Heat supply charging system and method based on convolutional neural network

Abstract

The invention discloses a heat supply charging system and method based on a convolutional neural network, and relates to the technical field of heat supply management and control, and the system comprises a convolutional neural network analysis module, a configuration module, a distributed computing server and a heat trend analysis module. The convolutional neural network analysis module, the configuration module and the heat trend analysis module are integrated in a distributed computing server; and the convolutional neural network analysis module constructs a convolutional neural network model, and the convolutional neural network analysis module performs deep learning on a heat consumption mode of a user by using a large amount of historical data to establish a nonlinear mapping relationship between the heat supply amount and influence factors. According to the method, accurate user behavior analysis is realized by using the convolutional neural network, the problem that the general classification of a traditional heat supply charging system cannot clearly reflect the actual heat consumption behavior difference of each user individual is solved, the user experience is improved, and a more personalized charging scheme and accurate heat consumption suggestions are provided for each user.

Description

Technical Field

[0001] The present invention relates to the technical field of heat supply control and management, and in particular to a heat supply charging system and method based on a convolutional neural network. Background Art

[0002] With the rapid development of science and technology, intelligence has become an important trend in social development. In various fields, such as transportation, medical care, education, etc., intelligent technology is gradually penetrating and changing people's lifestyles. As part of urban infrastructure, the heating fee collection system is also facing the need for intelligent transformation. As a major energy consumer, the intelligent management of the heating system is of great significance to achieving energy conservation and emission reduction goals.

[0003] As an important algorithm in the field of deep learning, convolutional neural network has achieved remarkable results in image processing, speech recognition and other fields. Its powerful feature extraction and classification capabilities provide strong support for the intelligentization of heating charging system. Through machine learning algorithms such as convolutional neural network, users' heating demand can be more accurately predicted and calculated, so as to achieve precise heating. This can not only improve the comfort and satisfaction of heating, but also reduce energy waste. However, it is necessary to collect multi-source data and integrate them. The convolutional neural network model has a complex structure and is difficult to train. At the same time, it is necessary to improve users' acceptance and trust in the intelligent charging system.

[0004] Therefore, it is necessary for this application to propose a heating fee collection system and method based on convolutional neural networks to solve the problems in the background.

[0005] Patent document CN111578368B discloses a heating network hydraulic regulation system and its intelligent control and heating billing method. The above patent overcomes the problems of hydraulic imbalance in the heating network, hydraulic imbalance in buildings and hydraulic imbalance in users, as well as inaccurate charging caused by many existing heating systems. The heating network hydraulic regulation system includes a primary network water supply regulating valve, a secondary network water supply pressure sensor, a secondary network variable frequency circulation pump, a front-of-building water supply variable frequency circulation pump, a front-of-building water supply flow meter, a front-of-building water supply temperature sensor, a front-of-building return water temperature sensor, an indoor water supply regulating valve, an indoor temperature sensor, a front-of-building pressure difference sensor, a front-of-building metering control device and a remote control device. However, the above patent cannot achieve accurate user behavior analysis.

[0006] Patent document CN104978795B discloses a fully dynamic energy-saving control management method and system based on IC card prepaid heating. The above patent realizes an IC card reading device, a prepaid indoor thermostat and an intelligent dynamic controller for a temperature control valve, and has the advantages of convenient operation, dynamic control, fair charging, energy saving, and prevention of heat loss. However, the above patent cannot realize the introduction of a "basic + on-demand" charging model.

[0007] Patent document CN114023002B discloses a household heating billing valve control system and method. The above patent realizes that the existing technology does not take into account the heat conduction effect of adjacent users, the calculation is not accurate enough, and the user's heating data cannot be guaranteed to be accurate. The energy consumption of heating is seriously wasted, which increases the burden of operation and regulation of the heating pipeline network. By obtaining user needs, personalized heating services can be provided, which are manifested in users setting their own room temperature and heating time; considering the heat conduction effect of adjacent users, accurately calculating the heat consumption of each user and the relationship with the water flow, and ensuring the accuracy of the user's heating data; charging by heat can encourage each user to reduce the energy consumption of heating as much as possible while ensuring their own heating comfort, and can also promote the demand for energy-saving buildings, save energy and reduce emissions, enhance resource awareness, and establish a healthy lifestyle. However, the above patent cannot achieve association with external systems to obtain data that affects heating.

