Intelligent heat supply system optimization scheduling platform considering man-machine cooperation

Abstract

The invention discloses an intelligent heat supply system optimization scheduling platform considering man-machine cooperation, and the platform comprises a digital twinning unit which is used for building a heat supply system digital twinning model through employing a mechanism modeling and data identification method; the heat supply AI unit comprises a machine decision module which is used for outputting a heat supply system optimization scheduling strategy for machine decision; the manual analysis unit comprises a model manual verification module and is used for constructing a scheduling service model performance evaluation index by a dispatcher and evaluating and verifying the scheduling service model; the system also comprises an artificial decision module which is used for a dispatcher to output an artificial decision heat supply system optimization scheduling strategy. The man-machine collaborative decision-making unit is used for obtaining a heat supply system optimization scheduling strategy of man-machine collaborative decision-making; the control unit is also used for carrying out dispatcher state perception and intention understanding through a multi-channel man-machine interaction technology and assisting the heat supply AI unit to make a decision; and the mode switching unit is used for carrying out heat supply system optimization scheduling mode switching of manual decision making and machine decision making.

Description

Technical Field

[0001] The present invention belongs to the technical field of heating systems, and specifically relates to an intelligent heating system optimization scheduling platform that takes human-machine collaboration into consideration. Background Art

[0002] As the scale of heating systems continues to expand and upgrade, system optimization and scheduling has become increasingly complex. Currently, artificial intelligence technology is used to learn massive operating data and experience, and to explore the inherent laws and knowledge of various businesses during heating system optimization and scheduling to assist in the operation of the heating system and achieve operational status awareness and intelligent decision-making of the heating system.

[0003] However, in the actual operation of the heating system, the trained prediction model and the established optimization scheduling model will lack the credibility of the model due to additional untrustworthy and uncontrollable factors. In addition, the current artificial intelligence technology is still in the primary stage of perceptual intelligence. Facing the uncertain and complex heating system, the problem of heating system optimization scheduling cannot be solved by artificial intelligence technology alone. The heating system optimization scheduling process is also inseparable from the dominance of human experience and knowledge. Therefore, human-machine collaboration for heating system optimization scheduling has become a new working mode, using machines to quickly process massive data and complete complex calculations, and using human unique thinking and experience knowledge to handle complex and uncertain problems. However, there are relatively few studies on heating system optimization scheduling based on human-machine collaboration. How to give full play to the advantages of humans and machines, coordinate heating dispatchers and intelligent machines, learn and support each other, improve the human-machine collaboration ability in heating scheduling business, and promote human-machine integration is an urgent problem to be solved. Summary of the invention

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide an intelligent heating system optimization scheduling platform that takes human-machine collaboration into consideration, which can tightly couple the experience knowledge and understanding of the heating business of the manual dispatcher with the machine algorithm, so that the two can adapt to and promote each other to form two-way communication and scheduling control; in addition, it can actively understand the dispatcher's intentions and utilize the dispatcher's manual knowledge information to improve the efficiency and accuracy of machine decision-making, enhance the human-machine collaboration capability in the heating scheduling business, and promote human-machine integration.

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

[0006] The present invention provides an intelligent heating system optimization scheduling platform considering human-machine collaboration, including:

[0007] The digital twin unit is used to establish a digital twin model of the heating system using mechanism modeling and data identification methods, and to monitor, simulate and analyze the actual operating status of the system and to preview the optimized operation decisions through the digital twin model of the heating system;

[0008] The heating AI unit includes a machine decision module, which is used to obtain relevant data from the data processing module according to the day-ahead scheduling, intraday supply and demand balance scheduling, real-time optimization scheduling and emergency scheduling business types, and design a model algorithm to train the scheduling business-related prediction model and establish the optimization scheduling model, and output the machine-decided heating system optimization scheduling strategy;

[0009] The manual analysis unit includes a model manual verification module, which is used by the dispatcher to construct the performance evaluation index of the dispatching business model and evaluate and verify the dispatching business model; it also includes a manual decision-making module, which is used by the dispatcher to obtain relevant data through the heating dispatching business interactive interface, and output the heating system optimization dispatching strategy made by manual decision after manual induction, deductive reasoning and simulation analysis;

[0010] A human-machine collaborative decision-making unit is used to collaborate with the manual analysis unit and the heating AI unit, integrate the manual knowledge and decision output information of the manual analysis unit into the heating AI unit, and finally obtain an optimal scheduling strategy for the heating system based on human-machine collaborative decision-making; it is also used to actively perceive the state and understand the intention of the dispatcher through multi-channel human-machine interaction technology, and assist the heating AI unit in making decisions;

[0011] The mode switching unit is used to switch the optimal scheduling mode of the heating system between manual decision-making and machine decision-making. When the heating system scheduling business is decided by the machine, the dispatcher acts as a supervisor to monitor the system operation status and adjust the system heating decision. When the decision is wrong or the system operation is abnormal, the mode is switched to manual intervention in the heating decision.

