Heat supply management method and system based on deep learning and storage medium

Abstract

The invention provides a heat supply management method and system based on deep learning and a storage medium, and the method comprises the steps: obtaining the working state of a heat exchange station, judging whether the working state is a heat supply state, and if not, generating a shutdown instruction to enable a heat source plant to stop working; if yes, obtaining a historical calorie value obtained by each user side, and storing the historical calorie value in a dynamic database; a historical calorie value group corresponding to each user side is obtained from a dynamic database, and the historical calorie value group comprises a plurality of ordered historical calorie values; processing the historical heat value group by using a preset learning model to obtain a predicted heat value required by each user side at the next moment; and determining a second pipe network heat value required to be provided for each user side by the heat exchange station and a first pipe network heat total value produced by the heat source plant based on all the predicted heat values. By means of the method, the situation that waste and lag exist in centralized heating can be relieved.

Description

Technical Field

[0001] The present application relates to the field of heating technology, and in particular to a heating management method, system and storage medium based on deep learning. Background Art

[0002] Central heating is a way of organizing energy regional production by building centralized heat sources in industrial production areas or urban residential areas to provide production and living heat to enterprises, units and residents in and around the area. Central heating can be mechanized to solve problems such as smoke pollution and ash and fuel storage, and has obvious environmental and social benefits, making central heating an important part of urban public utilities.

[0003] At present, the management of centralized heating needs to rely on the first pipeline network and the second pipeline network. The heat source plant will transmit all the heat generated to the heat exchange station through the first pipeline network, and the heat exchange station will distribute the received heat to the corresponding users through the second pipeline network. Among them, first, the heat exchange station will store enough heat, so that when the user has a heat demand, he will send the heat demand value to the heat exchange station, and then the heat exchange station will control the corresponding valve opening degree according to the demand value to provide the corresponding user with the heat corresponding to the demand value, thereby completing the work of centralized heating for the user.

[0004] Although the existing method can realize centralized heating for users, there is heat waste in the process of centralized heating because the heat source plant still supplies enough heat to the heat exchange station in advance; on the other hand, it is necessary to control the corresponding valve to provide heat to the user after receiving the demand value for heat, which will have a certain lag. Therefore, the existing centralized heating work completed by managing the pipeline network will lead to waste and lag in centralized heating. Summary of the invention

[0005] In order to alleviate the waste and lag in centralized heating, the embodiments of the present application provide a control method, system and storage medium for centralized heating.

[0006] In a first aspect, this embodiment provides a method for controlling centralized heating, the method comprising: Obtaining the working state of the heat exchange station, determining whether the working state is a heating state, and if not, generating a shutdown instruction to stop the heat source plant; If so, obtain the historical calorific value obtained by each user terminal, and store the historical calorific value in a dynamic database; Acquire a historical calorific value group corresponding to each user terminal from the dynamic database, wherein the historical calorific value group includes a plurality of ordered historical calorific values; Processing the historical calorific value group using a preset learning model to obtain a predicted calorific value required by each user terminal at the next moment; Based on all predicted heat values, determine the second pipeline network heat value that the heat exchange station needs to provide for each user end, as well as the total heat value of the first pipeline network produced by the heat source plant.

[0007] In some embodiments, the dynamic database is provided with a historical calorific value group corresponding to each user terminal, and storing the historical calorific value in the dynamic database includes: Obtain the target subscript for storing historical calorific values ​​in each historical calorific value group, and the terminal subscript for storing the last historical calorific value, and respectively determine whether the target subscript corresponding to the same historical calorific value group is equal to the terminal subscript. If not, store the historical calorific value corresponding to the historical calorific value group at the subscript next to the target subscript in the historical calorific value group; If they are equal, clear the historical calorific value stored at the minimum index in the historical calorific value group, store other historical calorific values ​​in the historical calorific value group at the previous index, and store the historical calorific value at the target index in the historical calorific value group.

[0008] In some embodiments, the processing of the historical calorific value group using a preset learning model includes: Obtain the storage quantity of the historical calorific value stored in each historical calorific value group, and determine whether the difference between the storage quantity corresponding to the same user terminal and the terminal index is one. If it is one, send the historical calorific value group to the input end of the learning model to complete the work of processing the historical calorific value group; If it is not one, obtaining the quantity difference between the terminal index and the stored quantity, uniformly moving all historical calorific values ​​in the historical calorific value group backward by the quantity difference minus one position, and filling the blank positions in the historical calorific value group with zeros to obtain a new historical calorific value group; The new historical calorific value group is sent to the input end of the learning model to complete the work of processing the historical calorific value group.