[0008] Patent document CN103631240B discloses a heat metering information application service management platform for the heating industry based on the Internet of Things. The above patent realizes the ability to statistically monitor the operating status of heat meters, analyze heating conditions, and provide management and information services, but the above patent cannot realize real-time monitoring of the energy consumption of the entire heating system.

[0009] In summary, the above patents cannot achieve accurate user behavior analysis, real-time monitoring of the energy consumption of the entire heating system, linking with external systems to obtain data affecting heating, and launching a "basic + on-demand" charging model, resulting in large heat loss in pipelines or excessive heating in certain areas, users are confused about the payment situation, waste resources, and the general classification of the traditional heating charging system cannot clearly reflect the actual differences in heating behavior of each individual user; To this end, this application proposes a heating charging system and method based on a convolutional neural network that can achieve accurate user behavior analysis, real-time monitoring of the energy consumption of the entire heating system, link with external systems to obtain data that affects heating, and launch a "basic + on-demand" charging model. Summary of the invention

[0010] The purpose of the present invention is to provide a heating fee charging system and method based on convolutional neural networks to solve the technical problems proposed in the above background technology, such as the inability to achieve accurate user behavior analysis, real-time monitoring of the energy consumption of the entire heating system, association with external systems to obtain data affecting heating and the introduction of a "basic + on-demand" charging model, resulting in large heat loss in pipelines or excessive heating in certain areas, users' unclear payment situation, waste of resources and the general classification of traditional heating fee charging systems that cannot clearly reflect the actual differences in heating behavior of each individual user.

[0011] To achieve the above object, the present invention provides the following technical solution: a heating charging system based on convolutional neural network, comprising: a convolutional neural network analysis module, a configuration module, a distributed computing server and a heat trend analysis module; the convolutional neural network analysis module is wirelessly combined with the heat trend analysis module, and the convolutional neural network analysis module, the configuration module and the heat trend analysis module are integrated in the distributed computing server; The convolutional neural network analysis module constructs a convolutional neural network model. The convolutional neural network analysis module uses a large amount of historical data to perform in-depth learning on the user's heat usage pattern and establish a nonlinear mapping relationship between the heat supply and the influencing factors; The nonlinear mapping relationship between the heating supply and the influencing factors: After multiple convolution and pooling layers, the data is flattened into a vector F, and the weight matrix is ​​W fc , nonlinear activation function (•), the bias vector is B fc (size is n), heat supply Q, output vector of the fully connected layer , the weight vector is W Q , with a bias of b Q , Q=W Q +b Q =W Q (W fc F+B fc )+b Q .

[0012] Preferably, the heating charging system based on convolutional neural network also includes: a parameter input module, a charging calculation module, a fee output module, an energy situation awareness module, an external system association module and a personalized interaction module; The parameter input module provides data to the convolutional neural network analysis module and the configuration module, and the convolutional neural network analysis module receives the data from the parameter input module; Energy situation awareness module: real-time awareness of the energy consumption of the entire heating system; External system association module: associate with external systems, obtain the operating parameters of heating equipment, and discover the increase in energy consumption caused by equipment failure / abnormal operation; Personalized interactive module: Provide users with suggestions on temperature settings and heating time adjustment, formulate personalized charging strategies and detailed fee details, and provide certain fee discounts for users with lower heating demand.

[0013] Preferably, the parameter input module is used to receive the user's heating demand information and determine the user parameters and environmental parameters of heating; The parameter input module includes: a parameter sensing unit and a parameter conditioning unit; Cost output module: used to provide users with heating and charging information and generate mobile heating fee bills; The expense output module includes a customer service unit, which is used to connect with users and provide personalized solutions.

[0014] Preferably, the convolutional neural network analysis module extracts period intervals and peak features from the user's heat usage pattern by analyzing time series data, and the convolutional neural network analysis module uses thermal imaging images to identify areas of heat loss in the building; To extract periodic intervals and peak features, the preprocessed heating charge data is input into the convolution layer. Each convolution kernel in the convolution layer will slide on the data to perform convolution operations. The first layer of convolution kernels extracts the periodic features in the heat usage data, and the second layer of convolution kernels extracts the comprehensive features between the user's heat usage behavior pattern and payment pattern.