[0012] Furthermore, the intelligent heating system optimization and scheduling platform also includes: a human-machine task allocation unit, which is used to consider the experience and knowledge of the heating dispatcher and the calculation and storage factors of the AI ​​machine, conduct decision-making ability assessment, and allocate the heating system optimization scheduling tasks to the manual analysis unit and the heating AI unit.

[0013] Furthermore, the heating AI unit also includes a knowledge graph module for identifying entities, extracting relationships and attributes from data using a machine learning algorithm, and establishing a knowledge graph for heating system scheduling;

[0014] The heating AI unit also includes a data processing module for obtaining historical operating data and real-time operating data of heat sources and heating networks, and using a preset machine learning algorithm to extract data features, annotate data labels, and preprocess data.

[0015] Furthermore, the manual analysis unit also includes a manual data assistance module, which is used to use the digital twin model of the heating system, and the dispatcher inputs typical operating condition parameters according to the dispatching experience knowledge and dispatching business characteristics to generate simulated operation data; and is also used for the dispatcher to select key characteristic variables of the heating dispatching business and label data sample labels according to experience knowledge;

[0016] The manual analysis unit also includes a knowledge graph manual assistance module, which is used for dispatchers to assist in entity recognition, relationship extraction and attribute extraction, and to establish a manual knowledge reasoning and knowledge graph quality assessment mechanism based on the dispatcher's experience and knowledge in optimizing the scheduling of the heating system.

[0017] Furthermore, the dispatcher obtains relevant data through the heating dispatch business interactive interface, and outputs the heating system optimization dispatch strategy based on manual decision-making after manual induction, deductive reasoning and simulation analysis, including:

[0018] The dispatcher can obtain relevant operation data and outdoor environment data of the heating system through the heating dispatching business interactive interface, and consult historical data to analyze the operation characteristics of the heating system in different time periods;

[0019] The dispatcher identifies the problems in the system by manually summarizing the data, and uses professional knowledge to deduce the problems, analyze the root causes of the problems and set different dispatch strategies;

[0020] The dispatcher inputs different scheduling strategies into the digital twin model of the heating system, monitors the model output results, and formulates an optimized scheduling strategy for the heating system based on manual decision-making.

[0021] Furthermore, the manual analysis unit and the heating AI unit are coordinated to integrate the manual knowledge and decision output information of the manual analysis unit into the heating AI unit, and finally an optimal scheduling strategy for the heating system with human-machine collaborative decision-making is obtained, including:

[0022] Integrate the evaluation and verification feedback results of the scheduling business model by the manual analysis unit into the machine decision module of the heating AI unit to determine whether the scheduling business model meets the expected results. If not, continuously adjust the model parameters;

[0023] The heating system optimization scheduling strategy manually decided by the manual analysis unit and the heating system optimization scheduling strategy output by the heating AI unit with machine decisions are simulated through the digital twin model respectively to judge the heating indicators after the execution of the strategy and select the optimal operation strategy; or the operation strategies of manual decision and machine decision are integrated and coordinated to output the optimal operation strategy.

[0024] Furthermore, the operation strategies of both manual decision-making and machine decision-making are integrated and coordinated to output the optimal operation strategy, including:

[0025] The intelligent agent and reinforcement learning algorithm technology are used to establish the optimal scheduling model of the heating system, and the optimal scheduling strategy of the heating system with output machine decision is solved;

[0026] The action corresponding to the optimal scheduling strategy of the heating system decided by manual decision is defined as a r,t , define the action corresponding to the machine-determined heating system optimization scheduling strategy as a j,t , and a r,t 、a j,t Fusion generates human-machine collaborative decision-making actions for heating systems t ; The actions include optimizing the dispatching of the heating system to determine the equipment output of each unit at each stage, the start and stop of the units, and whether the heating network is disconnected or disconnected;

