Data processing method and device

Abstract

The invention discloses a data processing method and device. The method comprises the steps of obtaining target time sequence data corresponding to a heat exchange device of a target building object; processing the target time sequence data to obtain target features; the target features at least comprise a first feature, a second feature, a third feature and a fourth feature; the first feature represents time information associated with the heating demand in the target time sequence data; the second feature represents data associated with heat storage of the target building object in the target time sequence data; the third feature represents overall temperature conditions corresponding to key space points with different weight factors in the target building object; the fourth characteristic represents the change trend of the heat supply quantity in a specified time period; the target features are processed based on a target model, and heating parameters are obtained; the heating parameters are used for adjusting operation parameters of the heat exchange device.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a data processing method and device. Background Art

[0002] In building energy management, the energy consumption of HVAC systems always occupies a dominant position, especially in the composition of annual energy consumption, the energy consumption proportion of its heating function is particularly prominent. The realization of the heating function mainly depends on the key component in the system - the heat exchange station. The operating efficiency of the heat exchange station is not only directly related to the heating performance, but also has a profound impact on the energy consumption level of the entire HVAC system. Therefore, how to achieve energy optimization control of the heat exchange station to improve the overall energy efficiency of the HVAC system has become a key issue that needs to be solved in the current field of energy saving and consumption reduction of heating systems. Summary of the invention

[0003] The technical solutions provided by this application are as follows:

[0004] The first aspect of the present application provides a data processing method, comprising:

[0005] Obtain target time series data corresponding to the heat exchange device of the target building object;

[0006] The target time series data is processed to obtain target features; the target features include at least a first feature, a second feature, a third feature and a fourth feature; the first feature represents time information associated with heating demand in the target time series data; the second feature represents data associated with heat storage of the target building object in the target time series data; the third feature represents the overall temperature conditions corresponding to key spatial points with different weight factors in the target building object; the fourth feature represents the change trend of heating heat in a specified time period;

[0007] The target characteristics are processed based on the target model to obtain heating parameters; the heating parameters are used to adjust the operating parameters of the heat exchange device.

[0008] The first feature includes at least one of the following:

[0009] The first time feature represents the global time information associated with the heating demand in the target time series data;

[0010] The second time feature characterizes a key time point in the target time series data associated with the internal temperature change of the target building object.

[0011] The target time series data is processed to obtain target features, including at least one of the following:

[0012] Encode the date in the target time series data to obtain a date encoding feature, and add a category feature to the date encoding feature to obtain a first time feature; the category feature represents the holiday to which the date belongs;

[0013] A key time point is selected from the target time series data, and a one-hot encoding feature is constructed for the key time point as a second time feature.

[0014] The target time series data is processed to obtain target features, including:

[0015] Acquire the temperature value of each key spatial point in each area of ​​at least one area in the target building object from the target time series data;

[0016] Based on the weight factors of the key spatial points, the temperature values ​​of the key spatial points are processed to obtain the temperature value of the area;

[0017] Based on the temperature values ​​of each of the regions, determining a plurality of temperature attribute values; the plurality of temperature attribute values ​​characterizing the temperature distribution of the at least one region;

[0018] The multiple temperature attribute values ​​are processed to obtain a third feature.

[0019] The target time series data is processed to obtain target features, including:

[0020] Based on the target time series data, obtaining the current heat characteristics of the heat supply source of the heat exchange device;

[0021] Based on the current heat characteristics, determining a heat component; the thermal component is used to reflect the change of the current heat data in the time domain;

[0022] Determining, based on the heat component, a predicted heat characteristic of a heat supply source of the heat exchange device in a specified time interval after a current time interval;

[0023] The current heat feature and the predicted heat feature are used as the fourth feature.

[0024] The target features also include:

[0025] The fifth feature is used to characterize the rate of change of heat consumption of the heat supply source of the heat exchange device.

[0026] The target model is trained in the following way:

[0027] Obtain sample time series data and target heating parameters corresponding to the heat exchange device of the target building object;

[0028] Processing the sample time series data to obtain sample features;

[0029] Processing the sample features based on a plurality of first learning modules to obtain first heating parameters determined by each of the first learning modules;

[0030] Based on the first losses corresponding to the first learning modules, adjusting the parameters of the first learning modules; the first losses represent the difference between the first heating parameters and the target heating parameters;

[0031] Fixing the parameters of each of the first learning modules, inputting the first heating parameters determined by each of the first learning modules into the second learning module, and obtaining the second heating parameters determined by the second learning module;

[0032] Based on a second loss, the parameters of the second learning module are adjusted; the second loss represents the difference between the second heating parameter and the target heating parameter.

[0033] The data processing method further includes:

[0034] Based on the target feature and the heating parameter, each of the first learning modules and the second learning modules is evaluated to obtain an index value of each of the first learning modules and the second learning modules; the index value is used to characterize the accuracy of each of the first learning modules and the second learning modules;

[0035] It is determined that the indicator value does not meet the set indicator threshold, and the first learning module and the second learning module are retrained.

[0036] The data processing method further includes at least one of the following:

[0037] Performing a first output constraint on the heating parameter so that the regulation frequency of the heating parameter by the heat exchange device satisfies a set regulation frequency threshold;

[0038] Performing a second output constraint on the heating parameter so that a change process of the heat exchange device from the current heating parameter to the heating parameter is relatively stable;

[0039] Performing a third output constraint on the heating parameter so that the heating parameter obtained by the heat exchange device is within a set upper and lower limit range;

[0040] A fourth output constraint is performed on the heating parameters so that the heat obtained by the target building object meets the overall heat demand and the heating flow and heating temperature of the target monitoring area in the target building object meet the set area demand.