[0015] Preferably, the charging calculation module supports time-sharing billing, tiered billing, and "basic + on-demand" mode; The fee calculation module includes: User information unit: records the heating mode selected by the user and associates the basic data affecting heating; Metering component: responsible for recording the consumer's calorie consumption; Billing component: Calculates the final fee by setting rules, combining user base and flow data; supports multiple billing modes, including "basic + on-demand" mode, the formula is: Total cost = basic heating cost + heating cost on demand Basic heating cost = basic heating unit price × basic heat consumption On-demand heating cost = on-demand heating unit price × on-demand heat consumption.

[0016] Preferably, the distributed computing server is used to build a distributed training and prediction environment for a convolutional neural network; Distributed computing servers include: Computing node cluster: Each computing node has its own independent memory, which is used to store the parameters, input data and intermediate calculation results of the convolutional neural network; Storage node: used to save user heat usage and payment records; Distributed computing framework: Distribute the computing tasks of convolutional neural networks to different computing nodes for parallel computing.

[0017] Preferably, the configuration module is used to retrieve the convolutional neural network model from a preset model database and load it into the model computing space of the distributed computing server, and supports dynamic updating of the configuration; Heat trend analysis module: used to predict the heat supply based on the information of the convolutional neural network model and parameter input module.

[0018] Preferably, the personalized interaction module includes: Cost warning unit: When the user's on-demand heating cost approaches the user's preset budget / exceeds a certain proportion, the user is reminded via SMS or APP message; the cost warning unit is linked to the fee calculation module.

[0019] Preferably, the method of use comprises the following steps: (1) Parameter input: Users submit their heating demand information through the parameter input module and associated external sensors; (2) Convolutional neural network model training: The convolutional neural network analysis module uses the preprocessed data to build and train the convolutional neural network model; (3) Heat supply trend analysis: The heat supply trend analysis module uses the trained convolutional neural network model, combined with real-time input data and external factors, to determine the user's heat supply demand in the future and provide a basis for the formulation of the heating plan; (4) Charging model selection: Users can select a suitable charging model based on their own needs through the configuration module to meet their personalized needs; (5) Cost calculation: The fee calculation module calculates the user's heating fee based on the predicted heat supply and actual heat supply, combined with the heating company's charging standards and the charging model selected by the user; (6) Cost warning and notification: When the user's on-demand heating costs are close to the preset budget / exceed a certain proportion, the cost warning unit of the personalized interactive module will promptly remind the user through SMS, APP messages, etc.

[0020] Preferably, the method of use further comprises the following steps: (7) Mobile bill generation: After the heat supply trend analysis and charge calculation are completed, the fee output module provides users with detailed heat supply and charge information and generates a mobile heat supply charge bill; (8) Heat data adjustment: The parameter conditioning unit adjusts the data required by the charging system, unifies the data scale, and extracts key features that affect heat supply and charging; (9) Energy dynamic tracking: The energy situation awareness module monitors the energy consumption of the entire heating system in real time; (10) Equipment operation monitoring: The external system association module is linked with the monitoring system of the heating equipment to obtain the equipment's operating parameters and promptly detect the increase in energy consumption caused by equipment failure / abnormal operation; (11) Personalized heating recommendations: The personalized interaction module provides users with recommendations on temperature settings and heating time adjustments based on their heating habits and historical data. The customer service unit helps users strike a balance between comfort and energy saving.