[0009] In some embodiments, determining the second pipe network heat value that needs to be provided by the heat exchange station to each user terminal based on all predicted heat values ​​includes: Obtain the distance value between each user end and the heat exchange station, the heat loss rate per unit distance in the second pipe network, and the valve opening ratio of the valves set between the heat exchange station and each user end, and divide the predicted heat value by the distance value, heat loss rate, and valve opening ratio corresponding to the predicted heat value, respectively, to obtain the second pipe network heat value that the heat exchange station needs to provide for the user end corresponding to the predicted heat value.

[0010] In some embodiments, after determining the total heat value of the first pipe network that needs to be generated by the heat source plant, the method further includes: Determine whether feedback information based on the predicted calorific value is received from the user end, wherein the feedback information includes the feedback calorific value, and if received, generate an adjustment instruction to adjust the weight value in the learning model based on the feedback calorific value as a subsequent learning model; If not received, continue to obtain the working status of the heat exchange station.

[0011] In some embodiments, adjusting the weight value in the learning model based on the feedback heat value includes: Determine the output layer weight error value corresponding to the last layer network in the learning model based on the feedback heat value and the historical heat value group; The output layer weight value in the last layer network of the learning model is obtained, and the output layer weight value is added to the output layer weight error value to adjust the weight value in the learning model.

[0012] In some embodiments, the method further comprises: If the working state is not the heating state, a closing instruction is also generated to close the valves in the first pipe network and the second pipe network.

[0013] In a second aspect, this embodiment provides a centralized heating control system, the system comprising: a status diagnosis module, a management module, a database module and a processing module; wherein: The state diagnosis module is used to obtain the working state of the heat exchange station and determine whether the working state is a heating state; The management module is used to generate a shutdown instruction to stop the heat source plant if the working state is not a heating state; The database module is used to obtain the historical heat value obtained by each user terminal if the working state is the heating state, and store the historical heat value in the dynamic database; The processing module is used to obtain a historical calorific value group corresponding to each user terminal from the dynamic database, wherein the historical calorific value group includes a plurality of ordered historical calorific values; and use a preset learning model to process the historical calorific value group to obtain a predicted calorific value required by each user terminal at the next moment; The management module is also used to determine the second pipe network heat value that the heat exchange station needs to provide for each user end and the total heat value of the first pipe network produced by the heat source plant based on all predicted heat values.

[0014] In some of the embodiments, the management module is further configured to generate a closing instruction to close the valves in the first pipe network and the second pipe network if the working state is not a heating state.

[0015] In a third aspect, this embodiment provides a computer-readable storage medium on which is stored a computer program that can be run on a processor, and when the computer program is executed by the processor, a method for controlling centralized heating as described in the first aspect is implemented.

[0016] By adopting the above method, the present application first obtains the working status of the heat exchange station, determines whether the working status is a heating status, and if not, generates a shutdown instruction to stop the heat source plant. Cut off the generation of heat from the source to minimize the waste of energy and heat. If the working status is a heating status, obtain the historical heat value obtained by each user end, and store the historical heat value in a dynamic database. In this way, all the historical heat values ​​obtained by the user end for this heating work are sorted out, so as to better predict the user end's demand for heat in the future.

[0017] Then, a historical calorific value group corresponding to each user terminal is obtained from the dynamic database, wherein the historical calorific value group includes a number of ordered historical calorific values, thereby providing reference information for subsequent prediction of the predicted calorific value required by the user terminal at the next moment.

[0018] Then, the preset learning model is used to process the historical heat value group to obtain the predicted heat value required by each user at the next moment. Knowing the user's demand for heat value in advance can provide a reference for the subsequent determination of the heat required by the heat source, thereby reducing the heat waste caused by the heat source over-preparing heat to ensure heating; on the other hand, based on the predicted heat value, the heat supply to the user can be guaranteed in advance, alleviating the delay in centralized heating.

[0019] Finally, based on all the predicted heat values, the heat value of the second pipe network that the heat exchange station needs to provide for each user end and the total heat value of the first pipe network produced by the heat source plant are determined. Not only can the required heat be provided to each user end in advance, alleviating the delay in centralized heating; but the heat value that needs to be generated by the heat source plant can also be calculated in advance, alleviating the waste of centralized heating. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic diagram of the central heating operation provided in the embodiment of the present application.

[0021] Figure 2 This is a block diagram of a heat management method based on deep learning provided in an embodiment of the present application.

[0022] Figure 3This is a flowchart of storing historical calorific values ​​in a dynamic database provided by an embodiment of the present application.

[0023] Figure 4 This is a flow chart of processing a historical calorific value group using a preset learning model provided in an embodiment of the present application.

