Heating estimation management method and management system based on artificial intelligence

Abstract

The present application provides a heating prediction management method and management system based on artificial intelligence. Based on the heat consumption behavior parameters of the object to be estimated in the heating period P2 and the heat consumption behavior parameters in the heating period P1, the estimated heat usage behavior standard coefficient of the object to be estimated in the heating period P1 is estimated, and the heat usage behavior standard coefficient of the object to be estimated in the heating period P2 is obtained. Based on the estimated heat usage behavior standard coefficient in the heating period P1 and the heat usage behavior standard coefficient in the heating period P2, the heat usage behavior standard coefficient of the object to be estimated in the heating period P1 is determined. Through the target heat usage behavior standard coefficient in the heating period P1, the heating scheduling method for the use of heating energy by the object to be estimated in the heating period P1 can be determined.

Description

Technical Field

[0001] The present application relates to the field of data processing, and more specifically, to a heating estimation management method and management system based on artificial intelligence. Background Art

[0002] With the increasing awareness of energy conservation and environmental protection, intelligent management of heating systems has become an important means to improve energy efficiency and reduce energy consumption. Traditional heating management methods often rely on fixed heating strategies and cannot be dynamically adjusted according to users' actual heating usage behaviors, resulting in energy waste and inefficient management. In order to overcome the shortcomings of the above-mentioned traditional heating management methods, the application of artificial intelligence technology in the field of heating management has gradually attracted attention in recent years. Artificial intelligence technology, especially machine learning algorithms and neural network models, has powerful data analysis and pattern recognition capabilities. It can learn users' heating usage behavior patterns based on a large amount of historical data, and predict future heating demand and the degree of behavioral norms based on these patterns.

[0003] However, there are still some challenges in the existing AI-based heating management methods. On the one hand, how to accurately obtain and evaluate the degree of standardization of users' heating usage behavior is a key issue. Traditional evaluation methods are often based on fixed indicators and standards, which cannot fully reflect the diversity and variability of user behavior. On the other hand, how to dynamically adjust the heating strategy and pricing mechanism according to the degree of standardization of users' heating usage behavior is also a challenge. Traditional heating management systems often adopt fixed strategies and mechanisms, which cannot effectively incentivize and constrain users according to their actual behavior. Summary of the invention

[0004] In view of this, the embodiments of the present application at least provide a heating estimation management method and management system based on artificial intelligence.

[0005] According to one aspect of an embodiment of the present application, there is provided a heating estimation management method based on artificial intelligence, comprising: obtaining heat consumption behavior parameters of an object to be estimated in a heating period P1, and heat consumption behavior parameters in a heating period P2, wherein the heating period P2 is the heating period before the heating period P1; estimating an estimated heat usage behavior standard coefficient of the object to be estimated in the heating period P1 based on the heat consumption behavior parameters in the heating period P1 and the heat consumption behavior parameters in the heating period P2; obtaining the heat usage behavior standard coefficient of the object to be estimated in the heating period P2, and determining the heat usage behavior standard coefficient of the object to be estimated based on the heat usage behavior standard coefficient in the heating period P2 and the estimated heat usage behavior standard coefficient in the heating period P1. The heat usage behavior standard coefficient of the estimated object in the heating period P1; wherein the heat usage behavior standard coefficient in the heating period P1 is configured to determine the heating scheduling method for the use of heating energy by the object to be estimated in the heating period P1; the heat usage behavior standard coefficient in the heating period P1 is configured to determine the heat usage behavior standard coefficient of the object to be estimated in the heating period P3 together with the estimated heat usage behavior standard coefficient in the heating period P3, and the heating period P3 is the period after the heating period P1; the estimated heat usage behavior standard coefficient in the heating period P3 is determined by the heat consumption behavior parameters of the object to be estimated in the heating period P3 and the heat consumption behavior parameters in the heating period P1.

[0006] According to another aspect of an embodiment of the present application, a heating forecast management system is provided, comprising: a processor; and a memory, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the method described above.

[0007] The beneficial effects of the present application include at least: the present application provides a heating prediction management method and management system based on artificial intelligence, which estimates the heat usage behavior standard coefficient of the object to be estimated in the heating period P1 based on the heat consumption behavior parameters of the object to be estimated in the heating period P2 and the heat consumption behavior parameters in the heating period P1, obtains the heat usage behavior standard coefficient of the object to be estimated in the heating period P2, and determines the heat usage behavior standard coefficient of the object to be estimated in the heating period P1 based on the estimated heat usage behavior standard coefficient in the heating period P1 and the heat usage behavior standard coefficient in the heating period P2; and through the target heat usage behavior standard coefficient in the heating period P1, the heating scheduling method for the use of heating energy by the object to be estimated in the heating period P1 can be determined. It can be understood that in the use of heating energy, the present application can accurately measure the heating standard coefficient of the object to be estimated in the current period through the heat consumption behavior parameters of the object to be estimated in the previous heating period and the heat consumption behavior parameters of the current period, and obtain an accurate estimated heat usage behavior standard coefficient, and then obtain the accurate heat usage behavior standard coefficient of the object to be estimated in the current period through the heat usage behavior standard coefficient in the previous heating period and the estimated heat usage behavior standard coefficient in the current period, and then the heat usage behavior standard coefficient can be used to determine the heating scheduling method of the object to be estimated for the use of heating energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A schematic diagram of an application scenario provided for an embodiment of the present application.

[0009] Figure 2 A schematic diagram of the implementation flow of a heating estimation management method based on artificial intelligence provided in an embodiment of the present application.

[0010] Figure 3 A schematic diagram of the hardware entity of a heat supply prediction management system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0011] The heat supply estimation management method based on artificial intelligence provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 (i.e., the device to be predicted) communicates with the heat supply prediction management system 104 through the network. The data storage system can store the data that the heat supply prediction management system 104 needs to process. The data storage system can be integrated on the heat supply prediction management system 104, or it can be placed on the cloud or other network servers. Among them, the terminal 102 can be, but is not limited to, an intelligent thermostat or a heating system control terminal.

[0012] The artificial intelligence-based heat supply estimation management method provided in the embodiment of the present application is applied to a heat supply prediction management system, and specifically includes the following steps: Step S10: Obtain the heat consumption behavior parameters of the object to be estimated in the heating period P1 and the heat consumption behavior parameters in the heating period P2, where the heating period P2 is the heating period before the heating period P1.

[0013] The object to be estimated is, for example, a specific heating user, such as a residential household or a commercial building. Two key heating periods are determined: P1 and P2. P1 is the current or upcoming heating period, and P2 is the previous heating period of P1. For example, assuming that the P1 period is from December 1 to December 15, 2023, then the P2 period is the previous heating cycle, such as November 16 to November 30, 2023. After determining these two heating periods, collect the heat consumption behavior parameters of the object to be estimated during these two periods. These parameters include but are not limited to the operating time of the heating equipment, energy consumption, set temperature, indoor temperature changes, etc.

[0014] The heat supply forecast management system communicates with the interfaces of heating equipment, smart meters, heat meters, thermostats and other related equipment to collect these heat consumption behavior parameters in real time or regularly. These parameters will be stored in the database of the heat supply forecast management system in the form of data for subsequent analysis and processing.

[0015] Step S20: estimating the estimated heat consumption behavior standard coefficient of the object to be estimated in the heating period P1 based on the heat consumption behavior parameter in the heating period P1 and the heat consumption behavior parameter in the heating period P2.

[0016] In order to estimate the heat usage behavior standard coefficient in the P1 period, the heating prediction management system uses a pre-trained machine learning model - a neural network is used as an example here. The neural network model is trained on the basis of a large amount of historical heating data, and learns to extract key features from the heat consumption behavior parameters, and evaluates whether the user's heat usage behavior is standardized based on these features. In step S20, the heating prediction management system converts the heat consumption behavior parameters in the P1 period into a feature vector. This feature vector contains key information that can reflect the user's heat usage behavior, such as the average value of the equipment running time, the rate of change of energy consumption, the stability of the set temperature, etc. Step S20 does not directly involve the data conversion of the P2 period, but during the model training process, the data of the P2 period is also used to extract features and train the model. Then, this feature vector is input into the trained neural network model. Based on its internal weights and bias parameters, the neural network model performs a series of complex nonlinear transformations and calculations on the input feature vector, and finally outputs a value between 0 and 1, that is, the estimated heat usage behavior standard coefficient. The higher this coefficient is, the more standardized the user's heat usage behavior in the P1 period is.

[0017] For example, if the device running time of the estimated object is stable, the energy consumption is moderate, and the set temperature changes little during the P1 period, then these characteristics will be reflected in the feature vector accordingly. When this feature vector is input into the neural network model, an estimated heat usage behavior standard coefficient close to 1 is output, indicating that the user's heat usage behavior is very standard. On the contrary, if the user's device running time is too long, the energy consumption is too high, or the set temperature changes greatly during the P1 period, then the estimated heat usage behavior standard coefficient may be low, reflecting that the user's heat usage behavior during this period is not standard.

[0018] Through step S20, the heat supply prediction management system can accurately estimate the degree of standardization of the user's heat use behavior in the current heating period based on the user's heat consumption behavior parameters in two consecutive heating periods, providing an important basis for the subsequent formulation of personalized heat scheduling methods.

[0019] Step S30: Obtain the heat usage behavior standard coefficient of the object to be estimated in the heating period P2, and determine the heat usage behavior standard coefficient of the object to be estimated in the heating period P1 based on the heat usage behavior standard coefficient in the heating period P2 and the estimated heat usage behavior standard coefficient in the heating period P1.

[0020] Among them, the heat usage behavior normative coefficient in the heating period P1 is configured to determine the heating scheduling method for the use of heating energy by the object to be estimated in the heating period P1; the heat usage behavior normative coefficient in the heating period P1 is configured to determine the heat usage behavior normative coefficient of the object to be estimated in the heating period P3 together with the estimated heat usage behavior normative coefficient in the heating period P3, and the heating period P3 is the period after the heating period P1; the estimated heat usage behavior normative coefficient in the heating period P3 is determined by the heat consumption behavior parameters of the object to be estimated in the heating period P3 and the heat consumption behavior parameters in the heating period P1.

[0021] Step S30 comprehensively determines the final heat usage behavior normative coefficient of the object to be estimated in the heating period P1 based on the actual heat usage behavior normative coefficient of the object to be estimated in the heating period P2 and the estimated heat usage behavior normative coefficient in the heating period P1. First, the heating prediction management system has estimated the heat usage behavior normative coefficient of the object to be estimated in the heating period P1 according to step S20. At the same time, the actual heat usage behavior normative coefficient of the object to be estimated in the heating period P2 is also obtained. This actual normative coefficient may be calculated based on the heat consumption behavior parameters of the user in the P2 period and the heating management specifications. In step S30, these two normative coefficients are combined to determine the final heat usage behavior normative coefficient of the object to be estimated in the heating period P1. For example, it is achieved through a weighted or fusion mechanism to ensure that both coefficients have an impact on the final result. For example, the heating forecast management system can assign a higher weight to the estimated specification coefficient of the P1 period, because it is based on the latest heat consumption data and can better reflect the user's current heating behavior characteristics; at the same time, it also assigns a certain weight to the actual specification coefficient of the P2 period to take into account the user's historical behavior patterns and the continuity of the heating management policy.

[0022] Specifically, it is assumed that the estimated heat usage behavior standard coefficient of the object to be estimated in the heating period P1 is 0.8 (indicating that it is relatively standardized), and the actual heat usage behavior standard coefficient in the heating period P2 is 0.7 (indicating that it is relatively less standardized). In step S30, a weighted average method may be used to determine the final standard coefficient of the P1 period, such as assigning a weight of 0.6 to the estimated coefficient and a weight of 0.4 to the actual coefficient. In this way, the final calculated standard coefficient of the P1 period may be 0.76, which reflects the user's relatively standardized heating behavior in the P1 period and also takes into account the user's relatively less standardized historical behavior in the P2 period.