[0027] The human-machine collaborative decision-making action of the heating system a t Input into the heating system agent environment, obtain agent feedback information after executing actions, and generate experience tuples (s t ,a t ,r t ,s t+1 ) is stored in the experience buffer; s t is the state of the heating system agent at the decision time t, including the operating parameters of the heating system optimization scheduling related equipment, outdoor meteorological parameters, heat load and the state and intention of the dispatcher; r t is the reward of the heating system agent at decision time t, which is related to the economic goal, supply and demand balance goal and environmental protection goal set during the optimization scheduling;

[0028] Pre-training agent: extract a batch of experience from the experience buffer and calculate it to update the parameters of the reinforcement learning network, repeating the cycle until the algorithm converges;

[0029] Learning from the environment: The heating system agent interacts with the environment based on experience and knowledge, continuously outputs actions through the strategy network, iteratively updates, obtains the optimal reward in the learning process, and uses the action corresponding to the optimal reward as the optimal operation strategy.

[0030] Furthermore, the reinforcement learning algorithm is an improved DDPG algorithm: an Actor network and a Critic network with an LSTM layer, and the input layer of the Actor network is the state s t The hidden layer is the LSTM network and the fully connected layer network, and the output layer is the human-machine collaborative decision action a of the heating system t ; The input layer of the Critic network is state s tThe human-machine collaborative decision-making action of the heating system output by the Actor network a t , the hidden layer is the LSTM network and the fully connected layer network, and the output layer is the action value function Q(s,a).

[0031] Furthermore, the multi-channel human-computer interaction technology is used to actively perceive the state and understand the intention of the dispatcher, and assist the heating AI unit in making decisions, including:

[0032] Through the voice recognition channel and the touch recognition channel, during the day-ahead dispatching stage, the system can perceive and understand the dispatcher's intention to start and stop the backup heat source and the peak-shaving heat source; during the intraday supply and demand balance dispatching stage, when the supply and demand of the heating system is unbalanced, the system can perceive and understand the dispatcher's intention to judge whether to start the real-time optimization dispatching; during the real-time optimization dispatching stage, when the actual heating capacity of the heat source is insufficient, the system can perceive and understand the dispatcher's intention to judge whether to jump to the emergency dispatching; during the emergency dispatching stage, the system can perceive and understand the dispatcher's intention to de-list and cut off the heating network.

[0033] Furthermore, the smart heating system optimization and scheduling platform also includes:

[0034] The decision confidence evaluation module is used to input the heating system optimization scheduling strategy of machine decision and manual collaborative decision into the digital twin model of the heating system to obtain the heating effect after machine decision and manual collaborative decision, including operating cost, supply and demand balance deviation and carbon dioxide emissions. When the heating effect of manual collaborative decision is better than the heating effect of machine decision, the confidence of manual decision is improved and the confidence of machine decision is reduced when optimizing scheduling at the next moment; when the heating effect of manual collaborative decision is worse than the heating effect of machine decision, the confidence of manual decision is reduced and the confidence of machine decision is improved when optimizing scheduling at the next moment; wherein, the heating effect is comprehensively calculated according to the weights of operating cost, supply and demand balance deviation and carbon dioxide emissions.

[0035] The beneficial effects of the present invention are:

[0036] The present invention collaborates the manual analysis unit and the heating AI unit through the human-machine collaborative decision-making unit, integrates the heating system optimization scheduling strategy of manual decision-making and the heating system optimization scheduling strategy of machine decision-making, and obtains the heating system optimization scheduling strategy of human-machine collaborative decision-making. It can tightly couple the experience knowledge and understanding of the heating business of the manual dispatcher with the machine algorithm, and the two can adapt to and promote each other to form two-way communication and scheduling control; in addition, it can actively understand the dispatcher's intentions, and use the dispatcher's manual knowledge information to improve the efficiency and accuracy of machine decision-making, enhance the human-machine collaboration ability in the heating scheduling business, and promote human-machine integration.

[0037] Other features and advantages will be described in the following description, and partly become apparent from the description, or understood by practicing the invention. The purpose and other advantages of the invention are realized and obtained by the structures particularly pointed out in the description and the drawings.

[0038] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0040] Figure 1 The present invention provides an intelligent heating system optimization scheduling platform that takes human-machine collaboration into consideration. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are 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.