[0024] Figure 5 This is a framework diagram of a heating management system based on deep learning provided in an embodiment of the present application. DETAILED DESCRIPTION

[0025] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. However, it should be understood by those of ordinary skill in the art that the present application can be implemented without these details. For those of ordinary skill in the art, it is obvious that various changes can be made to the embodiments disclosed in the present application, and without departing from the principles and scope of the present application, the general principles defined in the present application can be applied to other embodiments and application scenarios. Therefore, the present application is not limited to the embodiments shown, but conforms to the broadest scope consistent with the scope claimed for protection in the present application.

[0026] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0027] Figure 1 This is a schematic diagram of the central heating operation provided by the embodiment of the present application. Figure 1 As shown in the figure, the central heating work requires the use of heat source plants, heat exchange stations, user terminals, primary pipe networks and secondary pipe networks. Among them, the first pipe network refers to the pipe network between the heat source plant and the heat exchange station, which is mainly responsible for transporting heat from the heat source plant (such as a thermal power plant, a regional boiler room or a centralized boiler room) to the heat exchange station. The heat exchange station is responsible for converting the heat transported from the primary pipe network into hot water at a lower temperature through a heat exchanger. The secondary pipe network refers to the pipeline from the heat exchange station to each user terminal, which is responsible for transporting the hot water at a lower temperature converted by the heat exchange station to the user terminal and providing heat for the user terminal. The first pipe network is also provided with a first valve, and the opening and closing of the first valve can be used to manage whether the heat source plant can transfer heat to the heat exchange station. The second pipe network is also provided with a second valve, and the opening and closing of the second valve can be used to manage whether the heat exchange station can transfer heat to the user terminal.

[0028] Figure 2 is a block diagram of a heat supply management method based on deep learning provided in an embodiment of the present application. Figure 2 As shown, the heating management method based on deep learning includes the following steps: Step S100, obtaining the working state of the heat exchange station, determining whether the working state is a heating state, and if not, generating a shutdown instruction to stop the heat source plant.

[0029] In the centralized heating work, a management terminal is also provided, which can manage the working status of the heat source plant and the heat exchange station, and whether the first valve and each second valve are opened and the degree of opening. It is provided with a program corresponding to the deep learning-based heating management method of this application. This embodiment describes the deep learning-based heating management method from the perspective of the management terminal.

[0030] There are two corresponding working states of the heat exchange station, one is the heating state, and the other is the non-heating state, that is, the resting state. Among them, the heat exchange station is provided with a status code used to characterize its working state, and the status codes corresponding to the heating state and the resting state are different. The management end can obtain the status code of the heat exchange station through the communication channel set up with the heat exchange station to obtain the working state of the heat exchange station. Then the obtained status code is compared with the feature code corresponding to the heating state. If the two feature codes are the same, it indicates that the working state is the heating state. If the two feature codes are different, it indicates that the working state is not the heating state.

[0031] When the working state of the heat exchange station is not the heating state, it indicates that there is no need to provide heat to the user end. At this time, in order to reduce the waste of heat and energy, the management end will generate a shutdown command and send this shutdown command to the heat source plant to stop working, cut off the generation of heat from the source, and minimize the waste of energy and heat.

[0032] Step S200: If yes, obtain the historical calorific value obtained by each user terminal, and store the historical calorific value in a dynamic database.

[0033] The above historical heat value refers to the actual demand value of heat at the user end in the most recent time. When the working state of the heat exchange station is the heating state, it indicates that the user end has a heat demand and needs to provide heat. Since the demand for heat at the user end will not change suddenly, it is based on the original heat. The future demand for heat is affected by the existing heat. Therefore, in order to facilitate the subsequent prediction of the user end's demand for heat, it is necessary to sort out all the historical heat values ​​that the user end has obtained for this heating work.

[0034] The management end can obtain the historical calorific value obtained by each user end by sending an acquisition request to each user end and receiving the corresponding historical calorific value provided by each user end according to the acquisition request. A timer can also be set at each user end so that after each timer is reached, the user end actively sends the historical calorific value to the management end, so that the management end can obtain the historical calorific value obtained by each user end. No further limitation is made here on obtaining the historical calorific value obtained by each user end.

