does not match the instruction category of the previous management instruction, it is determined that the heating station management instruction is different from the previous management instruction, and the heating station management instruction is received.
[0088] In this embodiment, the heating station heating management command refers to the start command for intelligent prediction of heating station heating management. When the heating station heating management command is received, the prediction of secondary water supply temperature begins.
[0089] In this embodiment, the previous management commands also include heating control commands, load regulation commands, etc.
[0090] In this embodiment, the preset time difference is preferably 5 hours.
[0091] The beneficial effects of the above technical solution are as follows: When the time node difference is less than the preset time node difference, it may be due to staff misoperation, issuing the heating station heating management instruction again within the preset time node difference. It may also not be due to staff misoperation, but a secondary water supply temperature prediction needs to be performed again within the preset time node difference. Therefore, a log reminder is generated for relevant staff to confirm, avoiding missed or repeated operations. This can ensure the normal prediction of secondary water supply temperature and avoid repeated predictions, thus avoiding waste of manpower and resources.
[0092] S120: Reorganize the data of each group of heating forecast impact data to obtain multiple heating forecast impact data groups of the same dimension, and determine the degree of data fluctuation of each heating forecast impact data in the same dimension heating forecast impact data group;
[0093] In this embodiment, when reorganizing the data for each group of heating forecast impact data, it means combining all heating forecast impact data of the same type. For example, as mentioned above, the secondary return water pressure corresponding to each sub-fixed sampling time segment is combined to obtain a heating forecast impact data group of the same dimension regarding the secondary return water pressure. Similarly, the secondary supply water pressure corresponding to each sub-fixed sampling time segment is combined to obtain a heating forecast impact data group of the same dimension regarding the secondary supply water pressure.
[0094] In some embodiments of this application, determining the degree of data fluctuation for each heating forecast impact data in the same-dimensional heating forecast impact data group includes:
[0095] Based on the sub-fixed sampling time segment, the heating prediction impact data group of the same dimension is sorted in time order, and the initial heating prediction impact data and the final heating prediction impact data are determined.
[0096] One heating prediction impact data point is randomly extracted from the same-dimensional heating prediction impact data group and used as the baseline heating prediction impact data.
[0097] Calculate the initial data difference between the baseline heating forecast impact data and the initial heating forecast impact data, wherein the initial data difference is the absolute value of the difference between the baseline heating forecast impact data and the initial heating forecast impact data;
[0098] Calculate the difference between the baseline heating forecast impact data and the final heating forecast impact data, wherein the difference between the final data is the absolute value of the difference between the baseline heating forecast impact data and the final heating forecast impact data.
[0099] The degree of data fluctuation in the baseline heating forecast impact data is calculated based on the initial data difference and the final data difference.
[0100] The remaining heating forecast impact data in the same dimension heating forecast impact data group are extracted and calculated to obtain the corresponding data fluctuation degree.
[0101] In this embodiment, each heating forecast impact data corresponds to a sub-fixed sampling time segment. Therefore, the time order can be sorted based on the sub-fixed sampling time segment.
[0102] In this embodiment, by repeating the above steps, the degree of data fluctuation corresponding to each heating forecast impact data can be obtained.
[0103] The beneficial effects of the above technical solution are: the present invention calculates the degree of data fluctuation of the baseline heating forecast impact data based on the difference between the initial data and the difference between the final data. The degree of data fluctuation can characterize the fluctuation of each heating forecast impact data relative to all heating forecast impact data, thereby ensuring the accuracy of intelligent prediction of heating management at the heating station.
[0104] In some embodiments of this application, when calculating the degree of data fluctuation in the baseline heating forecast impact data based on the initial data difference and the final data difference, the method includes:
[0105] The degree of data fluctuation affecting the baseline heating forecast data is calculated using the following formula:
[0106]
[0107] Where q represents the data fluctuation degree of the baseline heating forecast impact data, w1 represents the initial data difference, w2 represents the final data difference, r represents the number of heating forecast impact data in the same dimension heating forecast impact data group, and y t+1 For the (t+1)th heating forecast impact data in the same dimension heating forecast impact data group, y t This is the t-th heating forecast impact data in the same dimension heating forecast impact data group.
[0108] S130: Calculate the sub-secondary water supply temperature prediction factor of the heating station based on the degree of fluctuation of all data, and calculate the comprehensive secondary water supply temperature prediction factor of the heating station based on all sub-secondary water supply temperature prediction factors.