[0042] Example 1

[0043] like Figure 1 As shown, this embodiment 1 provides an intelligent heating system optimization scheduling platform considering human-machine collaboration, including:

[0044] The digital twin unit is used to establish a digital twin model of the heating system using mechanism modeling and data identification methods, and to monitor, simulate and analyze the actual operating status of the system and to preview the optimized operation decisions through the digital twin model of the heating system;

[0045] The heating AI unit includes a machine decision module, which is used to obtain relevant data from the data processing module according to the day-ahead scheduling, intraday supply and demand balance scheduling, real-time optimization scheduling and emergency scheduling business types, and design a model algorithm to train the scheduling business-related prediction model and establish the optimization scheduling model, and output the machine-decided heating system optimization scheduling strategy;

[0046] The manual analysis unit includes a model manual verification module, which is used by the dispatcher to construct the performance evaluation index of the dispatching business model and evaluate and verify the dispatching business model; it also includes a manual decision-making module, which is used by the dispatcher to obtain relevant data through the heating dispatching business interactive interface, and output the heating system optimization dispatching strategy made by manual decision after manual induction, deductive reasoning and simulation analysis;

[0047] A human-machine collaborative decision-making unit is used to collaborate with the manual analysis unit and the heating AI unit, integrate the manual knowledge and decision output information of the manual analysis unit into the heating AI unit, and finally obtain an optimal scheduling strategy for the heating system based on human-machine collaborative decision-making; it is also used to actively perceive the state and understand the intention of the dispatcher through multi-channel human-machine interaction technology, and assist the heating AI unit in making decisions;

[0048] The mode switching unit is used to switch the optimal scheduling mode of the heating system between manual decision-making and machine decision-making. When the heating system scheduling business is decided by the machine, the dispatcher acts as a supervisor to monitor the system operation status and adjust the system heating decision. When the decision is wrong or the system operation is abnormal, the mode is switched to manual intervention in the heating decision.

[0049] In this embodiment, the intelligent heating system optimization and scheduling platform also includes: a human-machine task allocation unit, which is used to consider the experience and knowledge of the heating dispatcher and the calculation and storage factors of the AI ​​machine, conduct decision-making ability assessment, and allocate the heating system optimization scheduling tasks to the manual analysis unit and the heating AI unit.

[0050] In this embodiment, the heating AI unit also includes a knowledge graph module for identifying entities, extracting relationships and attributes from data using a machine learning algorithm, and establishing a knowledge graph for heating system scheduling;

[0051] The heating AI unit also includes a data processing module for obtaining historical operating data and real-time operating data of heat sources and heating networks, and using a preset machine learning algorithm to extract data features, annotate data labels, and preprocess data.

[0052] In actual applications, the optimal dispatching of heating systems involves services such as heat load forecasting, fault diagnosis, operation status monitoring and day-ahead dispatching, intraday supply and demand balance dispatching, real-time optimization dispatching and emergency dispatching. Therefore, it is necessary to establish knowledge graphs related to each business. In the heating AI unit, computers mainly use machine learning algorithms to conduct intelligent situational awareness and decision analysis on massive amounts of heating system operation-related data. In the manual analysis unit, the dispatcher's experience and knowledge and understanding of the heating business are mainly utilized. Through the human-computer interaction interface, the dispatcher can assist in decision support and is superior to the machine decision-making ability in some businesses. By closely coupling the dispatcher's cognitive ability, reasoning ability and business experience with machine decision-making, the two adapt to each other and promote each other, forming a two-way information exchange and control.

[0053] In this embodiment, the manual analysis unit also includes a manual data assistance module, which is used to use the digital twin model of the heating system, and the dispatcher inputs typical operating condition parameters according to the dispatch experience knowledge and dispatch business characteristics to generate simulated operation data; it is also used for the dispatcher to select key characteristic variables of the heating dispatch business according to experience knowledge and to label data sample labels;

[0054] The manual analysis unit also includes a knowledge graph manual assistance module, which is used for dispatchers to assist in entity recognition, relationship extraction and attribute extraction, and to establish a manual knowledge reasoning and knowledge graph quality assessment mechanism based on the dispatcher's experience and knowledge in optimizing the scheduling of the heating system.

[0055] In actual applications, the manual collaborative decision-making unit also includes: collaboratively integrating the manual data assistance module in the manual analysis unit with the data processing module in the heating AI unit to ensure the diversity, balance and availability of data samples; and, collaboratively integrating the knowledge graph manual assistance module in the manual analysis unit with the knowledge graph module in the heating AI unit to improve the accuracy and completeness of the knowledge graph.