[0021] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention uses convolutional neural networks to achieve accurate user behavior analysis, solves the problem that the general classification of the traditional heating charging system cannot clearly reflect the actual heating behavior differences of each individual user, improves the user experience, and provides each user with a more personalized charging plan and accurate heating suggestions; 2. The present invention realizes real-time monitoring of the energy consumption of the entire heating system through the energy situation awareness module, solves the problem of large heat loss in pipelines or excessive heating in certain areas, and the heating company can find the links of energy waste, check the conditions of heating equipment, pipeline insulation, etc. in time, and take corresponding energy-saving measures; 3. The present invention uses an external system association module to associate with an external system to obtain data that affects heating, thereby solving the problem of ambiguous energy consumption charges and improving user satisfaction; 4. The present invention realizes the introduction of a "basic + on-demand" charging model through a heat trend analysis module, solves the problem of users being unclear about payment conditions and wasting resources, classifies users more accurately, formulates charging standards that are more in line with the characteristics of user groups, and improves the rationality of charging and market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a schematic diagram of the structure of the heating charging system of the present invention; Figure 2 It is a schematic diagram of the structure of the use process of the heating charging system of the present invention; Figure 3 Schematic diagram of the convolutional neural network structure of the present invention; Figure 4 It is a schematic diagram of the structure of the distributed computer of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] See also Figure 1 and Figure 2The present invention provides an embodiment: a heating charging system based on a convolutional neural network, comprising: a convolutional neural network analysis module, a configuration module, a distributed computing server and a heat trend analysis module; the convolutional neural network analysis module is wirelessly combined with the heat trend analysis module, and the convolutional neural network analysis module, the configuration module and the heat trend analysis module are integrated in the distributed computing server; the convolutional neural network analysis module constructs a convolutional neural network model, and the convolutional neural network analysis module uses a large amount of historical data to perform in-depth learning on the user's heat usage pattern, and establishes a nonlinear mapping relationship between the heat supply and the influencing factors; The nonlinear mapping relationship between the heating supply and the influencing factors: After multiple convolution and pooling layers, the data is flattened into a vector F, and the weight matrix is ​​W fc , nonlinear activation function (•), the bias vector is B fc (size is n), heat supply Q, output vector of the fully connected layer , the weight vector is W Q , with a bias of b Q , Q=W Q +b Q =W Q (W fc F+B fc )+b Q ; The configuration module is responsible for managing and maintaining the operating parameters, environment variables, database connection information, etc. required by the service. By centrally managing these configurations, the configuration module can ensure the correct configuration and consistency of the service, support dynamic updates of configurations, and modify configuration information without restarting the service, and the configuration information will take effect in real time; Furthermore, the parameter input module receives room size, insulation status, outdoor temperature, humidity and behavior data, and performs data fusion processing. The convolutional neural network analysis module can mine the complex relationship between the data. When the weather is about to get cold, it can determine whether there are elderly people and children at home, which users may maintain the indoor temperature through other means, and predict which users' heating demand will increase significantly. The heating company can adjust the allocation of heating resources in advance and arrange energy supply reasonably. The distributed computing server builds a distributed training and prediction environment for the convolutional neural network, and deploys the model computing space in the server node; the configuration module retrieves the convolutional neural network model from the preset model database according to the user's instructions, and loads it into the model computing space of the distributed computing server, and supports dynamic updating of the configuration; the heat trend analysis module predicts the heat supply in the model computing space through the distributed computing server according to the convolutional neural network model and the parameter input module; the charge calculation module calculates the user's heating fee based on the heat supply provided by the heat trend analysis module and the actual heat supply, combined with the charging standards of the heating company; after the heat supply prediction and charge calculation are completed, the charge output module provides the user with heat supply and charge information, and generates a flowing heating charge bill.

[0025] See also Figure 3 , an embodiment provided by the present invention: a heating charging system based on a convolutional neural network, the parameter input module: used to receive the user's heating demand information, determine the user parameters and environmental parameters of the heating; the parameter input module includes: a parameter sensing unit and a parameter conditioning unit; the fee output module: used to provide the user with heating and charging information, and generate a flowing heating fee bill; the fee output module includes a customer service unit, and the customer service unit is used to connect with the user to provide personalized solutions; The convolutional neural network analysis module extracts periodic intervals and peak features from the user's heat usage pattern by analyzing time series data. The convolutional neural network analysis module uses thermal imaging images to identify areas of heat loss in buildings. The periodic intervals and peak features are extracted by inputting the preprocessed heating charge data into the convolution layer. Each convolution kernel in the convolution layer slides on the data to perform a convolution operation. The first convolution kernel extracts the periodic features in the heat usage data, and the second convolution kernel extracts the comprehensive features between the user's heat usage behavior pattern and the payment pattern. Furthermore, sensor networks are used to capture multi-dimensional data, edge computing technology is applied to perform preliminary screening and compression of data, the collected parameters are conditioned and standardized, the data are converted to a unified dimension, and a data cleaning algorithm is introduced to eliminate outliers; the customer service unit provides personalized expense reports and energy-saving suggestions by analyzing user data; the first layer of convolution kernels focuses on the periodic characteristics of heat usage data, and by setting appropriate kernel size and step size, the time intervals of user habits are identified, multi-level feature extraction is performed, and user behavior is deeply understood; thermal imaging technology is used to identify specific areas of heat loss in buildings, and convolutional neural networks are combined to automatically mark and classify heat loss in different areas, and generate detailed reports to help users optimize the building's insulation performance and heat use strategies.