[0035] After the historical calorific value is obtained, in order to facilitate the subsequent prediction of the user's demand for heat, the historical calorific value can be stored in a dynamic database. The dynamic database is provided with a historical calorific value group corresponding to each user. Figure 3 This is a flowchart of storing historical calorific values ​​in a dynamic database provided by an embodiment of the present application. Figure 3 As shown, storing the historical calorific value in the dynamic database includes the following steps: Step S201, obtain the target subscript that stores historical calorific values ​​in each historical calorific value group, and the terminal subscript that can store the last historical calorific value, and determine whether the target subscript corresponding to the same historical calorific value group is equal to the terminal subscript. If not, store the historical calorific value corresponding to the historical calorific value group at the subscript next to the target subscript in the historical calorific value group.

[0036] Step S202, if they are equal, clear the historical calorific value stored at the minimum index in the historical calorific value group, store other historical calorific values ​​in the historical calorific value group at the previous index, and store the historical calorific value at the target index in the historical calorific value group.

[0037] The historical calorific value group corresponding to each user end in the dynamic database is a blank array before the historical calorific value is stored. The historical calorific value group corresponding to each user end has the same number of subscripts, that is, the number of historical calorific values ​​that can be stored is the same. The subscripts of each historical calorific value group are 0, 1, 2, ..., n. When storing historical calorific values, each historical calorific value group stores them in order from small to large subscripts. For example, if there is no historical calorific value stored at the position with subscript 0 in the historical calorific value group, then the historical calorific value to be stored in the historical calorific value group is stored at the position with subscript 0. If there is a historical calorific value stored at the position with subscript 0 in the historical calorific value group, then continue to see if there is a historical calorific value stored at the position with subscript 1. If not, the historical calorific value to be stored in the historical calorific value group is stored at the position with subscript 1; if yes, continue to see if there is a historical calorific value stored at the position with subscript 2, and so on, so as to achieve storage in order from small to large subscripts.

[0038] The above-mentioned target index represents the largest index in the historical calorific value group. The terminal index represents the index of the historical calorific value that is most recently stored in the historical calorific value group. Among them, the management end can obtain the target index by checking the maximum number of historical calorific values ​​that can be stored in the historical calorific value group. And the terminal index is obtained by checking the number of historical calorific values ​​currently stored in the historical calorific value group. The terminal index is less than or equal to the target index.

[0039] After obtaining the target subscript and terminal subscript of each historical calorific value group, it is possible to determine whether the target subscript corresponding to the same historical calorific value group is equal to the terminal subscript by subtraction or division. For example, the target subscript is subtracted from the terminal subscript to obtain the subscript difference. If the subscript difference is equal to zero, then the target subscript is equal to the terminal subscript; if the subscript difference is not equal to zero, then the target subscript is not equal to the terminal subscript. Alternatively, the terminal subscript is placed at the target subscript to obtain the subscript ratio. If the subscript ratio is equal to one, then the target subscript is equal to the terminal subscript; if the subscript ratio is not equal to one, then the target subscript is not equal to the terminal subscript.

[0040] When the target index is not equal to the terminal index, it indicates that not all indexes in the historical calorific value group store historical calorific values. Then, after receiving the historical calorific value that needs to be stored in this historical calorific value group, the management end will directly store this historical calorific value at the index located next to the target index, that is, store the historical calorific value in the corresponding historical calorific value group in order. For example, the target index of a historical calorific value group is 5, the terminal index is 10, and the historical calorific value group is 4,5,4.5,5,4. If this historical calorific value group needs to store another historical calorific value with a value of 4.5, then this historical calorific value group is 4,5,4.5,5,4,4.5.

[0041] When the target index is equal to the terminal index, it indicates that all indexes in the historical calorific value group store historical calorific values. Then, after receiving the historical calorific value that needs to be stored in this historical calorific value group, the management end will give priority to deleting the earliest stored historical calorific value in this historical calorific value group to leave storage space for the historical calorific value just received. In order to enable all historical calorific values ​​to be stored in this historical calorific value group in an orderly manner according to the time sequence of receipt, after a position is vacated in this historical calorific value group, the current terminal index historical calorific value group can be moved forward one position in turn, so that the position corresponding to the terminal index is blank, and the historical calorific value just received is stored in the position corresponding to the terminal index, that is, the historical calorific value is stored in the dynamic database. For example, the target index of a historical calorific value group is 5, the terminal index is 10, and this historical calorific value group is 4,5,4.5,5,4,3.5,4,4.5,5,4.5. If this historical calorific value group needs to store another historical calorific value with a value of 4, then this historical calorific value group is 5,4.5,5,4,3.5,4,4.5,5,4.5,4. In this way, by setting the historical calorific value group corresponding to each user terminal in the dynamic database, on the one hand, the historical calorific value corresponding to each user terminal in the database can be stored separately, which increases the convenience of subsequent calls. On the other hand, the array method is used to effectively store the historical calorific value corresponding to each user terminal and delete the historical calorific value stored in the early stage, reducing the maintenance cost and workload of maintaining the database due to excessive storage of a large number of historical calorific value groups, and the reference value of the historical calorific value group that is old for the subsequent prediction of calorific values ​​is relatively low.