[0056] In this embodiment, the dispatcher obtains relevant data through the heating dispatch business interactive interface, and outputs the heating system optimization dispatch strategy based on manual decision-making after manual induction, deductive reasoning and simulation analysis, including:

[0057] The dispatcher can obtain relevant operation data and outdoor environment data of the heating system through the heating dispatching business interactive interface, and consult historical data to analyze the operation characteristics of the heating system in different time periods;

[0058] The dispatcher identifies the problems in the system by manually summarizing the data, and uses professional knowledge to deduce the problems, analyze the root causes of the problems and set different dispatch strategies;

[0059] The dispatcher inputs different scheduling strategies into the digital twin model of the heating system, monitors the model output results, and formulates an optimized scheduling strategy for the heating system based on manual decision-making.

[0060] In this embodiment, the manual analysis unit and the heating AI unit are coordinated to integrate the manual knowledge and decision output information of the manual analysis unit into the heating AI unit, and finally obtain the heating system optimization scheduling strategy of human-machine collaborative decision-making, including:

[0061] Integrate the evaluation and verification feedback results of the scheduling business model by the manual analysis unit into the machine decision module of the heating AI unit to determine whether the scheduling business model meets the expected results. If not, continuously adjust the model parameters;

[0062] The heating system optimization scheduling strategy manually decided by the manual analysis unit and the heating system optimization scheduling strategy output by the heating AI unit with machine decisions are simulated through the digital twin model respectively to judge the heating indicators after the execution of the strategy and select the optimal operation strategy; or the operation strategies of manual decision and machine decision are integrated and coordinated to output the optimal operation strategy.

[0063] In this embodiment, the operation strategies of manual decision-making and machine decision-making are integrated and coordinated to output the optimal operation strategy, including:

[0064] The intelligent agent and reinforcement learning algorithm technology are used to establish the optimal scheduling model of the heating system, and the optimal scheduling strategy of the heating system with output machine decision is solved;

[0065] The action corresponding to the optimal scheduling strategy of the heating system decided by manual decision is defined as a r,t , define the action corresponding to the machine-determined heating system optimization scheduling strategy as a j,t , and a r,t 、a j,t Fusion generates human-machine collaborative decision-making actions for heating systems t ; The actions include optimizing the dispatching of the heating system to determine the equipment output of each unit at each stage, the start and stop of the units, and whether the heating network is disconnected or disconnected;

[0066] The human-machine collaborative decision-making action of the heating system a t Input into the intelligent agent environment of the heating system, obtain the intelligent agent feedback information after executing the action, and generate the experience tuple (s t ,a t ,r t ,s t+1 ) is stored in the experience buffer; s tis the state of the heating system agent at the decision time t, including the operating parameters of the heating system optimization scheduling related equipment, outdoor meteorological parameters, heat load and the state and intention of the dispatcher; r t is the reward of the heating system agent at decision time t, which is related to the economic goal, supply and demand balance goal and environmental protection goal set during the optimization scheduling;

[0067] Pre-training agent: extract a batch of experience from the experience buffer and calculate it to update the parameters of the reinforcement learning network, repeating the cycle until the algorithm converges;

[0068] Learning from the environment: The heating system agent interacts with the environment based on experience and knowledge, continuously outputs actions through the strategy network, iteratively updates, obtains the optimal reward in the learning process, and uses the action corresponding to the optimal reward as the optimal operation strategy.

[0069] In this embodiment, the reinforcement learning algorithm is an improved DDPG algorithm: an Actor network and a Critic network with an LSTM layer, and the input layer of the Actor network is the state s t The hidden layer is the LSTM network and the fully connected layer network, and the output layer is the human-machine collaborative decision action a of the heating system t ; The input layer of the Critic network is state s t The human-machine collaborative decision-making action of the heating system output by the Actor network a t , the hidden layer is the LSTM network and the fully connected layer network, and the output layer is the action value function Q(s,a).

[0070] After the introduction of LSTM network, the input and output are no longer single states or actions, but continuous data of a certain length. The training data cannot be randomly sampled from the experience buffer, but a random batch sequence sampling method is used to transfer three consecutive state transfer data (s t-1 ,s t ,s t+1 ) as a group of training units; when the amount of data in the experience buffer meets the training threshold, the algorithm will randomly sample N groups of training units from the experience buffer for a batch of training inputs to the network, which can effectively capture the time correlation in the optimal scheduling process of the heating system and improve the model's understanding and response to the environment. Combining the LSTM network with the deep reinforcement learning algorithm DDPG can significantly improve the training network's fitting ability, active exploration ability, and temporal logic relationship mining ability.