[0026] See also Figure 4, an embodiment provided by the present invention: a heating charging system based on convolutional neural network, the distributed computing server is used to build a distributed training and prediction environment of convolutional neural network; the distributed computing server includes: a computing node cluster: each computing node has its own independent memory, and the independent memory is used to store the parameters, input data and intermediate calculation results of the convolutional neural network; a storage node: used to save the user's heat use and payment records; a distributed computing framework: the computing tasks of the convolutional neural network are assigned to different computing nodes for parallel computing; the charging calculation module supports time-sharing billing, tiered billing, and "basic + on-demand" mode; the charging calculation module includes: a user information unit: recording the user's selected heating mode, and associating the heating basic data; a metering component: responsible for recording the consumer's heat consumption; a billing component: through rule setting, combined with the user base and flow data, the final fee is calculated; multiple billing modes are supported, including the "basic + on-demand" mode, and the formula is: Total cost = basic heating cost + heating cost on demand Basic heating cost = basic heating unit price × basic heat consumption On-demand heating cost = on-demand heating unit price × on-demand heat consumption; Furthermore, the basic fee is calculated by area, the basic heating unit price × the user's house heating area = the basic heating fee; the basic fee is calculated according to a fixed amount: a fixed basic fee is set; Calculated based on additional heat consumption: on-demand heating unit price × the amount of heat consumed by the user beyond the basic heating range = heating cost for the on-demand portion; Calculated based on additional temperature demand and duration: the additional temperature increase required by the user × the duration of the additional temperature demand × the charging coefficient per unit temperature-time = the heating cost of the on-demand portion.

[0027] See also Figure 1 , an embodiment provided by the present invention: a heating charging system based on convolutional neural network, the heating charging system based on convolutional neural network also includes: a parameter input module, a charging calculation module, a fee output module, an energy situation perception module, an external system association module and a personalized interaction module; the parameter input module provides data to the convolutional neural network analysis module and the configuration module, and the convolutional neural network analysis module receives data from the parameter input module; the energy situation perception module: real-time perception of the energy consumption of the entire heating system; the external system association module: associates with the external system, obtains the operating parameters of the heating equipment, and discovers the increase in energy consumption caused by equipment failure / abnormal operation; the personalized interaction module: provides users with suggestions on temperature setting and heating time adjustment, formulates personalized charging strategies and detailed fee details, and gives certain fee discounts to users with lower heating demand; Furthermore, sensors installed at key nodes collect data including temperature and flow, and use convolutional neural networks to analyze data in real time to quickly identify changing trends in energy consumption. The energy situation awareness module monitors the energy consumption of the entire heating system in real time, and the external system association module interacts with the external system to obtain the operating parameters of the heating equipment. By connecting to the API interface of the equipment manufacturer or management system, the operating status of the equipment can be monitored in real time. When equipment failure or abnormal operation is detected, timely warnings are issued and the increase in energy consumption caused by the abnormality is recorded. The personalized interaction module proposes energy-saving suggestions and personalized charging strategies based on the temperature settings and heating time preferences entered by the user, combined with real-time data analysis. Through machine learning algorithms, the system can identify users' heating patterns, provide fee discounts for users with lower heating needs, and improve user satisfaction.

[0028] See also Figure 2 , an embodiment provided by the present invention: a method for using a heating charging system based on a convolutional neural network, comprising the following steps: (1) Parameter input: Users submit their heating demand information through the parameter input module and associated external sensors; (2) Convolutional neural network model training: The convolutional neural network analysis module uses the preprocessed data to build and train the convolutional neural network model; (3) Heat supply trend analysis: The heat supply trend analysis module uses the trained convolutional neural network model, combined with real-time input data and external factors, to determine the user's heat supply demand in the future and provide a basis for the formulation of the heating plan; (4) Charging model selection: Users can select a suitable charging model based on their own needs through the configuration module to meet their personalized needs; (5) Cost calculation: The fee calculation module calculates the user's heating fee based on the predicted heat supply and actual heat supply, combined with the heating company's charging standards and the charging model selected by the user; (6) Cost warning and notification: When the user's on-demand heating costs are close to the preset budget / exceed a certain proportion, the cost warning unit of the personalized interactive module will promptly remind the user through SMS, APP messages, etc.