[0023] This finally determined heat usage behavior standard coefficient will be used to determine the heating scheduling method for the use of heating energy by the object to be estimated in the heating period P1, including specific parameters such as how to heat, heating duration, heating temperature, and corresponding pricing rules. In this way, the heating management system can achieve precise regulation and optimization of the user's heating behavior, improve heating efficiency, and reduce energy waste. In addition, it is worth noting that the heat usage behavior standard coefficient in the heating period P1 is not only used for the heating scheduling decision of the current period, but also for the estimation and scheduling of subsequent heating periods. For example, in the heating period P3 (i.e., the period after P1), the heating prediction management system will combine the heat consumption behavior parameters of the object to be estimated in the P3 period and the heat consumption behavior parameters in the P1 period to estimate the heat usage behavior standard coefficient in the P3 period. This estimation process will again use the machine learning model and method described in step S20. In this way, continuous monitoring and dynamic adjustment of user heating behavior are achieved to ensure the scientificity and effectiveness of heating management.

[0024] As an implementation mode, step S20, based on the heat consumption behavior parameter in the heating period P1 and the heat consumption behavior parameter in the heating period P2, estimates the estimated heat usage behavior standard coefficient of the object to be estimated in the heating period P1, which may specifically include: Step S21: The heat consumption behavior parameters in the heating period P1 are passed into the heat usage behavior recognition network; the heat usage behavior recognition network is obtained by optimizing the heat usage behavior recognition network to be trained through the heat consumption behavior parameters in the heating period P2 and the normative coefficient mark of the behavior that has occurred in the heating period P2; the normative coefficient mark of the behavior that has occurred in the heating period P2 is determined by the statistics of the irregular behavior that has occurred in the heat consumption behavior parameters in the heating period P2.

[0025] Step S21 passes the heat consumption behavior parameters of the object to be estimated during the heating period P1 into a pre-trained heat use behavior recognition network. This network can estimate the heat use behavior standard coefficient based on the input behavior parameters, providing a scientific basis for heating scheduling.

[0026] First, the heat supply forecasting management system needs to obtain the heat consumption behavior parameters of the object to be estimated during the heating period P1. These parameters include, for example, the operating time, energy consumption, and set temperature of the heating equipment, which record in detail the heat use behavior of the user during the P1 period. These parameters can be expressed in the form of vectors, with each vector element corresponding to a specific behavior feature.

[0027] Next, these calorie consumption behavior parameters are passed to the calorie usage behavior recognition network, which is a pre-trained machine learning model that has the ability to extract key features from calorie consumption behavior parameters and predict calorie usage behavior norm coefficients. In the specific implementation, various machine learning algorithms or neural network structures can be used, such as multi-layer perceptron (MLP), convolutional neural network (CNN) or recurrent neural network (RNN). These network models are trained with a large amount of historical data and learn the mapping relationship from calorie consumption behavior parameters to behavior norm coefficients.

[0028] The training process of the heat use behavior recognition network is based on the data of the heating period P2 (such as November 1 to November 30, 2023). During the P2 period, the user's heat consumption behavior parameters are also collected, and the normative coefficient mark of the behavior that has occurred is determined by analyzing the statistics of the irregular behavior that has occurred in these parameters. The statistics of irregular behavior that has occurred may include behaviors that do not comply with the heating management specifications, such as abnormally high energy consumption, too high or too low set temperature. The system generates corresponding normative coefficient marks of the behavior that has occurred based on these statistics. These marks are expressed in the form of numerical values, for example, to indicate whether the user's heating use behavior in the P2 period is standardized. Then, the heat consumption behavior parameters of the P2 period and the normative coefficient marks of the behavior that has occurred are used to train the heat use behavior recognition network. The training process adjusts the parameters in the network (such as weights and biases) so that the network can accurately predict the corresponding heat use behavior normative coefficient based on the input heat consumption behavior parameters. This process includes multiple iterations and optimizations until the performance of the network reaches the preset standard.

[0029] In step S21, the heat forecast management system passes the heat consumption behavior parameters of the P1 period into the network. The network outputs an estimated heat usage behavior standard coefficient based on these parameters and the mapping relationship learned by itself. This coefficient reflects whether the user's heat usage behavior in the P1 period is standardized and to what extent. Based on this estimated coefficient, the heat management system can formulate corresponding heat scheduling strategies, such as adjusting parameters such as heating duration and temperature, to optimize heating efficiency and energy use.

[0030] Step S22: Obtain the network weight corresponding to the heating period P2 in the heat usage behavior recognition network, and obtain the weight second-order tensor corresponding to the network weight.

[0031] In step S22 of the embodiment of the present application, the heat prediction management system obtains the network weights corresponding to the heating period P2 in the heat usage behavior recognition network, and further obtains the weight second-order tensor (or weight matrix) corresponding to these weights. The heat usage behavior recognition network is a pre-trained machine learning model, which is trained based on the data of the heating period P2 (such as the previous heating cycle). During the training process, the network learns how to extract key features from the heat consumption behavior parameters and predict the heat usage behavior normative coefficients. This network contains a large number of parameters, the most important of which is the network weight.

[0032] The network weights determine how the network responds to input data, and they are automatically adjusted through the training process. In the scenario of heat estimation management, these weights reflect the network's understanding and learning of the user's heat consumption behavior during the heating period P2. That is, they capture the patterns and rules of the user's heating use behavior during the P2 period. When the heat prediction management system needs to estimate the user's heat use behavior normative coefficient during the heating period P1, it will use this trained heat use behavior recognition network again. In step S22, the network weights corresponding to the heating period P2 are obtained. These weights are optimized specifically for the data of the P2 period, so they can best reflect the characteristics of the user's heating use behavior during that period. Next, the weight second-order tensors corresponding to these network weights are further obtained (second-order tensors are matrices). In a neural network, weights exist in the form of matrices, for example, and each second-order tensor corresponds to a layer in the network. These weight second-order tensors store the connection strength between neurons in each layer of the network from the input layer to the output layer.

[0033] For example, assume that the heat usage behavior recognition network is a multi-layer perceptron (MLP) model. In this model, each layer contains a certain number of neurons, and the connections between neurons are determined by the weight matrix. When the network weights corresponding to the heating period P2 are obtained in step S22, the weight second-order tensors of each layer in these MLP models are actually obtained. Each element in these weight second-order tensors corresponds to a specific connection weight, which together determine how the network responds to the user's heat consumption behavior during the P2 period. By obtaining these weight second-order tensors, the heating prediction management system can calculate them with the heat consumption behavior parameters of the heating period P1 in step S23 to obtain the user's estimated heat usage behavior normative coefficient during the P1 period. This coefficient will serve as an important basis for subsequent heating scheduling decisions to help optimize heating efficiency and energy use.

[0034] Step S23: Obtain the behavior parameter second-order tensor corresponding to the heat consumption behavior parameter in the heating period P1, perform a tensor product of the behavior parameter second-order tensor and the weight second-order tensor to obtain the estimated heat usage behavior standard coefficient of the object to be estimated in the heating period P1.

[0035] Step S23 converts the heat consumption behavior parameters of the object to be estimated during the heating period P1 into a second-order tensor of behavior parameters (as understood in the aforementioned second-order tensor, it can be understood as a behavior parameter matrix), and performs a tensor product operation with the weighted second-order tensor in the heat usage behavior recognition network, thereby obtaining the estimated heat usage behavior standard coefficient of the object to be estimated during the heating period P1.

[0036] First, the heat forecasting management system needs to obtain the heat consumption behavior parameters in the heating period P1. These parameters record in detail the user's heating usage behavior during the P1 period, including the operating time, energy consumption, set temperature, etc. of the heating equipment. These parameters are expressed in the form of vectors, for example, and each vector element corresponds to a specific behavior feature. However, in order to operate with the weight matrix in the heat usage behavior recognition network, these parameters need to be converted into a second-order tensor of behavior parameters, that is, a behavior parameter matrix.

[0037] The construction of the behavior parameter matrix depends, for example, on the specific dimensions and number of the heat consumption behavior parameters. Assuming there are n heat consumption behavior features, the behavior parameter matrix is ​​an n×1 column vector, in which each element corresponds to an eigenvalue. Next, obtain the weight second-order tensor corresponding to the heating period P2 in the heat use behavior recognition network, that is, the weight matrix. This weight matrix is ​​learned through a large amount of data during the training process, and it contains the network's understanding and learning results of the user's heating use behavior during the P2 period. The size of the weight matrix depends on the structure and design of the network, for example, it is an m×n matrix, where m is the number of output neurons (in this scenario, there may be only one output neuron, that is, the heat use behavior specification coefficient), and n is the number of input features (the same as the number of columns in the behavior parameter matrix).

[0038] After obtaining the behavior parameter matrix and the weight matrix, the heat supply forecast management system performs a tensor product operation (i.e., matrix multiplication). The behavior parameter matrix is ​​multiplied on the left by the weight matrix to obtain a 1×1 matrix (i.e., a scalar). This scalar is the estimated heat use behavior standard coefficient of the object to be estimated in the heating period P1. This coefficient reflects whether the user's heating use behavior in the P1 period is standardized, and the degree of standardization.

[0039] For example, suppose the behavior parameter matrix is ​​[10, 50, 22] T, the weight matrix is ​​a 3×1 matrix [[0.2], [0.3], [0.5]] (understandably, this is just a simplified example). Then, through matrix multiplication, the estimated heat usage behavior standard coefficient is 100.2 + 500.3 + 22 0.5 = 27. The higher the value, the more standardized the user's heating behavior. Based on this estimated coefficient, the heating management system can formulate corresponding heating scheduling strategies, such as adjusting parameters such as heating duration and temperature, to optimize heating efficiency and energy use.

[0040] As an embodiment, the method further includes: Step S211: Obtain the heat consumption behavior parameters of the object to be estimated in the heating period P2 and the statistics of the irregular behaviors that have occurred in the heat consumption behavior parameters in the heating period P2.

[0041] In the embodiment of the present application, the heat supply forecast management system needs to clarify the specific scope of the object to be estimated and the heating period P2. Assume that the object to be estimated is a resident household, and the heating period P2 is the previous heating cycle, for example, from November 1, 2022 to November 30, 2022. Next, the heat consumption behavior parameters of the object to be estimated in the heating period P2 are collected. These parameters record in detail the user's heating use behavior in the P2 period, including but not limited to equipment operating time, energy consumption, set temperature, etc.

[0042] After obtaining the parameters of heat consumption behavior, the heating forecast management system further analyzes these parameters and counts the amount of irregular behavior that has occurred. Irregular behavior is defined, for example, according to heating management specifications, and may include abnormally high energy consumption, too long equipment operation time, too high or too low set temperature, etc. The system analyzes each parameter and counts the number or duration of these irregular behaviors. For example, the heating forecast management system may find that the household's energy consumption exceeds the set energy consumption limit (such as more than 100kWh per day) for many days during the P2 period, or the set temperature is too high during certain time periods (such as more than 25°C), which are considered irregular behaviors that have occurred. The system records the statistics of these irregular behaviors, such as the number of days exceeding the energy consumption limit, the length of the period when the set temperature is too high, etc.

[0043] Through step S211, the heat consumption behavior parameters of the object to be estimated in the heating period P2 and the statistics of the irregular behaviors that have occurred are obtained. These data will be used to generate the coefficient mark of the behavior norms that have occurred in the subsequent steps, and serve as an important basis for training the heat use behavior recognition network. In this way, the heat supply prediction management system can train and optimize the recognition network based on the user's historical heat use behavior data, thereby improving the prediction accuracy of future heat use behavior norms.

[0044] Step S212: Generate a normative coefficient mark of the behavior that has occurred for the object to be estimated in the heating period P2 based on the statistics of the irregular behaviors that have occurred.