[0071] In this embodiment, the multi-channel human-computer interaction technology is used to actively perceive the state and understand the intention of the dispatcher, and assist the heating AI unit in making decisions, including:

[0072] Through the voice recognition channel and the touch recognition channel, during the day-ahead dispatching stage, the system can perceive and understand the dispatcher's intention to start and stop the backup heat source and the peak-shaving heat source; during the intraday supply and demand balance dispatching stage, when the supply and demand of the heating system is unbalanced, the system can perceive and understand the dispatcher's intention to judge whether to start the real-time optimization dispatching; during the real-time optimization dispatching stage, when the actual heating capacity of the heat source is insufficient, the system can perceive and understand the dispatcher's intention to judge whether to jump to the emergency dispatching; during the emergency dispatching stage, the system can perceive and understand the dispatcher's intention to de-list and cut off the heating network.

[0073] In actual applications, the human-machine collaborative decision-making unit can proactively obtain relevant dispatching intentions or other system operation information from the dispatcher of the heating system, because the dispatcher is the main controller of the entire heating system regulation and control, and will obtain the global information of each heating station, heating network, heat users, and heat sources, as well as foresee possible emergencies in the future. In addition, the heating AI unit can use the experience knowledge and mechanism rules provided by humans to understand the dispatching business characteristics of the entire heating system and improve the effect of optimizing the dispatching decision of the heating system; and the dispatcher can deepen the control of machine decisions based on the inquiries and feedback of the heating AI unit. At the same time, the dispatcher monitors the operation strategy of the machine decision, intervenes when anomalies and risks are found, or replaces it with manual decision, to ensure the safety and stability of the operation of the heating system and achieve better human-machine hybrid two-way interactive decision-making.

[0074] In this embodiment, the smart heating system optimization and scheduling platform also includes:

[0075] The decision confidence evaluation module is used to input the heating system optimization scheduling strategy of machine decision and manual collaborative decision into the digital twin model of the heating system to obtain the heating effect after machine decision and manual collaborative decision, including operating cost, supply and demand balance deviation and carbon dioxide emissions. When the heating effect of manual collaborative decision is better than the heating effect of machine decision, the confidence of manual decision is improved and the confidence of machine decision is reduced when optimizing scheduling at the next moment; when the heating effect of manual collaborative decision is worse than the heating effect of machine decision, the confidence of manual decision is reduced and the confidence of machine decision is improved when optimizing scheduling at the next moment; wherein, the heating effect is comprehensively calculated according to the weights of operating cost, supply and demand balance deviation and carbon dioxide emissions.

[0076] It should be noted that when the heating effect of manual collaborative decision-making is better than that of machine decision-making, the gain factor α>1 is designed to amplify the confidence of manual decision-making, and the attenuation factor β<1 is designed to reduce the confidence of machine decision-making. Suppose the confidence of the current manual decision is C h , the confidence of the machine decision is C m , then the updated confidence is expressed as: C′ h =C h ×α、C′m =C m ×β;

[0077] When the heating effect of manual collaborative decision-making is inferior to that of machine decision-making, the confidence of machine decision-making is increased and the confidence of manual decision-making is reduced. The updated confidence is expressed as: C′ h =C h ×β、C′ m =C m ×α;

[0078] Among them, the values ​​of the gain factor α and the attenuation factor β are dynamically set according to the deviation between the heating effect of manual collaborative decision-making and the heating effect of machine decision-making, and the adjustment of confidence should take into account historical data rather than being based solely on the most recent decision effect.

[0079] In several embodiments provided in the present application, it should be understood that the disclosed system and method can also be implemented in other ways. The system embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and a part of the module, program segment or code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0080] In addition, the functional modules in each embodiment of the present invention can be integrated together to form an independent part, or each module can exist separately, or two or more modules can be integrated to form an independent part. If the function is implemented in the form of a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), disk or optical disk and other media that can store program codes.