[0029] (7) Mobile bill generation: After the heat supply trend analysis and charge calculation are completed, the fee output module provides users with detailed heat supply and charge information and generates a mobile heat supply charge bill; (8) Heat data adjustment: The parameter conditioning unit adjusts the data required by the charging system, unifies the data scale, and extracts key features that affect heat supply and charging; (9) Energy dynamic tracking: The energy situation awareness module monitors the energy consumption of the entire heating system in real time; (10) Equipment operation monitoring: The external system association module is linked with the monitoring system of the heating equipment to obtain the equipment's operating parameters and promptly detect the increase in energy consumption caused by equipment failure / abnormal operation; (11) Personalized heating suggestions: The personalized interaction module provides users with suggestions on temperature settings and heating time adjustments based on their heating habits and historical data; Furthermore, it receives the user's heating demand information, determines the user parameters and environmental parameters of heating, processes data, builds and trains a convolutional neural network model, calculates the user's heating costs, adopts a flexible charging strategy based on the user's actual heating usage, and the customer service provides heating suggestions and energy-saving tips.

[0030] Working principle: users submit their heating demand information through the parameter input module and associated external sensors. The convolutional neural network analysis module uses the preprocessed data to build and train the convolutional neural network model. The heat trend analysis module uses the trained convolutional neural network model, combined with real-time input data and external factors, to determine the user's heat demand in the future and provide a basis for the formulation of the heating plan. Users select a suitable charging mode according to their own needs through the configuration module. The charging calculation module calculates the user's heating fee based on the predicted heat supply and actual heat supply, combined with the charging standards of the heating company and the charging mode selected by the user. When the user's on-demand heating fee is close to the preset budget / exceeds a certain proportion, the fee warning unit of the personalized interaction module will promptly remind the user through SMS, APP messages, etc. After the heat supply trend analysis and fee calculation are completed, the fee output module provides users with detailed heat supply and fee information and generates a mobile heating fee bill. The personalized interaction module provides users with suggestions for temperature setting and heating time adjustment based on the user's heating habits and historical data. The external system association module is linked with the monitoring system of the heating equipment to obtain the operating parameters of the equipment and promptly discover the increase in energy consumption caused by equipment failure / abnormal operation.

[0031] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A heating charging system based on convolutional neural network, characterized in that , including: a convolutional neural network analysis module, a configuration module, a distributed computing server and a heat trend analysis module; the convolutional neural network analysis module is wirelessly combined with the heat trend analysis module, and the convolutional neural network analysis module, the configuration module and the heat trend analysis module are integrated in the distributed computing server; The convolutional neural network analysis module constructs a convolutional neural network model. The convolutional neural network analysis module uses a large amount of historical data to perform in-depth learning on the user's heat usage pattern and establish a nonlinear mapping relationship between the heat supply and the influencing factors; The nonlinear mapping relationship between the heating supply and the influencing factors: After multiple convolution and pooling layers, the data is flattened into a vector F, and the weight matrix is ​​W fc , nonlinear activation function (•), the bias vector is B fc (size is n), heat supply Q, output vector of the fully connected layer , the weight vector is W Q , with a bias of b Q , Q=W Q +b Q =W Q (W fc F+B fc )+b Q .

2. A heating charging system based on convolutional neural network according to claim 1, characterized in that: The heating charging system based on convolutional neural network also includes: a parameter input module, a charging calculation module, a fee output module, an energy situation awareness module, an external system association module and a personalized interaction module; the parameter input module provides data to the convolutional neural network analysis module and the configuration module, and the convolutional neural network analysis module receives data from the parameter input module; Energy situation awareness module: real-time awareness of the energy consumption of the entire heating system; External system association module: associate with external systems, obtain the operating parameters of heating equipment, and discover the increase in energy consumption caused by equipment failure / abnormal operation; Personalized interactive module: Provide users with suggestions on temperature settings and heating time adjustment, formulate personalized charging strategies and detailed fee details, and provide certain fee discounts for users with lower heating demand.

3. A heating charging system based on convolutional neural network according to claim 2, characterized in that: The parameter input module is used to receive the user's heating demand information and determine the user parameters and environmental parameters of the heating; The parameter input module includes: a parameter sensing unit and a parameter conditioning unit; Cost output module: used to provide users with heating and charging information and generate mobile heating fee bills; The expense output module includes a customer service unit, which is used to connect with users and provide personalized solutions.