[0109] In some embodiments of this application, the calculation of the sub-secondary water supply temperature prediction factor for the heating station based on all data fluctuation levels includes:
[0110] Determine the first standard data fluctuation level corresponding to all data fluctuation levels, wherein the first standard data fluctuation level is the average of all data fluctuation levels;
[0111] Determine the maximum data fluctuation level corresponding to all data fluctuation levels, and calculate the first ratio of the first standard data fluctuation level to the maximum data fluctuation level as the first calculation factor;
[0112] Determine the second standard data volatility corresponding to all data volatility levels, where the second standard data volatility level is the variance of all data volatility levels;
[0113] A first data identifier is generated for all data fluctuations exceeding the second standard data fluctuation level;
[0114] Generate a second data identifier for all data fluctuation levels that are less than or equal to the second standard data fluctuation level;
[0115] Count the number of first identifiers of the first data identifier generated, and count the number of second identifiers of the second data identifier generated;
[0116] Determine the exponential function template, construct a first exponential function based on the first number of identifiers and the exponential function template, and calculate the value of the first exponential function. The constant of the first exponential function is the first number of identifiers, the base is the second standard data fluctuation degree, and the independent variable is the maximum data fluctuation degree corresponding to the first data identifier.
[0117] A second exponential function is constructed based on the second identifier quantity and the exponential function template, and the value of the second exponential function is calculated. The constant of the second exponential function is the second identifier quantity, the base is the second standard data fluctuation degree, and the independent variable is the maximum data fluctuation degree corresponding to the second data identifier.
[0118] Calculate a second ratio between the first exponential function value and the second exponential function value, and use it as a second calculation factor;
[0119] The sum of the first calculation factor and the second calculation factor is determined as the sub-secondary water supply temperature prediction factor of the heating station.
[0120] In this embodiment, the template for the exponential function is f(x) = ab x , where a is a constant, b is the base, and x is the independent variable.
[0121] In this embodiment, by substituting the corresponding constants, bases, and independent variables into the above formula, the first exponential function and the second exponential function can be obtained. The values of the first exponential function and the second exponential function can then be calculated.
[0122] The beneficial effects of the above technical solution are: the present invention determines the sum of the first calculation factor and the second calculation factor as the sub-secondary water supply temperature prediction factor of the heating station. The sub-secondary water supply temperature prediction factor can characterize the influence of the secondary water supply temperature of the heating station on a data dimension, further ensuring the accuracy and comprehensiveness of intelligent prediction of heating station heating management.
[0123] In some embodiments of this application, calculating the comprehensive secondary water supply temperature prediction factor of the heating station based on all sub-secondary water supply temperature prediction factors includes:
[0124] Extract the same sub-secondary water supply temperature prediction factor from all the sub-secondary water supply temperature prediction factors, and obtain multiple sets of sub-secondary water supply temperature prediction factors.
[0125] The number of the first sub-set of secondary water supply temperature prediction factors in the statistical sub-set of secondary water supply temperature prediction factors.
[0126] Extract one sub-secondary water supply temperature prediction factor from each of the sets of sub-secondary water supply temperature prediction factors, and calculate the sum of the first sub-secondary water supply temperature prediction factors;
[0127] Obtain a preset sub-secondary water supply temperature prediction factor, remove all sub-secondary water supply temperature prediction factor sets that are smaller than the preset sub-secondary water supply temperature prediction factor, and count the number of second sub-secondary water supply temperature prediction factor sets in the remaining sub-secondary water supply temperature prediction factor sets.
[0128] Extract one sub-secondary water supply temperature prediction factor from the remaining set of sub-secondary water supply temperature prediction factors, and calculate the sum of the second sub-secondary water supply temperature prediction factors.
[0129] The comprehensive secondary water supply temperature prediction factor of the heating station is calculated based on the number of the first set of secondary water supply temperature prediction factors, the number of the second set of secondary water supply temperature prediction factors, the sum of the first and second sets of secondary water supply temperature prediction factors, and the sum of the second sets of secondary water supply temperature prediction factors.
[0130] In this embodiment, the preset sub-secondary water supply temperature prediction factor is the variance corresponding to all sub-secondary water supply temperature prediction factors.
[0131] The beneficial effects of the above technical solution are as follows: This invention calculates the comprehensive secondary water supply temperature prediction factor of the heating station based on the number of the first sub-secondary water supply temperature prediction factor set, the number of the second sub-secondary water supply temperature prediction factor set, the sum of the first sub-secondary water supply temperature prediction factors and the sum of the second sub-secondary water supply temperature prediction factors, thereby realizing the fusion of time variables and achieving multi-dimensional data analysis, ensuring the accuracy of the comprehensive secondary water supply temperature prediction factor, and laying the foundation for secondary water supply temperature prediction.