[0081] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. An intelligent heating system optimization scheduling platform considering human-machine collaboration, characterized in that: include: The digital twin unit is used to establish a digital twin model of the heating system using mechanism modeling and data identification methods, and to monitor, simulate and analyze the actual operating status of the system and to preview the optimized operation decisions through the digital twin model of the heating system; The heating AI unit includes a machine decision module, which is used to obtain relevant data from the data processing module according to the day-ahead scheduling, intraday supply and demand balance scheduling, real-time optimization scheduling and emergency scheduling business types, and design a model algorithm to train the scheduling business-related prediction model and establish the optimization scheduling model, and output the machine-decided heating system optimization scheduling strategy; The manual analysis unit includes a model manual verification module, which is used by the dispatcher to construct the performance evaluation index of the dispatching business model and evaluate and verify the dispatching business model; it also includes a manual decision-making module, which is used by the dispatcher to obtain relevant data through the heating dispatching business interactive interface, and output the heating system optimization dispatching strategy made by manual decision after manual induction, deductive reasoning and simulation analysis; A human-machine collaborative decision-making unit is used to collaborate with the manual analysis unit and the heating AI unit, integrate the manual knowledge and decision output information of the manual analysis unit into the heating AI unit, and finally obtain an optimal scheduling strategy for the heating system based on human-machine collaborative decision-making; it is also used to actively perceive the state and understand the intention of the dispatcher through multi-channel human-machine interaction technology, and assist the heating AI unit in making decisions; The mode switching unit is used to switch the optimal scheduling mode of the heating system between manual decision-making and machine decision-making. When the heating system scheduling business is decided by the machine, the dispatcher acts as a supervisor to monitor the system operation status and adjust the system heating decision. When the decision is wrong or the system operation is abnormal, the mode is switched to manual intervention in the heating decision.

2. The intelligent heating system optimization and scheduling platform according to claim 1 is characterized in that: The intelligent heating system optimization and scheduling platform also includes: a human-machine task allocation unit, which is used to consider the experience and knowledge of the heating dispatcher and the calculation and storage factors of the AI ​​machine, conduct decision-making ability assessment, and allocate the heating system optimization scheduling tasks to the manual analysis unit and the heating AI unit.

3. The intelligent heating system optimization and scheduling platform according to claim 1 is characterized in that: The heating AI unit also includes a knowledge graph module for identifying entities, extracting relationships and attributes from data using a machine learning algorithm, and establishing a knowledge graph for heating system scheduling; The heating AI unit also includes a data processing module for obtaining historical operating data and real-time operating data of heat sources and heating networks, and using a preset machine learning algorithm to extract data features, annotate data labels, and preprocess data.

4. The intelligent heating system optimization and scheduling platform according to claim 1 is characterized in that: The manual analysis unit also includes a manual data assistance module, which is used to use the digital twin model of the heating system, and the dispatcher inputs typical operating condition parameters according to the dispatch experience knowledge and dispatch business characteristics to generate simulated operation data; it is also used for the dispatcher to select key characteristic variables of the heating dispatch business and mark data sample labels according to experience knowledge; The manual analysis unit also includes a knowledge graph manual assistance module, which is used for dispatchers to assist in entity recognition, relationship extraction and attribute extraction, and to establish a manual knowledge reasoning and knowledge graph quality assessment mechanism based on the dispatcher's experience and knowledge in optimizing the scheduling of the heating system.

5. The intelligent heating system optimization and scheduling platform according to claim 1 is characterized in that: The dispatcher obtains relevant data through the heating dispatch business interactive interface, and outputs the heating system optimization dispatch strategy based on manual decision-making after manual induction, deductive reasoning and simulation analysis, including: The dispatcher can obtain relevant operation data and outdoor environment data of the heating system through the heating dispatching business interactive interface, and consult historical data to analyze the operation characteristics of the heating system in different time periods; The dispatcher identifies the problems in the system by manually summarizing the data, and uses professional knowledge to deduce the problems, analyze the root causes of the problems and set different dispatch strategies; The dispatcher inputs different scheduling strategies into the digital twin model of the heating system, monitors the model output results, and formulates an optimized scheduling strategy for the heating system based on manual decision-making.

6. The intelligent heating system optimization and scheduling platform according to claim 1 is characterized in that: The said coordination of the manual analysis unit and the heating AI unit, integrating the manual knowledge and decision output information of the manual analysis unit into the heating AI unit, and finally obtaining the heating system optimization scheduling strategy of the human-machine collaborative decision-making, includes: Integrate the evaluation and verification feedback results of the scheduling business model by the manual analysis unit into the machine decision module of the heating AI unit to determine whether the scheduling business model meets the expected results. If not, continuously adjust the model parameters; The heating system optimization scheduling strategy manually decided by the manual analysis unit and the heating system optimization scheduling strategy output by the heating AI unit with machine decisions are simulated through the digital twin model respectively to judge the heating indicators after the execution of the strategy and select the optimal operation strategy; or the operation strategies of manual decision and machine decision are integrated and coordinated to output the optimal operation strategy.