[0045] In step S212, the heat supply forecast management system needs to generate a behavior standard coefficient mark based on the statistics of the irregular behaviors that have occurred. This mark is a numerical value used to quantify whether the user's heating use behavior in the P2 period is standardized and the degree of standardization. The specific method of generating the mark can be designed according to the actual situation, for example, considering the following factors:

[0046] Irregular behavior threshold: Set one or more thresholds for irregular behavior. These thresholds are used to determine whether the user's behavior has reached an irregular level. For example, energy consumption exceeding 100kWh / day, set temperature exceeding 25°C, or device operation time exceeding 12 hours / day can be set as irregular behavior thresholds.

[0047] Analysis of statistics: Analyze the statistics of irregular behaviors that have occurred and calculate the number, duration or severity of irregular behaviors. These statistics will serve as an important basis for generating tags.

[0048] Quantification of marking: Based on the analysis of the threshold and statistics of irregular behavior, a standard coefficient mark of the behavior that has occurred can be quantified and generated. This mark is, for example, a value between 0 and 1, where 0 represents complete irregularity and 1 represents complete standardization. The system can design specific quantification algorithms or rules based on actual conditions.

[0049] For example, suppose that in the heating period P2, the statistics of the irregular behaviors of the object to be estimated show that the energy consumption of the user exceeds the threshold of 100kWh / day on certain days, and the set temperature exceeds 25°C in certain periods. Based on these statistics, the heating forecast management system can quantify and generate a normative coefficient mark of the behavior that has occurred, such as 0.7. This mark indicates that the user's heating use behavior in the P2 period is relatively standardized, but there are still some irregular behaviors.

[0050] The behavior norm coefficient mark generated in step S212 can be used as one of the important labels for training the heat use behavior recognition network. During the training process, the network will learn how to predict the corresponding behavior norm coefficient from the user's heat consumption behavior parameters, thereby achieving an accurate assessment of the user's heating use behavior. This will provide a scientific basis for subsequent heating scheduling decisions and help optimize heating efficiency and energy use.

[0051] Step S213: Based on the heat consumption behavior parameters in the heating period P2 and the normative coefficient marks of the behaviors that have occurred in the heating period P2, the heat use behavior recognition network to be trained is trained to obtain the heat use behavior recognition network.

[0052] In the embodiment of the present application, step S213 trains the heat usage behavior recognition network to be trained based on the heat consumption behavior parameters in the heating period P2 and the corresponding behavior norm coefficient marks that have occurred, thereby obtaining a mature network model that can accurately identify heat usage behavior norms.

[0053] Next, the heat supply forecasting management system selects a suitable machine learning algorithm or neural network structure to construct a heat usage behavior recognition network to be trained, such as a multi-layer perceptron (MLP) neural network. Then, the training network is trained using the data set of the P2 period.

[0054] The training process, for example, includes: preprocessing the heat consumption behavior parameters of the P2 period and converting them into numerical data that can be processed by the neural network. In the training process, forward propagation is first performed. The preprocessed feature vector is passed into the neural network as input data, and the network calculates the corresponding output value, that is, the predicted heat usage behavior normative coefficient according to its own weight and bias parameters. Then, the predicted normative coefficient is compared with the actual normative coefficient mark of the behavior that has occurred, and the prediction error (that is, loss) is calculated. This loss function measures the gap between the network prediction result and the actual situation. Next, back propagation is performed to adjust the weights and bias parameters in the network according to the gradient information calculated by the loss function to reduce the prediction error and improve the prediction accuracy of the network. This process is achieved by an optimization algorithm (such as a gradient descent algorithm). The above process is repeated for multiple iterations until the prediction performance of the network reaches a preset standard or meets a certain stop condition (such as reaching the maximum number of iterations). In each iteration, the heat supply prediction management system uses new data samples to train the network and gradually optimizes the network parameters to improve the prediction performance.

[0055] Through the training process of step S213, the heat supply forecasting management system obtains a mature heat use behavior recognition network model. This model can accurately predict the heat use behavior standard coefficient based on the user's heat consumption behavior parameters, providing a scientific basis for subsequent heat supply scheduling decisions.

[0056] As an implementation mode, step S212, based on the statistics of the irregular behaviors that have occurred, generates the normative coefficient mark of the behavior that has occurred for the object to be estimated in the heating period P2, which may specifically include: Step S2121: Obtain statistics of irregular behaviors of a number of existing objects in the heating period P2, wherein the number of existing objects are objects that generate heat consumption in the heating period P2, and the number of existing objects include objects to be estimated.

[0057] In the embodiment of the present application, the existing objects refer to objects that actually generate heat consumption during the heating period P2, such as residential households, office buildings, factories, etc. These objects will generate a series of heat consumption behavior data during the heating process.

[0058] During the execution of step S2121, it is necessary to first determine the heating period P2, which is the time range for data collection. For example, the P2 period can be the last heating cycle, that is, a period of time in the last winter, such as November 1, 2022 to March 31, 2023. Next, all existing objects that have generated heating consumption in the P2 period are retrieved from the database. These objects may include thousands of households, businesses, and other institutions that have used heating services in the P2 period. Then, for each existing object, the heating forecast management system needs to collect statistics of its irregular behaviors in the P2 period. Irregular behaviors refer to those behaviors that do not comply with heating management specifications or standards, such as abnormally high energy consumption, too high or too low set temperatures, and long-term operation of equipment. These behaviors will not only increase heating costs, but may also cause damage to the heating system.

[0059] In order to accurately count irregular behaviors, the heat supply forecast management system, for example, makes judgments based on a series of preset rules and thresholds. Similarly, the heat supply forecast management system can also set upper and lower limits for the set temperature, and behaviors beyond these ranges are also regarded as irregular behaviors. In the process of collecting irregular behavior statistics, it is necessary to traverse each existing object and check each heat consumption behavior data in the P2 time period. For each piece of data, the heat supply forecast management system makes judgments based on preset rules and thresholds, and counts the number or duration of irregular behaviors of the object in the P2 time period. Finally, the existing objects collected in step S2121 include not only the objects to be estimated (i.e., the objects whose heating behavior standardization currently needs to be evaluated), but also all other existing objects that generate heat consumption in the P2 time period.

[0060] Step S2122: Determine the number of objects corresponding to a number of existing objects, and determine the average number of unplanned behaviors in the heating period P2 based on the statistics of unregulated behaviors and the number of objects.

[0061] In step S2121, the irregular behavior statistics of several existing objects in the heating period P2 have been obtained. These existing objects are objects that actually generate heating consumption in the P2 period, including all users including the object to be estimated. The irregular behavior statistics of each object reflect its behavioral characteristics in the heating process, such as abnormally high energy consumption, too high or too low set temperature, etc.

[0062] In step S2122, the heat supply forecast management system determines the total number of these existing objects. This number represents the sample size involved in the statistics, which is crucial for the subsequent calculation of the average number of unplanned behaviors. For example, assuming that there are 1,000 existing objects that have generated heat consumption during the P2 period, then these objects are the total number of samples for the irregular behavior statistics. Next, the heat supply forecast management system needs to calculate the average number of unplanned behaviors in the heating period P2 based on the irregular behavior statistics of each object and the total number of objects. This average value is a statistical indicator that reflects the average degree of irregular behavior of all objects participating in the statistics during the P2 period. The method of calculating the average value is, for example, to add up the irregular behavior statistics of all objects and then divide it by the total number of objects.

[0063] By calculating the average number of unplanned behaviors, the heat forecast management system can provide an important reference benchmark for subsequent steps. This benchmark will help determine the degree of behavioral standardization of the object to be estimated during the P2 period, and compare and analyze it with the estimated heat usage behavior standardization coefficient during this period. In this way, the heat forecast management system can more accurately evaluate the user's heat usage behavior and provide a scientific basis for heat scheduling decisions.

[0064] Step S2123: Based on the statistics of the irregular behaviors that have occurred and the average number of unplanned behaviors, determine the normative coefficient mark of the behavior that has occurred for the object to be estimated in the heating period P2.

[0065] In step S2122, the average number of unplanned behaviors in the heating period P2 has been calculated. This average value represents the average degree of irregular behaviors of all objects participating in the statistics in this period. At the same time, in step S2121, the heating forecast management system also obtains the statistics of irregular behaviors that have occurred in the P2 period of the object to be estimated. In step S2123, these two data are used to determine the normative coefficient mark of the behavior that has occurred in the P2 period of the object to be estimated. This mark is a quantitative value used to reflect the degree of standardization of the heating use behavior of the object to be estimated in this period.

[0066] Specifically, the statistics of the irregular behaviors that have occurred for the object to be estimated are compared with the average number of unplanned behaviors. If the statistics of the object to be estimated are lower than or equal to the average value, it means that the degree of its irregular behavior is lower than or close to the average level, so it can be considered that its behavior is relatively standardized. In this case, the heat supply forecasting management system assigns a higher behavior norm coefficient mark to the object to be estimated, such as a value close to 1. On the contrary, if the statistics of the irregular behaviors that have occurred for the object to be estimated are higher than the average value, it means that the degree of its irregular behavior is more serious and exceeds the average level. In this case, the heat supply forecasting management system will assign a lower behavior norm coefficient mark to the object to be estimated, such as a value close to 0. Through the evaluation of step S2123, a quantitative standardization index can be provided for the heating usage behavior of the object to be estimated during the heating period P2.

[0067] As an implementation mode, step S2123, based on the statistics of the irregular behaviors that have occurred and the average number of unplanned behaviors, determines the normative coefficient mark of the behavior that has occurred in the heating period P2 of the object to be estimated, which may specifically include: Step S21231: Compare the statistics of irregular behaviors that have occurred with the average number of irregular behaviors.

[0068] Step S21232: When the statistic of the irregular behaviors that have occurred is greater than or equal to the average number of unplanned behaviors, the first preset value is used as the normative coefficient mark of the behavior that has occurred of the object to be estimated in the heating period P2. The first preset value is used to indicate that the heating use behavior of the object to be estimated in the heating period P2 is irregular behavior.

[0069] In step S21231, the statistics of irregular behaviors that have occurred in the object to be estimated during the heating period P2 are obtained. These statistics may include the number of times the preset energy consumption threshold is exceeded, the time when the set temperature exceeds the comfortable range, etc., which directly reflect whether the user's heating use behavior in the P2 period is standardized. At the same time, based on the irregular behavior statistics of all objects participating in the statistics (including the object to be estimated) in the P2 period, the average number of unplanned behaviors in the period is calculated. This average value represents the average level of irregular behaviors of all objects in the P2 period.

[0070] Next, the statistics of the irregular behaviors that have occurred for the object to be estimated are compared with this average value. The purpose is to determine whether the degree of behavioral norms of the object to be estimated is lower or higher than the average level. If the statistics of the irregular behaviors that have occurred for the object to be estimated are greater than or equal to the average number of unplanned behaviors, it means that the degree of irregularity in the heating use behavior of the object in the P2 period is high, even exceeding the overall level. In this case, a preset first value is used as a marker for the normative coefficient of the behavior that has occurred for the object in the P2 period. This first preset value is, for example, a relatively low value, which is used to clearly indicate that the heating use behavior of the object to be estimated in this period is irregular behavior. For example, the first preset value can be set to 0.5, which is low in the range of 0 to 1 and can intuitively reflect the irregularity of the user's behavior. Through the execution of steps S21231 and S21232, the heating prediction management system can make an accurate assessment of the user's heating use behavior based on quantitative data, and provide valuable label data for subsequent machine learning model training.

[0071] Step S21233: When the statistic of the irregular behaviors that have occurred is less than the average number of unplanned behaviors, the second preset value is used as the normative coefficient mark of the behavior that has occurred in the heating period P2 of the object to be estimated, and the second preset value is used to indicate that the heating use behavior of the object to be estimated in the heating period P2 is a normative behavior.