4. The heating charging system based on convolutional neural network according to claim 1 is characterized in that: The convolutional neural network analysis module extracts period intervals and peak features from the user's heat usage pattern by analyzing time series data, and the convolutional neural network analysis module uses thermal imaging images to identify areas of heat loss in the building; To extract periodic intervals and peak features, the preprocessed heating charge data is input into the convolution layer. Each convolution kernel in the convolution layer will slide on the data to perform convolution operations. The first layer of convolution kernels extracts the periodic features in the heat usage data, and the second layer of convolution kernels extracts the comprehensive features between the user's heat usage behavior pattern and payment pattern.

5. The heating charging system based on convolutional neural network according to claim 2 is characterized in that: The charging calculation module supports time-sharing billing, tiered billing, and "basic + on-demand" mode; The fee calculation module includes: User information unit: records the heating mode selected by the user and associates the basic data affecting heating; Metering component: responsible for recording the consumer's calorie consumption; Billing component: Calculates the final fee by setting rules, combining user base and flow data; supports multiple billing modes, including "basic + on-demand" mode, the formula is: Total cost = basic heating cost + heating cost on demand Basic heating cost = basic heating unit price × basic heat consumption On-demand heating cost = on-demand heating unit price × on-demand heat consumption.

6. The heating charging system based on convolutional neural network according to claim 1 is characterized in that: The distributed computing server is used to build a distributed training and prediction environment for convolutional neural networks; Distributed computing servers include: Computing node cluster: Each computing node has its own independent memory, which is used to store the parameters, input data and intermediate calculation results of the convolutional neural network; Storage node: used to save user heat usage and payment records; Distributed computing framework: Distribute the computing tasks of convolutional neural networks to different computing nodes for parallel computing.

7. The heating charging system based on convolutional neural network according to claim 1 is characterized in that: The configuration module is used to retrieve the convolutional neural network model from a preset model database, load it into the model computing space of the distributed computing server, and support dynamic update of the configuration; Heat trend analysis module: used to predict the heat supply based on the information of the convolutional neural network model and parameter input module.

8. The heating charging system based on convolutional neural network according to claim 2 is characterized in that: The personalized interaction module includes: Cost warning unit: When the user's on-demand heating cost approaches the user's preset budget / exceeds a certain proportion, the user is reminded via SMS or APP message; the cost warning unit is linked to the fee calculation module.

9. A method for using a heating charging system based on a convolutional neural network, applicable to a heating charging system based on a convolutional neural network according to any one of claims 1 to 8, characterized in that: The method of use comprises the following steps: (1) Parameter input: Users submit their heating demand information through the parameter input module and associated external sensors; (2) Convolutional neural network model training: The convolutional neural network analysis module uses the preprocessed data to build and train the convolutional neural network model; (3) Heat supply trend analysis: The heat supply trend analysis module uses the trained convolutional neural network model, combined with real-time input data and external factors, to determine the user's heat supply demand in the future and provide a basis for the formulation of the heating plan; (4) Charging model selection: Users can select a suitable charging model based on their own needs through the configuration module to meet their personalized needs; (5) Cost calculation: The fee calculation module calculates the user's heating fee based on the predicted heat supply and actual heat supply, combined with the heating company's charging standards and the charging model selected by the user; (6) Cost warning and notification: When the user's on-demand heating costs are close to the preset budget / exceed a certain proportion, the cost warning unit of the personalized interactive module will promptly remind the user through SMS, APP messages, etc.

10. The method for using the heating charging system based on convolutional neural network according to claim 9, characterized in that: The method of use also includes the following steps: (7) Mobile bill generation: After the heat supply trend analysis and charge calculation are completed, the fee output module provides users with detailed heat supply and charge information and generates a mobile heat supply charge bill; (8) Heat data adjustment: The parameter conditioning unit adjusts the data required by the charging system, unifies the data scale, and extracts key features that affect heat supply and charging; (9) Energy dynamic tracking: The energy situation awareness module monitors the energy consumption of the entire heating system in real time; (10) Equipment operation monitoring: The external system association module is linked with the monitoring system of the heating equipment to obtain the equipment's operating parameters and promptly detect the increase in energy consumption caused by equipment failure / abnormal operation; (11) Personalized heating recommendations: The personalized interaction module provides users with recommendations on temperature settings and heating time adjustments based on their heating habits and historical data. The customer service unit helps users strike a balance between comfort and energy saving.

Images