[0132] In some embodiments of this application, when calculating the comprehensive secondary water supply temperature prediction factor of the heating station based on the number of the first sub-secondary water supply temperature prediction factor set, the number of the second sub-secondary water supply temperature prediction factor set, the sum of the first sub-secondary water supply temperature prediction factors and the sum of the second sub-secondary water supply temperature prediction factors, the calculation includes:
[0133] The comprehensive secondary water supply temperature prediction factor for the heating station is calculated using the following formula:
[0134]
[0135] Where u is the comprehensive secondary water supply temperature prediction factor of the heating station, i1 is the number of the first sub-secondary water supply temperature prediction factor set, i2 is the number of the second sub-secondary water supply temperature prediction factor set, p1 is the sum of the first sub-secondary water supply temperature prediction factors, and p2 is the sum of the second sub-secondary water supply temperature prediction factors.
[0136] S140: Collect historical heating operation data, extract time-related features from the historical heating operation data, and construct a secondary water supply prediction model for the heating station. Input the comprehensive secondary water supply temperature prediction factor into the secondary water supply prediction model for the heating station to obtain the predicted secondary water supply temperature of the heating station.
[0137] In some embodiments of this application, the process of collecting historical heating operation data, extracting time-related features from the historical heating operation data, and constructing a secondary water supply prediction model for the heating station includes:
[0138] Extract time-related features from historical heating operation data, wherein the time-related features include time period features, historical control mode features, and heating system response delay features;
[0139] The sliding window correlation analysis method was used to determine the time-varying correlation between the correlation features at each time and the secondary water supply temperature, and the feature subset was dynamically selected.
[0140] Construct a hybrid neural network structure based on LSTM-attention mechanism, where LSTM layers are used to capture temporal dependencies and the attention mechanism is used to identify key time nodes;
[0141] The feature subset is divided according to the heating regulation period. Sub-models are trained for different periods and hybrid neural network structures. The sub-models are dynamically combined through a gating mechanism to obtain the secondary water supply prediction model of the heating station.
[0142] In this embodiment, the time period feature is as follows: the day is divided into multiple heating control periods, such as morning peak, midday trough, and evening peak, and each period is encoded as a unique vector or sine / cosine periodic code.
[0143] In this embodiment, the historical control mode features are: extracting control parameters from the same time period over the past N days, such as valve opening and pump frequency statistics.
[0144] In this embodiment, the response delay characteristic of the heating system is: the time delay from the issuance of the control command to the actual change in the secondary water supply temperature is calculated.
[0145] In this embodiment, the sliding window design is as follows:
[0146] Window size: Set according to the inertial characteristics of the heating network, usually 3-6 heating adjustment cycles.
[0147] Step size: The window sliding step size is consistent with the prediction frequency, such as 15 minutes.
[0148] Time-varying correlation calculation: combining Pearson correlation coefficient (linear relationship) and maximum information coefficient (MIC, non-linear relationship).
[0149] In this embodiment, the feature subset is a combination of high-value predictive variables dynamically selected from the original features through sliding window correlation analysis.
[0150] In this embodiment, the LSTM layer is designed as follows:
[0151] Structure: Double-layer stacked LSTM, the number of hidden layer neurons is set according to the complexity of the heating system (usually 32-64).
[0152] Input: Standardized time series data, in the shape of (batch, time step, number of features).
[0153] Attention mechanism:
[0154] Function: Automatically identify key time points (such as data in the first 10 minutes after valve adjustment).
[0155] In this embodiment, the gating mechanism is implemented as follows: a fully connected network, with time-segment codes as input and weights of each sub-model as output.
[0156] In this embodiment, the comprehensive secondary water supply temperature prediction factor is used as a single input feature and input into the secondary water supply prediction model of the heating station. The secondary water supply prediction model of the heating station outputs the predicted secondary water supply temperature of the heating station through linear calculation.
[0157] The beneficial effects of the above technical solution are: by training the secondary water supply prediction model of the heating station, the present invention can provide reliable technical support for the intelligent prediction of the heating station. The predicted secondary water supply temperature of the heating station is obtained through the secondary water supply prediction model, which effectively ensures the prediction accuracy and efficiency of the secondary water supply temperature, eliminates prediction errors, and comprehensively considers time variables.