[0072] In the embodiment of the present application, step S21233 involves how to assign a behavior standard coefficient mark to the object to be estimated in a specific heating period (such as P2 period) based on the comparison result of the statistics of the irregular behaviors that have occurred and the average number of irregular behaviors. In step S21233, the statistics of the irregular behaviors that have occurred in the heating period P2 of the object to be estimated are compared with the average number of irregular behaviors in the period. If the statistics of the irregular behaviors that have occurred are less than the average number of irregular behaviors, it means that the heating use behavior of the object to be estimated in the period is relatively standardized, and the degree of its irregular behavior is lower than the average level.

[0073] Based on this comparison result, a preset second value is assigned to the object to be estimated as a standard coefficient mark of its behavior in the heating period P2. This second preset value is, for example, a higher value, which is used to clearly indicate that the heating use behavior of the object to be estimated in this period is a standard behavior. By assigning a higher coefficient mark, the good performance of the object to be estimated in terms of heating use can be accurately reflected.

[0074] For example, suppose that in the heating period P2, the statistics of the irregular behaviors that have occurred in the object to be estimated are 2 times, and the average number of unplanned behaviors in this period is 3 times. Since the statistics of the object to be estimated are lower than the average value, the second preset value (such as 0.8) is used as its behavior norm coefficient mark according to step S21233. This higher value indicates that the heating use behavior of the object to be estimated in the P2 period is relatively standardized and meets the requirements of heating management. The implementation of this step not only helps to quantitatively evaluate the heating use behavior of the object to be estimated, but also provides valuable label data for subsequent machine learning model training. By assigning different coefficient marks to objects with different degrees of behavioral norms, the machine learning model can learn more accurate mapping relationships and improve the accuracy of predicting future heating use behavior norms.

[0075] As an implementation mode, step S213, based on the heat consumption behavior parameters in the heating period P2 and the behavior norm coefficient mark that has occurred in the heating period P2, the heat use behavior recognition network to be trained is trained to obtain the heat use behavior recognition network, which may specifically include: Step S2131: The heat consumption behavior parameters in the heating period P2 are passed into the heat usage behavior recognition network to be trained, and the training network weights in the heat usage behavior recognition network to be trained and the heat consumption behavior parameters in the heating period P2 are used to estimate the heat usage behavior standard coefficient of the object to be estimated in the heating period P2.

[0076] In step S2131 of the embodiment of the present application, the heat consumption behavior parameters in the heating period P2 are used as input data, and forward propagation is performed through the heat use behavior recognition network to be trained, and the training network weights in the network are used to estimate the heat use behavior normative coefficient of the object to be estimated in the P2 period. First, the heat consumption behavior parameters of the object to be estimated in the heating period P2 are collected. These parameters record in detail the user's heating use behavior in the P2 period, including but not limited to the equipment running time, energy consumption, set temperature, etc. These parameters are represented in the form of vectors, for example, and each vector element corresponds to a specific behavior feature. Next, these heat consumption behavior parameters are passed to the heat use behavior recognition network to be trained. This network is a pre-designed machine learning model for extracting features from input data and predicting heat use behavior normative coefficients. In practical applications, the heat use behavior recognition network can adopt a variety of different algorithms or neural network structures, such as multi-layer perceptron (MLP), convolutional neural network (CNN) or recurrent neural network (RNN). Here, MLP is used as an example for illustration.

[0077] In the MLP network, the input layer receives the parameter vector of calorie consumption behavior, which is then transformed nonlinearly through multiple neurons in the hidden layer. Each neuron is connected to the neurons in the previous and next layers, and the connection weights determine the degree of influence of the input data on the neuron output. These weights are gradually adjusted during the training process, allowing the network to better fit the relationship between the input data and the output label.

[0078] In step S2131, the network to be trained uses the current training network weights to calculate the input data to obtain the estimated heat usage behavior standard coefficient. This process is achieved through forward propagation, that is, the input data is transmitted layer by layer in the network, and the final output value is obtained after calculation by each neuron. This output value is the estimated heat usage behavior standard coefficient of the object to be estimated in the P2 period, which reflects whether the user's heating use behavior in this period is standardized and the degree of standardization.

[0079] For example, suppose the input vector is [10 hours, 50kWh, 22°C], and the training network weights in the network to be trained have been preliminarily adjusted. When this vector is passed into the network, after calculation and processing by the hidden layer, the network finally outputs an estimated heat usage behavior standard coefficient value, such as 0.8. This value indicates that the heating usage behavior of the object to be estimated during the P2 period is relatively standard, but there is still room for improvement.

[0080] By executing step S2131, the heat consumption behavior parameters of the user in the heating period P2 can be estimated by using the heat use behavior recognition network to be trained, and the corresponding heat use behavior standard coefficient can be obtained. This estimation result will provide an important reference basis for subsequent network training, which will help the network to better learn and identify the user's heating use behavior pattern.

[0081] Step S2132: Determine the network training error based on the estimated heat usage behavior specification coefficient in the heating period P2 and the occurred behavior specification coefficient mark in the heating period P2.

[0082] Looking back at step S2131, in this step, the heat forecasting management system uses the heat usage behavior recognition network to be trained (such as a multi-layer perceptron MLP) and the heat consumption behavior parameters in the heating period P2 to estimate the heat usage behavior standard coefficient of the object to be estimated in the P2 period. This estimated coefficient reflects the network's understanding of the current input data and is an intermediate result in the training process. At the same time, in step S212, based on the statistics of irregular behaviors that have occurred in the heating period P2, a real normative coefficient mark of the behavior that has occurred has been generated for each object (including the object to be estimated). This mark is obtained by comparing with the average number of unplanned behaviors to indicate whether the object's heating usage behavior in the P2 period is standardized and to what extent.

[0083] In step S2132, the heat supply forecast management system compares the estimated heat use behavior norm coefficient with the actual behavior norm coefficient mark that has occurred to determine the network training error. This error reflects the difference between the network prediction result and the actual label, and is an important basis for guiding the adjustment of network weights. Specifically, the heat supply forecast management system first retrieves the behavior norm coefficient mark that has occurred for the object to be estimated in the heating period P2 from the database. Then, this mark is compared with the heat use behavior norm coefficient estimated in step S2131. In order to quantify this difference, the heat supply forecast management system, for example, uses a loss function (such as mean square error MSE) to calculate the error between the estimated value and the true value.

[0084] For example, suppose the estimated heat usage behavior norm coefficient is 0.8, and the actual behavior norm coefficient is marked as 0.9. The system uses a loss function to calculate the difference between these two values ​​and obtains a specific error value. This error value will be used as the target for subsequent network weight optimization to guide the network to adjust its parameters in the direction of reducing the error. Through the execution of step S2132, the heating forecasting management system can accurately evaluate the performance of the current network model and guide the further optimization of the model based on the training error. This process is one of the key steps in machine learning, which helps to continuously improve the prediction accuracy and generalization ability of the heat usage behavior recognition network.

[0085] Step S2133: Optimizing the training network weights based on the network training errors to obtain network weights, and using the heat usage behavior recognition network to be trained including the network weights as the heat usage behavior recognition network.

[0086] In the embodiment of the present application, step S2133 optimizes the training network weights based on the network training error to obtain the optimized network weights, and applies these weights to the network to be trained to make it a mature heat usage behavior recognition network. In step S2132, the network training error has been calculated based on the estimated heat usage behavior norm coefficient and the actual behavior norm coefficient mark. This error reflects the accuracy of the current network model in predicting the degree of user heating usage behavior norms, and is an important basis for guiding the adjustment of network weights.

[0087] In step S2133, an optimization algorithm (such as a gradient descent algorithm) is used to adjust the training network weight according to the network training error. The optimization algorithm adjusts the network weight iteratively so that the estimated result gradually approaches the true label, thereby reducing the training error. Specifically, the optimization algorithm calculates the gradient of the network training error to the network weight (i.e., the partial derivative of the error to the network weight), and then updates the network weight according to this gradient information. This process is repeated many times, and each iteration will move the network weight a small step in the direction of reducing the error until the error reaches a preset threshold or the number of iterations reaches an upper limit. When the optimization process is completed, a set of optimized network weights will be obtained. These weights are the result of multiple iterative adjustments, which can make the network model have higher accuracy in predicting the degree of standardization of user heating use behavior. Then, these optimized weights are applied to the heat use behavior recognition network to be trained, making it a mature model that can accurately predict the degree of standardization of user heating use behavior. Through the execution of step S2133, not only the network weights are optimized and the model prediction accuracy is improved, but also a mature heat use behavior recognition network is obtained. This network will be used in subsequent heat supply estimation and management tasks, providing a scientific basis for heat supply scheduling decisions and promoting the rational use of heat resources and energy conservation and emission reduction.

[0088] As an implementation mode, step S30, based on the heat usage behavior standard coefficient in the heating period P2 and the estimated heat usage behavior standard coefficient in the heating period P1, determines the heat usage behavior standard coefficient of the object to be estimated in the heating period P1, which may specifically include: Step S31: obtaining a first contribution factor corresponding to the heat usage behavior standard coefficient in the heating period P2, and a second contribution factor corresponding to the estimated heat usage behavior standard coefficient in the heating period P1.

[0089] In the embodiment of the present application, step S31 obtains two key weight parameters, namely contribution factors. These contribution factors will be used in subsequent steps to adjust the weights of the actual heat usage behavior standard coefficient in the heating period P2 and the estimated heat usage behavior standard coefficient in the heating period P1 in the final calculation. The contribution factor is a weight parameter used to measure the importance of different data or information in the decision-making process. In the scenario of heating estimation management, the contribution factor is used to determine the weights of historical data (actual behavior standard coefficient in the P2 period) and the current estimation result (estimated behavior standard coefficient in the P1 period) in the comprehensive evaluation. In step S31, the first contribution factor corresponding to the heat usage behavior standard coefficient in the heating period P2 is obtained. This factor reflects the importance of the actual behavior standard coefficient in the P2 period in the final evaluation. For example, this factor will be pre-set according to the actual application scenario and requirements. For example, if it is believed that the historical data has a high reference value for the prediction of the current heating behavior, the first contribution factor can be set to a higher value, such as 0.7.

[0090] At the same time, the second contribution factor corresponding to the estimated heat usage behavior specification coefficient in the heating period P1 is obtained. This factor measures the importance of the estimated behavior specification coefficient in the P1 period in the final evaluation. Similar to the first contribution factor, the second contribution factor is also set according to actual needs. If it is believed that the estimated result of the current period has a greater impact on the decision, the second contribution factor can be set to a higher value. In the process of setting the contribution factor, it should be noted that the sum of the values ​​of the two factors is, for example, 1, to ensure that the two behavior specification coefficients can be correctly weighted and summed in subsequent calculations. For example, if the first contribution factor is set to 0.7, then the second contribution factor should be 0.3 to ensure that the sum of the two is 1. Through the execution of step S31, the heating forecast management system obtains two key contribution factors, which provide important weight parameters for the subsequent calculation of the final heat usage behavior specification coefficient of the object to be estimated in the heating period P1. These parameters will be set based on actual application scenarios and needs to ensure the accuracy and rationality of the evaluation results.

[0091] Step S32: multiplying the heat usage behavior standard coefficient in the heating period P2 by the first contribution factor to obtain a first regulated heat usage behavior standard coefficient.

[0092] In step S31, the first contribution factor corresponding to the heat usage behavior standard coefficient in the heating period P2 has been obtained. In step S32, a specific weighted calculation will be performed. Specifically, the heat usage behavior standard coefficient in the heating period P2 is multiplied by the first contribution factor. The size of the actual behavior standard coefficient of the P2 period is adjusted according to the weight of the contribution factor to obtain an adjusted value, namely the first adjusted heat usage behavior standard coefficient. Through the execution of step S32, the historical data can be weighted according to the pre-set contribution factor to obtain a more reasonable and accurate adjustment coefficient. This coefficient will be combined with the estimated behavior standard coefficient of the P1 period in subsequent steps to jointly determine the final heat usage behavior standard coefficient of the object to be estimated in the heating period P1. It is helpful to comprehensively utilize historical data and current estimation results to improve the accuracy and effectiveness of heating estimation management.