7. The intelligent heating system optimization and scheduling platform according to claim 6 is characterized in that: The method of fusing and coordinating the operation strategies of manual decision-making and machine decision-making to output the optimal operation strategy includes: The intelligent agent and reinforcement learning algorithm technology are used to establish the optimal scheduling model of the heating system, and the optimal scheduling strategy of the heating system with output machine decision is solved; The action corresponding to the optimal scheduling strategy of the heating system decided by manual decision is defined as a r,t , define the action corresponding to the machine-determined heating system optimization scheduling strategy as a j,t , and a r,t 、a j,t Fusion generates human-machine collaborative decision-making actions for heating systems t ; The actions include optimizing the dispatching of the heating system to determine the equipment output of each unit at each stage, the start and stop of the units, and whether the heating network is disconnected or disconnected; The human-machine collaborative decision-making action of the heating system a t Input into the heating system agent environment, obtain agent feedback information after executing actions, and generate experience tuples (s t ,a t ,r t ,s t+1 ) is stored in the experience buffer; s t is the state of the heating system agent at the decision time t, including the operating parameters of the heating system optimization scheduling related equipment, outdoor meteorological parameters, heat load and the state and intention of the dispatcher; r t is the reward of the heating system agent at decision time t, which is related to the economic goal, supply and demand balance goal and environmental protection goal set during the optimization scheduling; Pre-training agent: extract a batch of experience from the experience buffer and calculate it to update the parameters of the reinforcement learning network, repeating the cycle until the algorithm converges; Learning from the environment: The heating system agent interacts with the environment based on experience and knowledge, continuously outputs actions through the strategy network, iteratively updates, obtains the optimal reward in the learning process, and uses the action corresponding to the optimal reward as the optimal operation strategy.

8. The intelligent heating system optimization and scheduling platform according to claim 7 is characterized in that: The reinforcement learning algorithm is an improved DDPG algorithm: an Actor network and a Critic network with an LSTM layer. The input layer of the Actor network is the state s t The hidden layer is the LSTM network and the fully connected layer network, and the output layer is the human-machine collaborative decision action a of the heating system t ; The input layer of the Critic network is state s t The human-machine collaborative decision-making action of the heating system output by the Actor network a t , the hidden layer is the LSTM network and the fully connected layer network, and the output layer is the action value function Q(s,a).

9. The intelligent heating system optimization and scheduling platform according to claim 1 is characterized in that: The multi-channel human-computer interaction technology is used to actively perceive the state and understand the intention of the dispatcher, and assist the heating AI unit in making decisions, including: Through the voice recognition channel and the touch recognition channel, during the day-ahead dispatching stage, the system can perceive and understand the dispatcher's intention to start and stop the backup heat source and the peak-shaving heat source; during the intraday supply and demand balance dispatching stage, when the supply and demand of the heating system is unbalanced, the system can perceive and understand the dispatcher's intention to judge whether to start the real-time optimization dispatching; during the real-time optimization dispatching stage, when the actual heating capacity of the heat source is insufficient, the system can perceive and understand the dispatcher's intention to judge whether to jump to the emergency dispatching; during the emergency dispatching stage, the system can perceive and understand the dispatcher's intention to de-list and cut off the heating network.

10. The intelligent heating system optimization and scheduling platform according to claim 1, characterized in that: The smart heating system optimization and scheduling platform also includes: The decision confidence evaluation module is used to input the heating system optimization scheduling strategy of machine decision and manual collaborative decision into the digital twin model of the heating system to obtain the heating effect after machine decision and manual collaborative decision, including operating cost, supply and demand balance deviation and carbon dioxide emissions. When the heating effect of manual collaborative decision is better than the heating effect of machine decision, the confidence of manual decision is improved and the confidence of machine decision is reduced when optimizing scheduling at the next moment; when the heating effect of manual collaborative decision is worse than the heating effect of machine decision, the confidence of manual decision is reduced and the confidence of machine decision is improved when optimizing scheduling at the next moment; wherein, the heating effect is comprehensively calculated according to the weights of operating cost, supply and demand balance deviation and carbon dioxide emissions.

Images