[0042] Step S300: obtaining a historical calorific value group corresponding to each user terminal from a dynamic database, wherein the historical calorific value group includes a plurality of ordered historical calorific values.

[0043] After the historical calorific value is stored in the dynamic database, the historical calorific value group corresponding to each user terminal can be obtained by viewing all the historical calorific value groups set in the dynamic database. Each historical calorific value group includes several ordered historical calorific values. This provides reference information for the subsequent prediction of the predicted calorific value required by the user terminal at the next moment.

[0044] Step S400, using a preset learning model to process the historical calorific value group to obtain the predicted calorific value required by each user terminal at the next moment.

[0045] After obtaining the historical calorie value group corresponding to each user end, each historical calorie value group is processed separately by using the preset learning model to predict the predicted calorie value required by the user end corresponding to each historical calorie value group at the next moment, so as to know the user end's demand for calorie value in advance. On the one hand, it provides a reference for the subsequent determination of the amount of heat that the heating source needs to prepare, so as to reduce the heat waste caused by excessive preparation of heat by the heating source to ensure heating; on the other hand, it can also supply heat to the user end in advance based on the obtained predicted calorie value, so as to alleviate the delay in centralized heating. Figure 4 1 is a flow chart of processing a historical calorific value group using a preset learning model provided in an embodiment of the present application. Figure 4 As shown, using a preset learning model to process a historical calorific value group includes the following steps: Step S401, obtain the storage quantity of historical calorific values ​​stored in each historical calorific value group, and determine whether the difference between the storage quantity corresponding to the same user terminal and the terminal index is one. If it is one, send the historical calorific value group to the input end of the learning model to complete the work of processing the historical calorific value group.

[0046] Step S402, if it is not one, obtain the quantity difference between the terminal index and the stored quantity, uniformly move all historical calorific values ​​in the historical calorific value group backward by the quantity difference minus one position, and fill the blank positions in the historical calorific value group with zeros to obtain a new historical calorific value group.

[0047] Step S403, sending the new historical calorific value group to the input end of the learning model to complete the work of processing the historical calorific value group.

[0048] The above-mentioned preset learning model refers to a trained deep learning model, which can be specifically a BP neural network model. Since the learning model is trained, the number of information at the input end and the number of information at the output end have also been fixed. If the number of information given to the input end of the learning model is different from the originally set number, it may cause the learning model to fail to work well, wherein the number of information at the input end set by the learning model is the maximum number of historical calorific values ​​that can be stored in the historical calorific value group, that is, the number of terminals plus one. Therefore, after obtaining the historical calorific value group corresponding to each user terminal from the dynamic database, it is also necessary to obtain the storage number of historical calorific values ​​stored in the historical calorific value group by checking the target subscript of each historical calorific value group, that is, the storage number of historical calorific values ​​stored in the historical calorific value group is the target subscript plus one. By comparing the difference between the storage number and the terminal subscript, it is determined whether the learning model can directly use the historical calorific value group. The difference is obtained by subtracting the terminal subscript from the storage number corresponding to the same user terminal, and it is determined whether the difference is one. If it is one, it indicates that the learning model can directly use the historical calorific value group corresponding to this user terminal. If it is not one, it indicates that the learning model cannot directly use the historical calorific value group corresponding to this user terminal.

[0049] When the difference between the storage quantity and the terminal index corresponding to the same user terminal is one, the historical calorific value group corresponding to the user terminal can be directly sent to the input end of the learning model as the input information of the learning model to complete the work of processing the historical calorific value group corresponding to the user terminal. Therefore, under the action of the learning model, the corresponding output information is output, and the output information output by the learning model is used as the predicted calorific value required by the user terminal at the next moment.