[0093] Step S33: multiplying the estimated heat usage behavior standard coefficient in the heating period P1 by the second contribution factor to obtain a second adjusted heat usage behavior standard coefficient.

[0094] In the embodiment of the present application, step S33 multiplies the estimated heat usage behavior standard coefficient in the heating period P1 by the second contribution factor to obtain the second adjustment heat usage behavior standard coefficient. The estimated heat usage behavior standard coefficient in the heating period P1 is estimated by a machine learning model (such as a heat usage behavior recognition network) based on the heat consumption behavior parameters of the current period P1. This estimated coefficient reflects the system's prediction of the degree of standardization of the user's heating usage behavior in the current period.

[0095] At the same time, in step S31, the heat supply forecast management system has obtained the second contribution factor. This factor is a weight parameter used to measure the importance of the P1 period forecast behavior specification coefficient in the final evaluation. The value of the second contribution factor is pre-set according to the actual application scenario and requirements, and it reflects the importance of the current forecast result relative to the historical data.

[0096] In step S33, a specific weighted calculation will be performed. Specifically, the estimated heat usage behavior specification coefficient in the heating period P1 is multiplied by the second contribution factor. The size of the estimated behavior specification coefficient is adjusted according to the weight of the contribution factor to obtain an adjusted value, namely the second adjusted heat usage behavior specification coefficient. Through the execution of step S33, the current estimated result can be weighted according to the preset contribution factor to obtain a more reasonable and accurate adjustment coefficient. This coefficient will be combined with the heat usage behavior specification coefficient of the P2 period in subsequent steps to jointly determine the final heat usage behavior specification coefficient of the object to be estimated in the heating period P1. This process helps to comprehensively utilize historical data and current estimated results to improve the accuracy and effectiveness of heating estimation management.

[0097] Step S34: summing the first regulating heat usage behavior standard coefficient and the second regulating heat usage behavior standard coefficient to obtain the heat usage behavior standard coefficient of the object to be estimated in the heating period P1.

[0098] In the embodiment of the present application, step S34 combines the first heat usage behavior regulation coefficient and the second heat usage behavior regulation coefficient calculated in the previous step to obtain the final heat usage behavior regulation coefficient of the object to be estimated in the heating period P1. In steps S32 and S33, the first heat usage behavior regulation coefficient and the second heat usage behavior regulation coefficient have been calculated respectively. These two coefficients are based on the actual heat usage behavior regulation coefficient in the heating period P2 and the estimated heat usage behavior regulation coefficient in the heating period P1, respectively, and are weighted by the corresponding contribution factor. In step S34, the heating forecast management system sums the two adjustment coefficients. Specifically, the first heat usage behavior regulation coefficient and the second heat usage behavior regulation coefficient are added to obtain the final heat usage behavior regulation coefficient of the object to be estimated in the heating period P1.

[0099] As an implementation method, the method provided in the embodiment of the present application also includes: Step S40: In the heating period P1, a heating energy usage application sent by the device to be predicted is obtained, and the device to be predicted is the device corresponding to the object to be estimated.

[0100] In the heating period P1, the user's heating demand may change with factors such as weather changes, indoor temperature, and personal comfort needs. In order to meet these changing needs, the user will, for example, send a heating energy use application to the heating management system through its device to be predicted (such as an intelligent thermostat, heating system control terminal, etc.). This application is essentially an instruction that clearly expresses the user's demand for heating services during the current heating period. After receiving this application, the heating forecast management system will be able to understand the user's specific heating needs during the current heating period P1. This information is crucial for subsequent heating estimation management because it will directly affect how the heating system provides services to users, including the duration of heating, temperature settings, etc.

[0101] It should be noted that the devices to be predicted are important interfaces for users to interact with the heating management system. These devices, for example, have networking functions and can transmit users' heating needs to the heating prediction management system in real time. At the same time, they may also have some intelligent functions, such as automatically adjusting heating parameters based on users' historical behavior data, thereby improving the personalization and comfort of heating services.

[0102] Step S50: Based on the application for heating energy use, obtain the heat use behavior standard coefficient in the heating period P1, and determine the heating execution strategy of the object to be estimated and the heating pricing mechanism corresponding to the heating execution strategy based on the heat use behavior standard coefficient in the heating period P1.

[0103] In the embodiment of the present application, step S50 determines the heating execution strategy of the object to be estimated and the corresponding heating pricing mechanism based on the user's heating energy use application and the heat use behavior normative coefficient in the heating period P1. First, when the user sends a heating energy use application through the device to be predicted (such as an intelligent thermostat) in the heating period P1, the heating prediction management system responds to this application and obtains relevant information, such as the user's desired heating temperature, heating duration and other specific requirements.

[0104] Next, query and obtain the heat usage behavior standard coefficient in the heating period P1. This coefficient is calculated by combining the actual behavior standard coefficient in the heating period P2 and the estimated behavior standard coefficient in the P1 period, and combining the corresponding contribution factor. It reflects whether the user's heating usage behavior in the current heating period meets the standard and the degree of standardization.

[0105] Once the heating energy application and heat use behavior coefficient are obtained, the heating forecast management system can start to formulate a heating implementation strategy. This strategy will be determined based on the user's specific needs and behavior coefficient. For example, if the behavior coefficient is high, it means that the user's heating behavior is relatively standardized, and the heating forecast management system may directly provide heating according to the heating temperature and duration applied by the user; if the behavior coefficient is low, the heating forecast management system may appropriately adjust the heating parameters, such as lowering the heating temperature or shortening the heating duration, to encourage users to save more energy.

[0106] At the same time, the heat forecast management system will also determine the corresponding heat pricing mechanism according to the heat execution strategy. The pricing mechanism, for example, calculates the heating fee based on the user's heat consumption and behavior norm coefficient. For example, users with a higher behavior norm coefficient may enjoy a lower unit heat price to encourage them to continue to maintain energy-saving behavior; while users with a lower behavior norm coefficient may need to pay a higher fee to encourage them to improve their heating usage behavior.

[0107] For example, suppose a user applies through a smart thermostat to raise the indoor temperature to 22°C for 8 hours during the heating period P1. After the system query, it is found that the user's heat usage behavior normative coefficient during the P1 period is 0.8 (high), indicating that his heating usage behavior is relatively normative. Therefore, the heating forecast management system may directly provide heating according to the user's application and provide him with a relatively low unit heat price. On the contrary, if the behavior normative coefficient is low (such as 0.5), the heating forecast management system may adjust the heating temperature to 21°C or shorten the heating time to 6 hours, and provide the user with a higher unit heat price.

[0108] By executing step S50, the heat supply forecast management system can formulate a personalized heat supply execution strategy and pricing mechanism according to the user's heat supply demand and behavior norm coefficient. This intelligent management method can not only meet the user's personalized needs, but also promote the user to develop energy-saving habits, improve heating efficiency and reduce energy consumption.

[0109] Step S60: The heating execution strategy and the heating pricing mechanism are used as the heating scheduling method for the heating energy usage of the object to be estimated, and the heating scheduling method is sent to the device of the object to be predicted.

[0110] In step S50, the heating execution strategy and corresponding heating pricing mechanism of the estimated object have been determined based on the user's heating energy use application and the heat use behavior norm coefficient in the heating period P1. These strategies and mechanisms are customized according to the user's personalized needs and behavioral norms, aiming to provide heating services that meet the user's expectations while encouraging energy-saving behavior.

[0111] In step S60, these heating execution strategies and heating pricing mechanisms are integrated into a complete heating scheduling method. This scheduling method details how to provide heating services to users, including specific parameters such as heating time and heating temperature, as well as the calculation method and standard of heating costs.

[0112] Next, this heating scheduling method is sent to the device to be predicted. The device to be predicted is the device associated with the object to be estimated (such as a user's home or enterprise), such as a smart thermostat, a heating system control terminal, etc. These devices can receive and execute the scheduling method sent by the system, automatically adjust the heating parameters, and ensure that the heating service is carried out according to the predetermined strategy.

[0113] For example, suppose the system determines the following heating scheduling method for the user: in the heating period P1, the indoor temperature is maintained between 20°C and 22°C, and the heating time is 8 hours per day; the heating cost is calculated based on the actual heat consumption of the user and the price of 0.5 yuan / kWh. The system will send this scheduling method to the user's smart thermostat. After receiving the scheduling method, the smart thermostat will automatically adjust the heating parameters to ensure that the indoor temperature is within the user's expected range, and calculate the heating cost based on the user's actual heat consumption.

[0114] By executing step S60, the heat supply forecast management system can apply the customized heat supply execution strategy and pricing mechanism to the user's heat supply service in real time. This intelligent management method can not only improve the efficiency and comfort of heat supply services, but also motivate users to save energy and promote sustainable development through a reasonable pricing mechanism.

[0115] Step S70: Compare the heat usage behavior standard coefficient in the heating period P1 with the reference heat usage behavior standard coefficient.

[0116] Step S80: When the heat usage behavior specification coefficient in the heating period P1 is less than the reference heat usage behavior specification coefficient, a heating rejection information is generated and sent to the device of the object to be predicted. The heating rejection information is used to inform the object to be predicted that the heat usage behavior specification coefficient in the heating period P1 does not meet the heating requirements.

[0117] In step S70, the heat use behavior standard coefficient in the heating period P1 is obtained. This coefficient is calculated by the machine learning model and historical data in the previous step, and it reflects the standard degree of the user's heating use behavior in the P1 period. Then, this behavior standard coefficient is compared with a preset reference heat use behavior standard coefficient. The reference coefficient is, for example, a threshold set based on factors such as heating management policies and energy efficiency standards, which is used to determine whether the user's heating use behavior meets the requirements. The purpose of the comparison is to evaluate whether the user's heating use behavior reaches the expected standard level. If the behavior standard coefficient is higher than or equal to the reference coefficient, it means that the user's heating use behavior is relatively standardized; if it is lower than the reference coefficient, it indicates that the user's heating use behavior may have problems and needs further adjustment or optimization. If it is found in the comparison of step S70 that the heat use behavior standard coefficient in the heating period P1 is less than the reference heat use behavior standard coefficient, it means that the user's heating use behavior does not meet the requirements. In this case, the heating prediction management system will generate a heating rejection message and send the message to the device to be predicted. The content of the heat rejection information includes, for example, a specific description of the user's irregular heat use behavior, the reasons for non-compliance, and possible suggestions for improvement. For example, the information may point out that the user's current heat energy consumption is too high, and recommends adjusting the heating temperature or duration to save energy. The purpose of issuing heat rejection information is to promptly inform users of problems with their heat use behavior and guide them to take improvement measures. This helps users understand the impact of their behavior on the heating system, promotes users to use heating energy more rationally, improves heating efficiency, and reduces energy consumption.

[0118] Step S90: When the heat usage behavior norm coefficient in the heating period P1 is not less than the reference heat usage behavior norm coefficient, the step of determining the heating execution strategy of the object to be estimated and the heating pricing mechanism corresponding to the heating execution strategy based on the heat usage behavior norm coefficient in the heating period P1 is executed.

[0119] In step S90, it is determined whether the heat usage behavior standard coefficient in the heating period P1 is not less than the reference heat usage behavior standard coefficient. This determination is based on the comparison result in step S70. If the behavior standard coefficient is not less than the reference coefficient, it means that the user's heating usage behavior meets the requirements of heating management, that is, the user's behavior in the current heating period is standardized and energy-saving.