[0158] like Figure 2 As shown, in another preferred embodiment based on the above embodiments, this implementation provides an intelligent predictive system for heating station management, including:
[0159] The data acquisition module is used to receive heating management instructions from the heating station, preset a fixed sampling time period, and collect multiple sets of heating prediction impact data of the heating station based on the fixed sampling time period, wherein the fixed sampling time period includes multiple sub-fixed sampling time segments;
[0160] The data analysis module is used to reorganize the data of each group of heating forecast impact data to obtain multiple heating forecast impact data groups of the same dimension, and to determine the degree of data fluctuation of each heating forecast impact data in the same dimension heating forecast impact data group.
[0161] The factor calculation module is used to calculate the sub-secondary water supply temperature prediction factor of the heating station based on the degree of fluctuation of all data, and to calculate the comprehensive secondary water supply temperature prediction factor of the heating station based on all sub-secondary water supply temperature prediction factors.
[0162] The temperature prediction module is used to collect historical heating operation data, extract time-related features from the historical heating operation data, and construct a secondary water supply prediction model for the heating station. The comprehensive secondary water supply temperature prediction factor is input into the secondary water supply prediction model of the heating station to obtain the predicted secondary water supply temperature of the heating station.
[0163] In some embodiments of this application, it also includes:
[0164] The instruction processing module is used to obtain the previous management instruction and determine the instruction category of the previous management instruction;
[0165] The heating station management instruction is matched with the instruction category of the previous management instruction, and it is determined whether the heating station management instruction is the same as the previous management instruction.
[0166] If the heating station management instruction matches the instruction category of the previous management instruction, then it is determined that the heating station management instruction is the same as the previous management instruction. The first time point of sending the heating station management instruction is collected, and the second time point of sending the previous management instruction is collected.
[0167] Calculate the time difference between the first time node and the second time node, and determine whether the heating station heating management instruction can be received based on the relationship between the time difference and the preset time difference.
[0168] If the time node difference is greater than or equal to the preset time node difference, it is determined that the heating station heating management instruction can be received.
[0169] If the time node difference is less than the preset time node difference, it is determined that the heating station heating management instruction cannot be received, and a log reminder is generated and sent.
[0170] If the heating station management instruction does not match the instruction category of the previous management instruction, it is determined that the heating station management instruction is different from the previous management instruction, and the heating station management instruction is received.
[0171] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0172] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0173] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0174] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0175] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention.
Claims
1. A method for intelligent prediction of heating management in a heat station, characterized in that, include: Receive heating management instructions from the heating station, preset a fixed sampling time period, and collect multiple sets of heating prediction impact data of the heating station based on the fixed sampling time period, wherein the fixed sampling time period includes multiple sub-fixed sampling time segments; Each set of heating forecast impact data is reorganized to obtain multiple heating forecast impact data groups of the same dimension, and the degree of data fluctuation of each heating forecast impact data in the same dimension heating forecast impact data group is determined. The sub-secondary water supply temperature prediction factor of the heating station is calculated based on the degree of fluctuation of all data, and the comprehensive secondary water supply temperature prediction factor of the heating station is calculated based on all sub-secondary water supply temperature prediction factors. Historical heating operation data is collected, time-related features are extracted from the historical heating operation data, and a secondary water supply prediction model for the heating station is constructed. The comprehensive secondary water supply temperature prediction factor is input into the secondary water supply prediction model of the heating station to obtain the predicted secondary water supply temperature of the heating station. When determining the degree of data fluctuation for each heating forecast impact data in the same-dimensional heating forecast impact data group, the following is included: Based on the sub-fixed sampling time segment, the heating prediction impact data group of the same dimension is sorted in time order, and the initial heating prediction impact data and the final heating prediction impact data are determined. One heating prediction impact data point is randomly extracted from the same-dimensional heating prediction impact data group and used as the baseline heating prediction impact data. Calculate the initial data difference between the baseline heating forecast impact data and the initial heating forecast impact data, wherein the initial data difference is the absolute value of the difference between the baseline heating forecast impact data and the initial heating forecast impact data; Calculate the difference between the baseline heating forecast impact data and the final heating forecast impact data, wherein the difference between the final data is the absolute value of the difference between the baseline heating forecast impact data and the final heating forecast impact data. The degree of data fluctuation in the baseline heating forecast impact data is calculated based on the initial data difference and the final data difference. The remaining heating forecast impact data in the same dimension heating forecast impact data group are extracted and calculated to obtain the corresponding data fluctuation degree.