[0050] When the difference between the storage quantity and the terminal subscript corresponding to the same user terminal is not one, it indicates that not every subscript in the historical calorific value group corresponding to this user terminal stores a historical calorific value group. The terminal subscript corresponding to this user terminal can be further subtracted from the storage quantity to obtain the difference in quantity between the two, and it can be obtained that there are still several positions in the historical calorific value group where no historical calorific value is stored. At the same time, in order to reflect the correspondence between the historical calorific value group and time, it is also necessary to uniformly move all the historical calorific values ​​stored in the historical calorific value group backward, so that the historical calorific value that was originally at the target subscript among these historical calorific values ​​is moved to the terminal subscript, and the historical calorific value that was originally at the 0 subscript is moved to the terminal subscript minus the target subscript. Then the other blank positions in the historical calorific value group that do not store historical calorific values ​​are filled with zeros, so as to obtain a new historical calorific value group, which contains the terminal subscript plus a historical calorific value, and completes the work of processing the historical calorific value group. After receiving this new historical calorific value group, the learning model can output the predicted calorific value required by the user at the next moment, that is, complete the prediction of the calorific value required by the user at the next moment. In this way, no matter how many historical calorific values ​​are contained in the historical calorific value group obtained from the dynamic database, the historical calorific value group adapted to the learning model can be obtained through processing, thereby improving the accuracy of the learning model in processing the historical calorific value group and outputting the predicted calorific value, and providing relatively accurate predicted calorific values ​​for subsequent mitigation of waste and lag in centralized heating.

[0051] Step S500, based on all predicted heat values, determine the second pipe network heat value that the heat exchange station needs to provide for each user end, and the total heat value of the first pipe network produced by the heat source plant.

[0052] The above-mentioned second pipe network heat value refers to the heat value stored in the heat exchange station after cooling, which can just meet the heat demand of the user end, wherein each user end corresponds to a second pipe network heat value. The total heat value of the first pipe network refers to the heat value stored in the heat source plant that can just meet the heat demand of the heat exchange station. Among them, based on all predicted heat values, determining the second pipe network heat value that the heat exchange station needs to provide for each user end includes: obtaining the distance value between each user end and the heat exchange station, the heat loss rate per unit distance in the second pipe network, and the valve opening ratio value of the valve set between the heat exchange station and each user end, and dividing the preset heat value by the distance value, heat loss rate and valve opening ratio value corresponding to the predicted heat value, respectively, to obtain the second pipe network heat value that the heat exchange station needs to provide for the user end corresponding to the predicted heat exchange value.

[0053] Once the equipment used for centralized heating is installed and fixed, the distance value between the user end and the heat exchange station has been determined, and the heat loss rate per unit distance in the second pipeline has also been fixed. The above distance value and heat loss rate are stored in the management end. Therefore, by viewing the information stored in the management end, the distance value between each user end and the heat exchange station and the heat loss rate per unit distance in the second pipe network can be obtained. In addition, the valve opening ratio value of the valve set between the heat exchange station and each user end can be obtained by reading the opening value of the valve on each second pipe network. In this embodiment, the valve on the second pipe network is preferably at the maximum opening degree. Then, the distance value, heat loss rate and valve opening ratio value corresponding to the same user end are substituted into the formula of the second pipe network heat value * distance value * heat loss rate * valve opening ratio value = predicted heat value, and the second pipe network heat value that the heat exchange station needs to provide for this user end can be obtained. Thus, the distance value, heat loss rate and valve opening ratio value corresponding to each user end are substituted into the above formula, and the second pipe network heat value that the heat exchange station needs to provide for each user end can be obtained. In this way, each valve on the second pipe network can always maintain the same opening degree, reducing the service life of the valve due to frequent rotation.

[0054] In addition, after obtaining the second pipe network heat value that the heat exchange station needs to provide for each user end, all the second pipe network heat values ​​can be added together to obtain the total heat value stored by the heat exchange station to just meet the heat demand of all users. Then, according to the conversion ratio of the heat exchanger to convert heat, the high-temperature heat value received by the heat exchange station is obtained. Finally, the high-temperature heat value is divided by the distance value between the heat exchange station and the heat source plant and the heat loss rate per unit distance in the first pipe network to obtain the total heat value of the first pipe network produced by the heat source plant. In this way, by using the predicted heat value to obtain the first pipe network heat value that the heat exchange station needs to provide for each user end, and the total heat value of the first pipe network produced by the heat source plant in advance, not only can the required heat be provided to each user end in advance, alleviating the delay in centralized heating; but also the heat value that needs to be generated by the heat source plant can be calculated in advance to alleviate the waste of centralized heating.

[0055] In addition, in some other embodiments, after determining the total heat value of the first pipeline network that needs to be produced by the heat source plant, it also includes: judging whether feedback information based on the predicted heat value feedback from the user end is received, wherein the feedback information includes the feedback heat value; if received, generating an adjustment instruction to adjust the weight value in the learning model based on the feedback heat value as a subsequent learning model; if not received, continuing to obtain the working status of the heat exchange station.

[0056] After obtaining the total heat value of the first pipe network and the heat value of the second pipe network, the management end will control the heat source plant and the heat exchange station to perform corresponding work according to these two values, so that each user end can obtain the corresponding predicted heat value. Among them, the predicted heat value is a predicted value after all, and there may be a certain deviation from the heat value actually required by the user end. Therefore, in the case of deviation, the user end will send a feedback message containing the feedback heat value to the management end. In the case of no deviation, the user end will not send feedback information to the management end.