[0120] When this condition is met, the step of determining the heating execution strategy and the corresponding heating pricing mechanism of the object to be estimated based on the heat usage behavior norm coefficient in the heating period P1 is executed. This step is similar to step S50, but is executed here as a conditional branch. Specifically, a personalized heating execution strategy is formulated according to the user's heat usage behavior norm coefficient. This strategy takes into account the user's actual needs and energy-saving goals, including specific parameters such as heating time and heating temperature. For example, if the behavior norm coefficient is high, the heating forecast management system may provide heating according to the temperature and duration requested by the user; if the coefficient meets the specification but is low, the heating forecast management system may appropriately fine-tune the heating parameters to encourage the user to maintain energy-saving behavior. At the same time, the heating forecast management system will also determine the corresponding heating pricing mechanism according to the heating execution strategy. The pricing mechanism, for example, calculates the heating cost based on the user's heat consumption and behavior norm coefficient. Since the user's heating use behavior meets the specification, the heating forecast management system may give the user a certain fee discount, such as a lower unit heat price, to encourage the user to continue to maintain energy-saving behavior.

[0121] For example, suppose a user applies to raise the indoor temperature to 22°C for 8 hours through the smart thermostat during heating period P1. The system calculates the heat usage behavior standard coefficient to be 0.8, and the reference heat usage behavior standard coefficient is also 0.8. Since the behavior standard coefficient is not less than the reference coefficient, the heating forecast management system determines that the user's heating usage behavior meets the standard.

[0122] Therefore, the heating forecast management system will formulate heating execution strategies based on the user's application and behavioral norm coefficient, such as directly providing heating according to the temperature and duration requested by the user. At the same time, the heating forecast management system will provide users with a relatively low unit heat price to encourage users to continue energy-saving behavior. In this way, users can not only obtain comfortable heating services, but also enjoy the economic benefits brought by energy saving.

[0123] As an implementation mode, step S50, based on the heat usage behavior norm coefficient in the heating period P1, determines the heating execution strategy of the object to be estimated, and the heating pricing mechanism corresponding to the heating execution strategy, which may specifically include: Step S51: obtaining a heating strategy index set, wherein the heating strategy index set includes a matching condition between a first heat usage behavior specification coefficient range and a heating execution strategy, and each first heat usage behavior specification coefficient range corresponds to a heating execution strategy.

[0124] In the embodiment of the present application, the heating strategy index set is used to guide how to determine a specific heating execution strategy based on the user's heat usage behavior standard coefficient.

[0125] First, the heating strategy index set is a database or data structure containing multiple entries, each of which describes the matching relationship between a first heat usage behavior norm coefficient range and a corresponding heating execution strategy. This range is set, for example, based on the needs of heating management and historical data of user behavior, to distinguish user groups with different levels of behavioral norms. For example, the heating strategy index set may contain the following entries:

[0126] When the heat usage behavior specification coefficient is between 0.9 and 1.0, the "full power standard heating strategy" is adopted, that is, full power heating is provided according to the temperature and time set by the user.

[0127] When the heat usage behavior regulation coefficient is between 0.7 and 0.9, the "energy-saving mode heating strategy" is adopted, that is, the heating power is appropriately reduced or the heating time is adjusted while ensuring comfort.

[0128] When the heat usage behavior norm coefficient is lower than 0.7, the heating forecast management system may consider adopting a "heating restriction strategy", such as reducing the heating time or lowering the heating temperature, to encourage users to improve their heating usage behavior.

[0129] In step S51, the heating forecast management system accesses the heating strategy index set and loads all the entries therein. These entries provide the system with a basis for determining the heating execution strategy based on the user's heat usage behavior norm coefficient. By matching the user's actual behavior norm coefficient with the range in the set, the heating forecast management system can quickly find the corresponding heating execution strategy and ensure that the provision of heating services meets both the user's needs and the goal of energy-saving management. This matching method based on the index set not only improves the response speed and accuracy of the system, but also makes the heating management strategy more flexible and configurable. Managers can adjust the range and strategy in the index set according to actual conditions to adapt to changes in different user groups and heating environments. At the same time, by continuously updating and optimizing the index set, the heating forecast management system can better adapt to changes in user behavior and provide more personalized heating services.

[0130] Step S52: determining a first heat usage behavior standard coefficient range corresponding to the heat usage behavior standard coefficient in the heating period P1 as a first target heat usage behavior standard coefficient range.

[0131] In the embodiment of the present application, step S52 matches the heat usage behavior norm coefficient in the heating period P1 with the first heat usage behavior norm coefficient range in the heating strategy index set to determine the corresponding first target heat usage behavior norm coefficient range.

[0132] In step S52, the heat supply forecast management system obtains the heat use behavior specification coefficient in the heating period P1. Next, the heat supply forecast management system matches this coefficient with the first heat use behavior specification coefficient range in the heating strategy index set. This matching process is completed by comparing the coefficient value with the upper and lower limits of each range in the set. Specifically, the heat supply forecast management system traverses each range in the set to check whether the current coefficient falls within the range. Once a matching range is found, the heat supply forecast management system determines it as the first target heat use behavior specification coefficient range. For example, assume that the heating strategy index set contains the following range and strategy matching relationship: Range 1: 0.9-1.0, corresponding to strategy A (full power standard heating); Range 2: 0.7-0.9, corresponding to strategy B (energy-saving mode heating); Range 3: 0.0-0.7, corresponding to strategy C (restricted heating).

[0133] Now, assume that the heat usage behavior standard coefficient in the heating period P1 is 0.85. The system will match this coefficient with the range in the set. Since 0.85 falls within range 2 (0.7-0.9), the system will determine range 2 as the first target heat usage behavior standard coefficient range, and use the corresponding strategy B (energy-saving mode heating) as the heating execution strategy for the object to be estimated in the P1 period.

[0134] Through the execution of step S52, the heat supply forecast management system can accurately determine the corresponding heat supply execution strategy range according to the user's heat use behavior standard coefficient, providing a basis for subsequent strategy selection and pricing mechanism determination. This range-based matching method not only improves the accuracy and efficiency of the system, but also makes the heat supply management strategy more flexible and configurable.

[0135] Step S53: Determine the heating execution strategy corresponding to the first target heat usage behavior specification coefficient range in the heating strategy index set as the heating execution strategy.

[0136] In the embodiment of the present application, step S53 selects the corresponding heating execution strategy from the heating strategy index set based on the first target heat usage behavior specification coefficient range determined in step S52. In step S52, the heating forecast management system has determined the corresponding first target heat usage behavior specification coefficient range based on the heat usage behavior specification coefficient in the heating period P1. This range reflects the degree of standardization of the user's heating usage behavior during the P1 period. In step S53, the heating forecast management system searches the heating strategy index set to find a heating execution strategy that matches the first target heat usage behavior specification coefficient range. This process is accomplished by comparing the range in the set with the first target range. Once a matching range is found, the heating forecast management system will use the heating execution strategy corresponding to the range as the final heating execution strategy.

[0137] If the first target heat usage behavior specification coefficient range determined in step S52 is range 2 (0.7-0.9), then in step S53, the heat supply forecast management system will select strategy B (energy-saving mode heating) corresponding to range 2 from the heat supply strategy index set as the final heat supply execution strategy. This strategy B (energy-saving mode heating) will be used to guide the heating system on how to provide heating services to users during the heating period P1. For example, the heat supply forecast management system may appropriately reduce the heating power or adjust the heating time according to the requirements of strategy B to achieve energy-saving goals while ensuring user comfort. Through the execution of step S53, the heat supply forecast management system can determine the appropriate heating execution strategy based on the user's heat usage behavior specification coefficient to achieve personalized heating management. This method of determining the heating execution strategy based on user behavior specifications helps to improve heating efficiency, reduce energy consumption, and promote users' energy-saving behavior.

[0138] Step S54: Obtain a pricing mechanism index set, where the pricing mechanism index set includes a matching condition between the second heat usage behavior specification coefficient range and the pricing step factor, and each second heat usage behavior specification coefficient range corresponds to a pricing step factor.

[0139] In the embodiment of the present application, step S54 obtains a predefined pricing mechanism index set. This set provides the system with a basis for determining the heating pricing mechanism based on the user's heat usage behavior standard coefficient. The pricing mechanism index set is a database or data structure containing multiple entries. Each entry describes the matching relationship between a second heat usage behavior standard coefficient range and a corresponding pricing ladder factor. These ranges are set based on the economic goals of heating management and historical data of user behavior, and are used to distinguish the heating cost calculation methods under different levels of behavioral standardization.

[0140] Specifically, the pricing ladder factor is a coefficient used to adjust the heating cost calculation formula. It is determined based on the user's heat usage behavior coefficient, reflecting the impact of user behavior on heating costs. When the user's behavior coefficient is high, the heating forecast management system may give a lower pricing ladder factor to reduce the user's heating costs and encourage energy-saving behavior; conversely, when the behavior coefficient is low, the heating forecast management system may increase the pricing ladder factor and increase the user's heating costs to encourage users to improve their heating usage behavior.

[0141] In step S54, the heat supply forecast management system first accesses and loads the pricing mechanism index set. This set contains the matching relationship between multiple second heat use behavior specification coefficient ranges and pricing ladder factors. For example, the set may contain the following entries: Range 1: 0.9-1.0, corresponding to pricing ladder factor 0.8 (low cost); Range 2: 0.7-0.9, corresponding to pricing ladder factor 0.9 (medium cost); Range 3: 0.0-0.7, corresponding to pricing ladder factor 1.1 (high cost).

[0142] These ranges and step factors can be adjusted according to the specific needs of heat supply management. For example, the management agency can set reasonable fees and step factors based on local energy prices, environmental protection policies and the economic affordability of users.

[0143] By obtaining the pricing mechanism index set, the heat forecast management system provides a basis for determining the heat pricing mechanism in the subsequent steps. This set makes the calculation of heating costs more flexible and personalized, and can be dynamically adjusted according to the degree of user behavior norms. This pricing mechanism based on user behavior norms helps to motivate users to save energy and improve the economy and sustainability of heat management.

[0144] Step S55: determining the second heat usage behavior standard coefficient range corresponding to the heat usage behavior standard coefficient in the heating period P1 as the second target heat usage behavior standard coefficient range.

[0145] In step S55, the heat supply forecast management system obtains the heat use behavior standard coefficient in the heating period P1. This coefficient is calculated by the previous steps, and it reflects the degree of standardization of the user's heat use behavior in the P1 period. Next, the heat supply forecast management system matches this coefficient with the second heat use behavior standard coefficient range in the pricing mechanism index set. This process is completed by comparing the coefficient value with the upper and lower limits of each range in the set. Specifically, the heat supply forecast management system will traverse each range in the set and check whether the current coefficient falls within the range. Once a matching range is found, the heat supply forecast management system will determine it as the second target heat use behavior standard coefficient range.

[0146] Assume that the heat usage behavior standard coefficient in the heating period P1 is 0.82. The system will match this coefficient with the range in the set. Since 0.82 falls within range 2 (0.7-0.9), the system will determine range 2 as the second target heat usage behavior standard coefficient range, and use the corresponding step factor 0.9 as the basis for subsequent calculation of heating costs. Through the execution of step S55, the heating forecasting management system can accurately determine the corresponding pricing step factor range based on the user's heat usage behavior standard coefficient, providing a basis for subsequent calculation of heating costs. This range-based matching method not only improves the accuracy and efficiency of the system, but also makes the heating pricing mechanism more flexible and configurable, and can be dynamically adjusted according to the user's degree of behavioral standardization.

[0147] Step S56: Determine the pricing step factor corresponding to the second target heat usage behavior specification coefficient range in the pricing mechanism index set as the target pricing step factor.

[0148] In the embodiment of the present application, step S56 is a key value-taking step, which selects the corresponding pricing ladder factor from the pricing mechanism index set as the target pricing ladder factor based on the second target heat usage behavior specification coefficient range determined in step S55. The following is a detailed explanation of this step in combination with a specific application scenario.