2. The intelligent prediction method for heating station management according to claim 1, characterized in that, Before receiving heating management instructions from the heating station and setting a fixed sampling time period, the process also includes: Obtain the previous management instruction and determine the instruction category of the previous management instruction; The heating station management instruction is matched with the instruction category of the previous management instruction, and it is determined whether the heating station management instruction is the same as the previous management instruction. If the heating station management instruction matches the instruction category of the previous management instruction, then it is determined that the heating station management instruction is the same as the previous management instruction. The first time point of sending the heating station management instruction is collected, and the second time point of sending the previous management instruction is collected. Calculate the time difference between the first time node and the second time node, and determine whether the heating station heating management instruction can be received based on the relationship between the time difference and the preset time difference. If the time node difference is greater than or equal to the preset time node difference, it is determined that the heating station heating management instruction can be received. If the time node difference is less than the preset time node difference, it is determined that the heating station heating management instruction cannot be received, and a log reminder is generated and sent. If the heating station management instruction does not match the instruction category of the previous management instruction, it is determined that the heating station management instruction is different from the previous management instruction, and the heating station management instruction is received.
3. The intelligent prediction method for heating station management according to claim 1, characterized in that, When calculating the degree of data fluctuation in the baseline heating forecast impact data based on the initial data difference and the final data difference, the following is included: The degree of data fluctuation affecting the baseline heating forecast data is calculated using the following formula: ； Where q represents the data fluctuation degree of the baseline heating forecast impact data, w1 represents the initial data difference, w2 represents the final data difference, r represents the number of heating forecast impact data in the same dimension heating forecast impact data group, and y t+1 For the (t+1)th heating forecast impact data in the same dimension heating forecast impact data group, y t This is the t-th heating forecast impact data in the same dimension heating forecast impact data group.
4. The intelligent prediction method for heating station management according to claim 1, characterized in that, When calculating the sub-secondary water supply temperature prediction factor for the heating station based on the degree of fluctuation of all data, the following is included: Determine the first standard data fluctuation level corresponding to all data fluctuation levels, wherein the first standard data fluctuation level is the average of all data fluctuation levels; Determine the maximum data fluctuation level corresponding to all data fluctuation levels, and calculate the first ratio of the first standard data fluctuation level to the maximum data fluctuation level as the first calculation factor; Determine the second standard data volatility corresponding to all data volatility levels, where the second standard data volatility level is the variance of all data volatility levels; A first data identifier is generated for all data fluctuations exceeding the second standard data fluctuation level; Generate a second data identifier for all data fluctuation levels that are less than or equal to the second standard data fluctuation level; Count the number of first identifiers of the first data identifier generated, and count the number of second identifiers of the second data identifier generated; Determine the exponential function template, construct a first exponential function based on the first number of identifiers and the exponential function template, and calculate the value of the first exponential function. The constant of the first exponential function is the first number of identifiers, the base is the second standard data fluctuation degree, and the independent variable is the maximum data fluctuation degree corresponding to the first data identifier. A second exponential function is constructed based on the second identifier quantity and the exponential function template, and the value of the second exponential function is calculated. The constant of the second exponential function is the second identifier quantity, the base is the second standard data fluctuation degree, and the independent variable is the maximum data fluctuation degree corresponding to the second data identifier. Calculate a second ratio between the first exponential function value and the second exponential function value, and use it as a second calculation factor; The sum of the first calculation factor and the second calculation factor is determined as the sub-secondary water supply temperature prediction factor of the heating station.
5. The intelligent prediction method for heating station management according to claim 1, characterized in that, When calculating the comprehensive secondary water supply temperature prediction factor for the heating station based on all sub-secondary water supply temperature prediction factors, the following is included: Extract the same sub-secondary water supply temperature prediction factor from all the sub-secondary water supply temperature prediction factors, and obtain multiple sets of sub-secondary water supply temperature prediction factors. The number of the first sub-set of secondary water supply temperature prediction factors in the statistical sub-set of secondary water supply temperature prediction factors. Extract one sub-secondary water supply temperature prediction factor from each of the sets of sub-secondary water supply temperature prediction factors, and calculate the sum of the first sub-secondary water supply temperature prediction factors; Obtain a preset sub-secondary water supply temperature prediction factor, remove all sub-secondary water supply temperature prediction factor sets that are smaller than the preset sub-secondary water supply temperature prediction factor, and count the number of second sub-secondary water supply temperature prediction factor sets in the remaining sub-secondary water supply temperature prediction factor sets. Extract one sub-secondary water supply temperature prediction factor from the remaining set of sub-secondary water supply temperature prediction factors, and calculate the sum of the second sub-secondary water supply temperature prediction factors. The comprehensive secondary water supply temperature prediction factor of the heating station is calculated based on the number of the first set of secondary water supply temperature prediction factors, the number of the second set of secondary water supply temperature prediction factors, the sum of the first and second sets of secondary water supply temperature prediction factors, and the sum of the second sets of secondary water supply temperature prediction factors.