[0057] If the management end does not receive feedback information sent by the user end, it indicates that the predicted heat value obtained in the above step S400 is accurate, and the prediction work performed by the learning model is also accurate. There is no need to adjust the learning model, and the operation of obtaining the working status of the heat exchange station is continued.

[0058] If the management end receives feedback information sent by the user end, it indicates that the predicted calorific value obtained in the above step S400 is inaccurate. At this time, the management end will generate an adjustment instruction to adjust the weight value in the learning model based on the received feedback calorific value as the subsequent learning model. Among them, adjusting the weight value in the learning model based on the feedback calorific value includes: determining the output layer weight error value corresponding to the last layer network in the learning model based on the feedback calorific value and the historical calorific value group; obtaining the output layer weight value in the last layer network in the learning model, and adding the output layer weight error value to the output layer weight value to adjust the weight value in the learning model.

[0059] The above-mentioned feedback heat value specifically represents the difference between the heat value actually required by the user end and the predicted heat value, that is, the prediction error value. After the learning model is trained, the weight value of each layer is known and fixed. According to the neural network calculation logic, the historical heat value group can be used to determine the heat value obtained by the previous layer of the output layer of the learning model, and then the heat value obtained by the previous layer of the output layer of the learning model can be multiplied by the output layer weight error value to obtain the feedback heat value. Therefore, after the heat value obtained by the previous layer of the output layer of the learning model is determined, the output layer weight error value corresponding to the last layer of the network in the learning model can be obtained by multiplying the inverse matrix of the feedback heat value by the heat value obtained by the previous layer of the output layer of the learning model. Thus, the output layer weight error value corresponding to the last layer of the network in the learning model is added to the output layer weight error value, and the accurate output layer weight value at this time can be obtained, and the output layer weight value in the original learning model is replaced to reach the weight value adjustment work in the learning model. In this way, when the next wave of calculation and prediction of heat value is calculated, the adjusted learning model can be used to improve the accuracy of the predicted heat value. And there is no need to train the learning model offline, reducing the impact on the original work rhythm of heating management.

[0060] In addition, if it is determined in the above step S100 that the working state of the heat exchange station is not the heating state, a closing instruction can also be generated to close the valves in the first pipeline network and the second pipeline network to reduce the impact on the user end caused by misoperation of the heat exchange station and the heat source plant.

[0061] Figure 5 is a framework diagram of a heating management system based on deep learning provided in an embodiment of the present application. Figure 5 As shown, the heating management system based on deep learning includes: a status diagnosis module, a management module, a database module and a processing module.

[0062] Among them, the state diagnosis module is used to obtain the working state of the heat exchange station and determine whether the working state is a heating state. The management module is used to generate a shutdown instruction if the working state is not a heating state, so as to stop the heat source plant. The database module is used to obtain the historical heat value obtained by each user terminal if the working state is a heating state, and store the historical heat value in a dynamic database. The processing module is used to obtain the historical heat value group corresponding to each user terminal from the dynamic database, wherein the historical heat value group includes a number of ordered historical heat values; the historical heat value group is processed using a preset learning model to obtain the predicted heat value required by each user terminal at the next moment. The management module is also used to determine the second pipe network heat value that the heat exchange station needs to provide for each user terminal based on all predicted heat values, as well as the total heat value of the first pipe network produced by the heat source plant. The management module is also used to generate a closing instruction if the working state is not a heating state, so as to close the valves in the first pipe network and the second pipe network.

[0063] The other functions performed by the above-mentioned status diagnosis module, management module, database module and processing module as well as the technical details of each function are the same or similar to the corresponding features in the deep learning-based heating management method described above, so they will not be repeated here.

[0064] An embodiment of the present application also provides a computer storage medium, which stores a computer program. When the computer storage medium is run on a computer, the computer can execute the steps of the deep learning-based heating management method described above.

[0065] It should be understood that, although the steps in the flowchart of the accompanying drawings are shown in sequence according to the instructions of the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless explicitly stated in this document, the execution of these steps is not strictly limited in order and can be performed in other orders.