[0149] In step S56, the heat supply forecast management system searches for the corresponding pricing ladder factor in the pricing mechanism index set according to the second target heat use behavior specification coefficient range. This process is completed by comparing the range in the set with the second target range. Once a matching range is found, the heat supply forecast management system determines the pricing ladder factor corresponding to the range as the target pricing ladder factor.

[0150] Assume that the second target heat usage behavior specification coefficient range determined in step S55 is range 2 (0.7-0.9). In step S56, the heat supply forecast management system searches for the step factor corresponding to range 2 in the pricing mechanism index set, which is 0.9. Therefore, the heat supply forecast management system determines 0.9 as the target pricing step factor.

[0151] This target pricing ladder factor will be used for subsequent calculations of heating costs. Specifically, it will be multiplied with the initial heating pricing mechanism corresponding to the heating execution strategy to obtain the final heating pricing mechanism. For example, if the price of a unit of heat in the initial heating pricing mechanism is 1 yuan / kWh, then after applying the target pricing ladder factor of 0.9, the price of a unit of heat will become 0.9 yuan / kWh. In this way, users need to pay according to this price when they actually use the heating service. Through the execution of step S56, the heating forecasting management system can determine a reasonable pricing ladder factor based on the user's heat usage behavior norm coefficient, making the calculation of heating costs more fair and reasonable. This pricing method based on user behavior norms helps to motivate users to save energy and improve the economy and sustainability of heating management.

[0152] Step S57: Obtain the initial heating pricing mechanism corresponding to the heating execution strategy, and multiply the initial heating pricing mechanism by the target pricing step factor to obtain the heating pricing mechanism.

[0153] In the embodiment of the present application, step S57 calculates the final heating pricing mechanism according to the initial heating pricing mechanism and the target pricing ladder factor corresponding to the heating execution strategy. In step S57, the heating forecast management system obtains the initial heating pricing mechanism corresponding to the heating execution strategy. This initial pricing mechanism is pre-set in the heating management strategy, and it includes, for example, elements such as the price of unit heat, fixed fees, and ladder prices, which are used to calculate the user's heating costs. Next, the heating forecast management system multiplies the relevant cost elements in the initial heating pricing mechanism with the target pricing ladder factor. The purpose of this operation is to adjust the calculation method of the heating cost according to the degree of user behavior standardization. Specifically, if the target pricing ladder factor is less than 1, it means that the user's heating use behavior is relatively standardized, and the heating forecast management system will give a certain fee discount; conversely, if the ladder factor is greater than 1, it means that the user's behavior is not standardized, and the heating forecast management system will increase the heating cost to encourage the user to improve his behavior.

[0154] As an implementation method, the target pricing ladder factor may be obtained by estimating based on machine learning instead of using the above steps S54 to S56, which may specifically include the following steps: Step S54': obtaining the reference calorie consumption behavior parameters of the object to be estimated, the reference calorie consumption behavior parameters are used as reference information when determining the pricing step factor.

[0155] In the embodiment of the present application, step S54' obtains the control heat consumption behavior parameters of the object to be estimated. The control heat consumption behavior parameters are the heat consumption data of the object to be estimated in the past heating period. These data record in detail the user's heating use behavior in a specific time period, including key features such as equipment running time, energy consumption, and set temperature. These features reflect the user's heating use habits and behavior patterns, and are an important basis for determining the pricing ladder factor. In step S54', the heating forecast management system obtains the control heat consumption behavior parameters of the object to be estimated by querying the database or calling the relevant interface. These parameters are represented in the form of feature vectors, for example, and each vector element corresponds to a specific behavior feature. For example, a feature vector may contain the following elements: [equipment running time (hours), energy consumption (kWh), set temperature (°C)]. The process of obtaining the control heat consumption behavior parameters needs to ensure the accuracy and completeness of the data. The heating forecast management system may need to preprocess the data, such as missing value filling, outlier processing, etc., to ensure the quality of the input data. In addition, in order to more comprehensively reflect the user's heating usage behavior, the heating forecasting management system may also need to obtain control parameters for multiple heating time periods to form a richer data set.

[0156] Specific examples of heat consumption behavior parameters may include: daily equipment operation time, energy consumption and set temperature data of a user in the past month. These data will be used as inputs of the machine learning model to calculate the pricing ladder factor of the user in the current or future heating period.

[0157] By obtaining the reference heat consumption behavior parameters of the object to be estimated, the heating forecast management system can provide accurate data support for the subsequent pricing ladder factor estimation. This will help the system to more accurately understand the user's heating usage behavior, thereby formulating a more reasonable heating pricing strategy, promoting users to save energy and improving the efficiency of heating management.

[0158] Step S55': using a module dimensions included in the multi-dimensional module in the pricing ladder factor estimation network, deep knowledge extraction is performed based on the control calorie consumption behavior parameters to obtain a dimensional output knowledge; a≥2, in the training stage of the pricing ladder factor estimation network, a module dimensions are respectively used to learn the deep knowledge of the calorie consumption behavior parameters corresponding to different pricing ladder factor intervals.

[0159] In an embodiment of the present application, the pricing step factor estimation network is a pre-trained machine learning model, which is used to infer the corresponding pricing step factor based on the user's calorie consumption behavior parameters. This network is, for example, a complex neural network structure, which contains modules of multiple dimensions (assuming there are a, where a≥2), and each module is responsible for extracting deep knowledge related to a specific pricing step factor interval from the input data. In step S55', the control calorie consumption behavior parameters of the object to be estimated are obtained. These parameters are the calorie consumption data of the user during the past heating period, including key features such as equipment operating time, energy consumption, and set temperature. Then, these control parameters are passed as input data to the pricing step factor estimation network.

[0160] Next, the a module dimensions in the network work in parallel to extract deep knowledge from the input data. Each module dimension corresponds to a specific pricing ladder factor interval, and has learned how to extract deep features related to this interval from the heat consumption behavior parameters during the network training phase. These features may be the distribution of equipment operating time, the trend of energy consumption, the stability of the set temperature, etc., which together constitute a multi-dimensional knowledge representation that describes the user's heating use behavior pattern. Specifically, each module dimension receives the same input data (i.e., the heat consumption behavior parameters), but due to their different internal structures and parameters, they process and analyze the input data from different angles to obtain their corresponding dimensional output knowledge. These output knowledge can be regarded as feature vectors of a channels, each of which contains important information related to the user's heating use behavior pattern.

[0161] For example, assuming that the pricing ladder factor estimation network contains two module dimensions (i.e., a=2), the first dimension is responsible for learning the features corresponding to the low pricing ladder factor interval, while the second dimension corresponds to the high pricing ladder factor interval. When the control heat consumption behavior parameters are passed in, the first dimension may focus on features such as the stability of energy consumption and the rationality of equipment operating time; while the second dimension may pay more attention to features such as the volatility of energy consumption and abnormal values ​​of set temperatures. Through the parallel processing of these two dimensions, the heating forecasting management system can obtain dimensional output knowledge from two different angles, which together reflect the user's heating usage behavior pattern and possible pricing ladder factor intervals.

[0162] By executing step S55', the multi-dimensional module in the pricing ladder factor estimation network can be used to extract deep knowledge from the calorie consumption behavior parameters of the estimated object to obtain multi-dimensional output knowledge. These output knowledge will be used as input for subsequent steps to further determine the target pricing ladder factor estimation value.

[0163] Step S56': according to the interval training adjustment control in the pricing ladder factor estimation network, according to the control calorie consumption behavior parameter, determine the contribution parameters corresponding to the a-dimensional output knowledge respectively.

[0164] In the embodiment of the present application, step S56' determines the contribution of different dimensional output knowledge (or features) to the final pricing ladder factor estimate value according to the interval training adjustment control in the pricing ladder factor estimation network. The pricing ladder factor estimation network is a trained machine learning model, which contains multiple module dimensions for extracting features related to different pricing ladder factor intervals from the control calorie consumption behavior parameters. During the training process, the interval training adjustment control in the network plays a key role. It is responsible for adjusting the network parameters according to the relationship between the features and the target pricing ladder factor to optimize the prediction performance of the model.

[0165] In step S56', the heat supply forecast management system first obtains the control heat consumption behavior parameters of the object to be estimated, and inputs them into the pricing ladder factor estimation network. The a module dimensions in the network process these parameters and output a dimensional feature vectors. These feature vectors contain key information related to different pricing ladder factor intervals. Next, the heat supply forecast management system determines the contribution parameters corresponding to these feature vectors according to the interval training adjustment control in the pricing ladder factor estimation network. The interval training adjustment control evaluates the importance or influence of each feature vector on the final pricing ladder factor estimate value based on the knowledge learned by the network during the training phase. Specifically, the control may analyze the statistical indicators such as the correlation and significance between each feature vector and the target pricing ladder factor, thereby assigning a contribution parameter to it. The contribution parameter reflects the weight or importance of the feature vector in determining the target pricing ladder factor estimate value. A higher contribution parameter means that the feature vector has a greater impact on the estimate value, while a lower contribution parameter means that its impact is smaller. In this way, the heat supply forecast management system can more accurately capture the actual contribution of different feature vectors to the target pricing ladder factor.

[0166] For example, assume that there are two module dimensions (i.e., a=2) in the pricing step factor estimation network, which output feature vectors F1 and F2 respectively. When analyzing these two feature vectors, the interval training adjustment control finds that F1 has a higher correlation with the target pricing step factor, while F2 has a lower correlation. Therefore, the control may assign a higher contribution parameter (such as 0.8) to F1 and a lower contribution parameter (such as 0.2) to F2. This means that when the target pricing step factor estimate is calculated later, the impact of F1 will be greater than that of F2. Through the execution of step S56', the heating forecasting management system can determine the contribution parameters for each dimension output knowledge based on the interval training adjustment control. These parameters will be used as inputs in subsequent steps to calculate the estimate of the target pricing step factor, thereby ensuring the accuracy and rationality of the prediction results.

[0167] Step S57': According to the classifier in the pricing ladder factor estimation network, according to the a-dimensional output knowledge and their corresponding contribution parameters, determine the target coding knowledge corresponding to the object to be estimated.

[0168] In an embodiment of the present application, step S57' utilizes the classifier in the pricing step factor estimation network to integrate multiple dimensional output knowledge and its corresponding contribution parameters, so as to determine the target coded knowledge corresponding to the object to be estimated. In step S57', the heating forecast management system will utilize the classifier in the estimation network to integrate these dimensional output knowledge and contribution parameters. The classifier is, for example, a trained machine learning model component that can make decisions or predictions based on input features and weights. Specifically, the classifier receives a dimensional output knowledge and its corresponding contribution parameters as input. Then, a comprehensive coding representation is calculated based on these inputs, and this coding representation is the target coding knowledge. During the calculation process, the classifier will consider the specific content of each dimensional output knowledge and the size of the contribution parameter to ensure that those features that have a greater impact on the target pricing step factor are more reflected in the coding representation.

[0169] For example, suppose there are two dimensions of output knowledge F1 and F2, which respectively represent the energy consumption stability of the object to be estimated and the rationality of the set temperature. If the contribution parameter of F1 is higher, it means that the energy consumption stability has a greater impact on the pricing ladder factor. Then, when the classifier calculates the target coding knowledge, the features of F1 will get more weight, thus occupying a more important position in the coding representation. Through this process, the heating forecasting management system can obtain a target coding knowledge that integrates multiple dimensional features and contribution parameters. This coding knowledge not only contains the key information of the object to be estimated in terms of heating usage behavior, but also reflects the importance of this information in determining the target pricing ladder factor.

[0170] Step S58': Determine the estimated value of the pricing ladder factor corresponding to the object to be estimated based on the target coding knowledge.