6. The intelligent prediction method for heating station management according to claim 5, characterized in that, When calculating the comprehensive secondary water supply temperature prediction factor of the heating station based on the number of the first sub-secondary water supply temperature prediction factor set, the number of the second sub-secondary water supply temperature prediction factor set, the sum of the first and second sub-secondary water supply temperature prediction factors, and the sum of the second sub-secondary water supply temperature prediction factors, the calculation includes: The comprehensive secondary water supply temperature prediction factor for the heating station is calculated using the following formula: ； Where u is the comprehensive secondary water supply temperature prediction factor of the heating station, i1 is the number of the first sub-secondary water supply temperature prediction factor set, i2 is the number of the second sub-secondary water supply temperature prediction factor set, p1 is the sum of the first sub-secondary water supply temperature prediction factors, and p2 is the sum of the second sub-secondary water supply temperature prediction factors.
7. The intelligent prediction method for heating station management according to claim 1, characterized in that, When collecting historical heating operation data, extracting time-related features from the historical heating operation data, and constructing a prediction model for the secondary water supply of the heating station, the following steps are included: Extract time-related features from historical heating operation data, wherein the time-related features include time period features, historical control mode features, and heating system response delay features; The sliding window correlation analysis method was used to determine the time-varying correlation between the correlation features at each time and the secondary water supply temperature, and the feature subset was dynamically selected. Construct a hybrid neural network structure based on LSTM-attention mechanism, where LSTM layers are used to capture temporal dependencies and the attention mechanism is used to identify key time nodes; The feature subset is divided according to the heating regulation period. Sub-models are trained for different periods and hybrid neural network structures. The sub-models are dynamically combined through a gating mechanism to obtain the secondary water supply prediction model of the heating station.
8. A heating station heating management intelligent prediction system, applied to the heating station heating management intelligent prediction method as described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to receive heating management instructions from the heating station, preset a fixed sampling time period, and collect multiple sets of heating prediction impact data of the heating station based on the fixed sampling time period, wherein the fixed sampling time period includes multiple sub-fixed sampling time segments; The data analysis module is used to reorganize the data of each group of heating forecast impact data to obtain multiple heating forecast impact data groups of the same dimension, and to determine the degree of data fluctuation of each heating forecast impact data in the same dimension heating forecast impact data group. The factor calculation module is used to calculate the sub-secondary water supply temperature prediction factor of the heating station based on the degree of fluctuation of all data, and to calculate the comprehensive secondary water supply temperature prediction factor of the heating station based on all sub-secondary water supply temperature prediction factors. The temperature prediction module is used to collect historical heating operation data, extract time-related features from the historical heating operation data, and construct a secondary water supply prediction model for the heating station. The comprehensive secondary water supply temperature prediction factor is input into the secondary water supply prediction model of the heating station to obtain the predicted secondary water supply temperature of the heating station.
9. The intelligent predictive system for heating station management according to claim 8, characterized in that, Also includes: The instruction processing module is used to obtain the previous management instruction and determine the instruction category of the previous management instruction; The heating station management instruction is matched with the instruction category of the previous management instruction, and it is determined whether the heating station management instruction is the same as the previous management instruction. If the heating station management instruction matches the instruction category of the previous management instruction, then it is determined that the heating station management instruction is the same as the previous management instruction. The first time point of sending the heating station management instruction is collected, and the second time point of sending the previous management instruction is collected. Calculate the time difference between the first time node and the second time node, and determine whether the heating station heating management instruction can be received based on the relationship between the time difference and the preset time difference. If the time node difference is greater than or equal to the preset time node difference, it is determined that the heating station heating management instruction can be received. If the time node difference is less than the preset time node difference, it is determined that the heating station heating management instruction cannot be received, and a log reminder is generated and sent. If the heating station management instruction does not match the instruction category of the previous management instruction, it is determined that the heating station management instruction is different from the previous management instruction, and the heating station management instruction is received.