[0066] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A heating management method based on deep learning, characterized in that: The method comprises: Obtaining the working state of the heat exchange station, determining whether the working state is a heating state, and if not, generating a shutdown instruction to stop the heat source plant; If so, obtain the historical calorific value obtained by each user terminal, and store the historical calorific value in a dynamic database; Acquire a historical calorific value group corresponding to each user terminal from the dynamic database, wherein the historical calorific value group includes a plurality of ordered historical calorific values; Processing the historical calorific value group using a preset learning model to obtain a predicted calorific value required by each user terminal at the next moment; Based on all predicted heat values, determine the second pipeline network heat value that the heat exchange station needs to provide for each user end, as well as the total heat value of the first pipeline network produced by the heat source plant.

2. The method according to claim 1, characterized in that The dynamic database is provided with a historical calorific value group corresponding to each user terminal, and the storing of the historical calorific value in the dynamic database includes: Obtain the target subscript for storing historical calorific values ​​in each historical calorific value group, and the terminal subscript for storing the last historical calorific value, and respectively determine whether the target subscript corresponding to the same historical calorific value group is equal to the terminal subscript. If not, store the historical calorific value corresponding to the historical calorific value group at the subscript next to the target subscript in the historical calorific value group; If they are equal, clear the historical calorific value stored at the minimum index in the historical calorific value group, store other historical calorific values ​​in the historical calorific value group at the previous index, and store the historical calorific value at the target index in the historical calorific value group.

3. The method according to claim 2, characterized in that The using a preset learning model to process the historical calorific value group includes: Obtain the storage quantity of the historical calorific value stored in each historical calorific value group, and determine whether the difference between the storage quantity corresponding to the same user terminal and the terminal index is one. If it is one, send the historical calorific value group to the input end of the learning model to complete the work of processing the historical calorific value group; If it is not one, obtaining the quantity difference between the terminal index and the stored quantity, uniformly moving all historical calorific values ​​in the historical calorific value group backward by the quantity difference minus one position, and filling the blank positions in the historical calorific value group with zeros to obtain a new historical calorific value group; The new historical calorific value group is sent to the input end of the learning model to complete the work of processing the historical calorific value group.

4. The method according to claim 1, characterized in that The method of determining the second pipe network heat value that needs to be provided by the heat exchange station to each user terminal based on all predicted heat values ​​includes: Obtain the distance value between each user end and the heat exchange station, the heat loss rate per unit distance in the second pipe network, and the valve opening ratio of the valves set between the heat exchange station and each user end, and divide the predicted heat value by the distance value, heat loss rate, and valve opening ratio corresponding to the predicted heat value, respectively, to obtain the second pipe network heat value that the heat exchange station needs to provide for the user end corresponding to the predicted heat value.

5. The method according to claim 3, characterized in that: After determining the total heat value of the first pipe network that needs to be generated by the heat source plant, the following steps are further included: Determine whether feedback information based on the predicted calorific value is received from the user end, wherein the feedback information includes the feedback calorific value, and if received, generate an adjustment instruction to adjust the weight value in the learning model based on the feedback calorific value as a subsequent learning model; If not received, continue to obtain the working status of the heat exchange station.

6. The method according to claim 5, characterized in that The adjusting the weight value in the learning model based on the feedback heat value comprises: Determine the output layer weight error value corresponding to the last layer network in the learning model based on the feedback heat value and the historical heat value group; The output layer weight value in the last layer network of the learning model is obtained, and the output layer weight value is added to the output layer weight error value to adjust the weight value in the learning model.

7. The method according to claim 1, characterized in that The method further comprises: If the working state is not the heating state, a closing instruction is also generated to close the valves in the first pipe network and the second pipe network.

8. A heating management system based on deep learning, characterized in that: The system includes: a status diagnosis module, a management module, a database module and a processing module; wherein, The state diagnosis module is used to obtain the working state of the heat exchange station and determine whether the working state is a heating state; The management module is used to generate a shutdown instruction to stop the heat source plant if the working state is not a heating state; The database module is used to obtain the historical heat value obtained by each user terminal if the working state is the heating state, and store the historical heat value in the dynamic database; The processing module is used to obtain a historical calorific value group corresponding to each user terminal from the dynamic database, wherein the historical calorific value group includes a plurality of ordered historical calorific values; and use a preset learning model to process the historical calorific value group to obtain a predicted calorific value required by each user terminal at the next moment; The management module is also used to determine the second pipe network heat value that the heat exchange station needs to provide for each user terminal based on all predicted heat values, and the total heat value of the first pipe network produced by the heat source plant.

9. The system according to claim 8, characterized in that The management module is further used to generate a closing instruction to close the valves in the first pipe network and the second pipe network if the working state is not a heating state.

10. A computer-readable storage medium having a computer program that can be run on a processor stored thereon, characterized in that: When the computer program is executed by the processor, the deep learning-based heating management method according to any one of claims 1 to 7 is implemented.

Images