[0171] In an embodiment of the present application, step S58' determines the estimated value of the pricing ladder factor corresponding to the object to be estimated based on the target coding knowledge generated in the previous step. In step S58', the heating forecast management system will use this target coding knowledge to determine the estimated value of the pricing ladder factor corresponding to the object to be estimated. This process, for example, involves a decoder or a regression model, which maps the target coding knowledge to a specific pricing ladder factor value. Specifically, the decoder or regression model will deduce a reasonable estimated value of the pricing ladder factor based on the specific content of the target coding knowledge, which will serve as the basis for subsequent heating cost calculations, and is used to reflect the degree of standardization of the heating use behavior of the object to be estimated during the current or future heating period, and adjust the heating costs accordingly to encourage users to save energy.

[0172] For example, it is assumed that the target coding knowledge is a vector containing multiple features, each of which corresponds to a certain aspect of the heating use behavior characteristics of the object to be estimated (such as energy consumption stability, reasonableness of set temperature, etc.), and each feature has a corresponding weight value (i.e., contribution parameter). The decoder or regression model will receive this feature vector as input and output a specific pricing ladder factor estimate through a series of calculations (such as linear regression, neural network prediction, etc.). For example, if the feature value representing energy consumption stability in the feature vector is high and the weight is also large, it means that the energy consumption behavior of the object to be estimated is relatively standardized; while the feature value representing the reasonableness of the set temperature is low and the weight is small, it means that there are some irregular behaviors in the set temperature. The decoder or regression model will comprehensively consider these features and their weights, and finally output a pricing ladder factor estimate that can reflect the overall behavioral standardization of the object to be estimated. Through step S58', the heating forecast management system can accurately determine the pricing ladder factor estimate corresponding to the object to be estimated based on the target coding knowledge. This estimate is not only based on the actual heating usage behavior data of the estimated object, but also has been deeply analyzed and predicted by the machine learning model, so it has high accuracy and rationality. This will provide strong support for the subsequent heating cost calculation, help promote energy conservation for users and improve the efficiency of heating management.

[0173] In one embodiment, a heat supply forecasting management system is provided. The heat supply forecasting management system may be a server, and its internal structure diagram may be as follows: Figure 3As shown. The heat supply forecast management system includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the heat supply forecast management system is used to provide computing and control capabilities. The memory of the heat supply forecast management system includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the heat supply forecast management system is used to exchange information between the processor and external devices. The communication interface of the heat supply forecast management system is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a heat supply forecast management method based on artificial intelligence is implemented. Those skilled in the art can understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the heat supply forecasting management system to which the scheme of the present application is applied. The specific heat supply forecasting management system may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0174] In one embodiment, a heat supply prediction management system is further provided, comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

Claims

1. A heating estimation management method based on artificial intelligence, characterized in that: include: Obtaining a heat consumption behavior parameter of the object to be estimated in a heating period P1 and a heat consumption behavior parameter in a heating period P2, wherein the heating period P2 is a heating period before the heating period P1; Based on the heat consumption behavior parameter in the heating period P1 and the heat consumption behavior parameter in the heating period P2, estimating the estimated heat usage behavior standard coefficient of the object to be estimated in the heating period P1; Obtaining a heat usage behavior standard coefficient of the object to be estimated in the heating period P2, and determining a heat usage behavior standard coefficient of the object to be estimated in the heating period P1 based on the heat usage behavior standard coefficient in the heating period P2 and the estimated heat usage behavior standard coefficient in the heating period P1; The heat usage behavior standard coefficient in the heating period P1 is configured to determine the heating scheduling mode for the heating energy usage of the object to be estimated in the heating period P1; the heat usage behavior standard coefficient in the heating period P1 is configured to determine the heat usage behavior standard coefficient of the object to be estimated in the heating period P3 together with the estimated heat usage behavior standard coefficient in the heating period P3, and the heating period P3 is the period after the heating period P1; The estimated heat usage behavior norm coefficient in the heating period P3 is determined by the heat consumption behavior parameter of the object to be estimated in the heating period P3 and the heat consumption behavior parameter in the heating period P1.

2. The method according to claim 1, characterized in that The estimating the estimated heat usage behavior standard coefficient of the object to be estimated in the heating period P1 based on the heat consumption behavior parameter in the heating period P1 and the heat consumption behavior parameter in the heating period P2 includes: The heat consumption behavior parameters in the heating period P1 are transmitted to the heat use behavior recognition network; The heat use behavior recognition network is obtained by optimizing the heat use behavior recognition network to be trained through the heat consumption behavior parameters in the heating period P2 and the normative coefficient markers of the behaviors that have occurred in the heating period P2; The normative coefficient mark of the behavior that has occurred in the heating period P2 is determined by the statistics of the irregular behavior that has occurred in the heat consumption behavior parameter in the heating period P2; Obtaining a network weight corresponding to the heating period P2 in the heat usage behavior recognition network, and obtaining a weight second-order tensor corresponding to the network weight; The behavior parameter second-order tensor corresponding to the heat consumption behavior parameter in the heating period P1 is obtained, and the behavior parameter second-order tensor is obtained by performing a tensor product with the weight second-order tensor to obtain the estimated heat usage behavior standard coefficient of the object to be estimated in the heating period P1.

3. The method according to claim 1, characterized in that The determining of the heat usage behavior standard coefficient of the object to be estimated in the heating period P1 based on the heat usage behavior standard coefficient in the heating period P2 and the estimated heat usage behavior standard coefficient in the heating period P1 includes: Obtaining a first contribution factor corresponding to the heat usage behavior standard coefficient in the heating period P2, and a second contribution factor corresponding to the estimated heat usage behavior standard coefficient in the heating period P1; The heat usage behavior standard coefficient in the heating period P2 is multiplied by the first contribution factor to obtain a first regulated heat usage behavior standard coefficient; The estimated heat usage behavior standard coefficient in the heating period P1 is multiplied by the second contribution factor to obtain a second regulated heat usage behavior standard coefficient; The first regulating heat usage behavior standard coefficient and the second regulating heat usage behavior standard coefficient are summed to obtain the heat usage behavior standard coefficient of the object to be estimated in the heating period P1.

4. The method according to claim 1, characterized in that The method further comprises: In the heating period P1, a heating energy use application sent by a device to be predicted is obtained, and the device to be predicted is a device corresponding to the object to be estimated; Based on the heating energy use application, obtaining the heat use behavior standard coefficient in the heating period P1, and determining the heating execution strategy of the object to be estimated and the heating pricing mechanism corresponding to the heating execution strategy based on the heat use behavior standard coefficient in the heating period P1; The heating execution strategy and the heating pricing mechanism are used as a heating scheduling method for the heating energy usage of the object to be estimated, and the heating scheduling method is sent to the device of the object to be estimated; Comparing the heat usage behavior standard coefficient in the heating period P1 with the reference heat usage behavior standard coefficient; When the heat usage behavior standard coefficient in the heating period P1 is less than the reference heat usage behavior standard coefficient, heat rejection information is generated and sent to the device of the object to be predicted, and the heat rejection information is used to inform the object to be predicted that the heat usage behavior standard coefficient in the heating period P1 does not meet the heating requirements; When the heat usage behavior norm coefficient in the heating period P1 is not less than the reference heat usage behavior norm coefficient, the step of determining the heating execution strategy of the object to be estimated and the heating pricing mechanism corresponding to the heating execution strategy based on the heat usage behavior norm coefficient in the heating period P1 is executed.

5. The method according to claim 4, characterized in that The method of determining the heating execution strategy of the object to be estimated based on the heat usage behavior norm coefficient in the heating period P1, and the heating pricing mechanism corresponding to the heating execution strategy, includes: Acquire a heating strategy index set, wherein the heating strategy index set includes a matching condition between a first heat usage behavior specification coefficient range and a heating execution strategy, and each first heat usage behavior specification coefficient range corresponds to a heating execution strategy; Determine a first heat usage behavior standard coefficient range corresponding to the heat usage behavior standard coefficient in the heating period P1 as a first target heat usage behavior standard coefficient range; Determine the heating execution strategy corresponding to the first target heat usage behavior specification coefficient range in the heating strategy index set as the heating execution strategy; Acquire a pricing mechanism index set, wherein the pricing mechanism index set includes a matching condition between the second heat usage behavior specification coefficient range and the pricing ladder factor, and each second heat usage behavior specification coefficient range corresponds to a pricing ladder factor; Determine a second heat usage behavior standard coefficient range corresponding to the heat usage behavior standard coefficient in the heating period P1 as a second target heat usage behavior standard coefficient range; Determine the pricing ladder factor corresponding to the second target heat usage behavior specification coefficient range in the pricing mechanism index set as the target pricing ladder factor; An initial heat supply pricing mechanism corresponding to the heat supply execution strategy is obtained, and the initial heat supply pricing mechanism is multiplied by the target pricing step factor to obtain the heat supply pricing mechanism.

6. The method according to claim 2, characterized in that The method further comprises: Obtaining the heat consumption behavior parameters of the object to be estimated in the heating period P2 and the statistics of the irregular behaviors that have occurred in the heat consumption behavior parameters in the heating period P2; Generate a behavior norm coefficient mark of the object to be estimated in the heating period P2 based on the statistics of the irregular behaviors that have occurred; Based on the heat consumption behavior parameters in the heating period P2 and the normative coefficient marks of the behaviors that have occurred in the heating period P2, the heat use behavior recognition network to be trained is trained to obtain the heat use behavior recognition network.

7. The method according to claim 6, characterized in that The generating of the behavior norm coefficient mark of the object to be estimated in the heating period P2 based on the statistics of the irregular behaviors that have occurred includes: Obtaining statistics of irregular behaviors of a plurality of existing objects in the heating period P2, wherein the plurality of existing objects are objects that generate heat consumption in the heating period P2, and the plurality of existing objects include the object to be estimated; Determine the number of objects corresponding to the plurality of existing objects, and determine the average number of unplanned behaviors in the heating period P2 based on the statistics of the unstandardized behaviors and the number of objects; Based on the statistics of the irregular behaviors that have occurred and the average number of unplanned behaviors, a normative coefficient mark of the behavior that has occurred of the object to be estimated in the heating period P2 is determined.

8. The method according to claim 7, characterized in that The determining, based on the statistics of the irregular behaviors that have occurred and the average number of irregular behaviors, of the normative coefficient mark of the behavior that has occurred in the heating period P2 of the object to be estimated includes: Comparing the statistics of the irregular behaviors that have occurred with the average number of irregular behaviors; When the statistic of the irregular behaviors that have occurred is greater than or equal to the average number of irregular behaviors, a first preset value is used as a normative coefficient mark of the behavior that has occurred of the object to be estimated in the heating period P2, and the first preset value is used to indicate that the heating use behavior of the object to be estimated in the heating period P2 is irregular behavior; When the statistic of the irregular behaviors that have occurred is less than the average number of unplanned behaviors, a second preset value is used as a normative coefficient mark for the behavior that has occurred of the object to be estimated in the heating period P2, and the second preset value is used to indicate that the heating usage behavior of the object to be estimated in the heating period P2 is a normative behavior.

9. The method according to claim 6, characterized in that The method of training the heat consumption behavior recognition network to be trained based on the heat consumption behavior parameter in the heating period P2 and the behavior norm coefficient mark that has occurred in the heating period P2 to obtain the heat consumption behavior recognition network includes: The heat consumption behavior parameters in the heating period P2 are transmitted to the heat usage behavior recognition network to be trained, and the estimated heat usage behavior standard coefficient of the object to be estimated in the heating period P2 is estimated by using the training network weights in the heat usage behavior recognition network to be trained and the heat consumption behavior parameters in the heating period P2; Determining a network training error based on the estimated heat usage behavior specification coefficient in the heating period P2 and the occurred behavior specification coefficient mark in the heating period P2; The training network weights are optimized based on the network training errors to obtain network weights, and the heat usage behavior recognition network to be trained including the network weights is used as the heat usage behavior recognition network.

10. A heat supply forecasting and management system, characterized in that: include: processor; and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the method as claimed in any one of claims 1 to 9.

Images