[0041] Another aspect of the present application provides a data processing device, comprising:

[0042] An acquisition module, used to obtain target time series data corresponding to a heat exchange device of a target building object;

[0043] The first processing module is used to process the target time series data to obtain target features; the target features include at least a first feature, a second feature, a third feature and a fourth feature; the first feature represents the time information associated with the heating demand in the target time series data; the second feature represents the data associated with the heat storage of the target building object in the target time series data; the third feature represents the overall temperature conditions corresponding to the key spatial points with different weight factors in the target building object; the fourth feature represents the change trend of the heating quantity in a specified time period;

[0044] The second processing module is used to process the target characteristics based on the target model to obtain heating parameters; the heating parameters are used to adjust the operating parameters of the heat exchange device. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0046] Figure 1 A flowchart of a data processing method provided in Example 1 of the present application;

[0047] Figure 2 A schematic diagram of comparison before and after data cleaning provided for this application;

[0048] Figure 3 A schematic diagram of a scenario of a heating system control page provided for this application;

[0049] Figure 4 Another schematic diagram of a heating system control page provided for this application;

[0050] Figure 5 A correlation heat diagram provided for this application;

[0051] Figure 6 A schematic diagram of the relationship between heat load and time provided for this application;

[0052] Figure 7 A schematic diagram of a curve of a secondary network water supply temperature setting value provided in this application;

[0053] Figure 8 A schematic diagram of the structure of a target model provided for this application;

[0054] Fig. 9 A performance diagram of a target model provided for this application;

[0055] Fig.10 A schematic diagram of the residual of a target model provided in this application;

[0056] Fig.11 A schematic diagram of a model monitoring page provided for this application;

[0057] Fig.12 A visual schematic diagram of the training of a target model provided in this application;

[0058] Fig.13 A schematic diagram of a simulation result of a target model provided in this application;

[0059] Fig.14 A schematic diagram of scheduling optimization of an edge computing server provided for this application;

[0060] Fig.15 A schematic diagram of the structure of a data processing device provided in this application. DETAILED DESCRIPTION

[0061] The following describes the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. The terms used in the implementation method section of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0062] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0063] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, which is only to describe the distinction mode adopted by the objects of the same attributes when describing in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0064] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0065] Reference Figure 1, is a flow chart of a data processing method provided in Example 1 of the present application, such as Figure 1 As shown, the method may include but is not limited to the following steps:

[0066] Step S101: Obtain target time series data corresponding to a heat exchange device of a target building object.

[0067] In this embodiment, the thermal interaction device may include key components such as a primary network, a plate heat exchanger and a secondary network.

[0068] The primary network may include a water supply network and a return network for connecting the heat source and the plate heat exchanger, as well as a circulating pump arranged on the water supply network and a regulating valve arranged on the return network.

[0069] The circulation pump can be used to promote the circulation of water from the heating source in the pipe network to ensure that the heat can be continuously and stably delivered to the plate heat exchanger.

[0070] The secondary network may include a pipe network connecting the plate heat exchanger and a heat user (eg, an air conditioning system in a target building object), and a pipe network connecting the plate heat exchanger and an air conditioning circulation pump.

[0071] The circulation pump in the primary network can be used to promote the circulation of water from the heating source in the pipe network to ensure that heat can be continuously and stably delivered to the plate heat exchanger.

[0072] The air conditioning circulation pump in the secondary network can be used to deliver cold water to the plate heat exchanger. The plate heat exchanger can use the heat from the primary network to heat the cold water and deliver the heated hot water to the heat users.

[0073] The regulating valve can be used to return the return water from the primary network to the heating source for reheating.

[0074] The target time series data can be used to record the changes of various parameters of the heat exchange device over time. The various parameters of the heat exchange device may include but are not limited to: meteorological parameters of the target building object (such as outdoor temperature, humidity, irradiance and wind speed, etc.), state parameters of the heat exchange device (such as the operating status of the circulation pump in the primary network and the air conditioning circulation pump in the secondary network, the opening of the plate heat exchanger, the instantaneous power of the regulating valve and the circulation pump in the primary network) and the thermal parameters of the primary network (such as the temperature difference between the supply water temperature and the return water temperature of the primary network, and the instantaneous flow rate of the fluid in the primary network, the daily heat supply in the primary network and the accumulated heat in the primary network).

[0075] In this embodiment, data of various parameters of the heat exchange device may be continuously collected to obtain time series data of various parameters, and the time series data of various parameters may be used as target time series data.

[0076] In this embodiment, the data of various parameters of the heat exchange device can also be continuously collected to obtain time series data of each parameter, and the time series data of each parameter can be verified and cleaned, and the cleaned time series data can be used as the target time series data.

[0077] In the actual production environment, there may be network fluctuations or equipment failures, which may cause the acquired data to have over-range, sudden zero values ​​or quality problems, so data verification is required. The main purpose of the verification is to identify and eliminate abnormal data in the time series data of each parameter that obviously does not conform to physical laws or system characteristics.

[0078] For abnormal data that still exists after verification, further data cleaning can be performed. In this embodiment, the error box method can be used to identify abnormal data.

[0079] Error box method:

[0080] For the time series data x1, x2, …, x n , select a window size k, calculate the deviation of the data in the window from the mean, and calculate the upper quantile Q3 and standard deviation σ of the deviation. The deviation size is used to determine whether the data is abnormal. Abnormal data can be supplemented by linear interpolation.

[0081] In this embodiment, the abnormal data is supplemented by the following relationship:

[0082]

[0083] in To calculate the deviation of the data in the window from the mean, Q3 and σ represent the upper quantile and standard deviation of the deviation in the window, k i represents the linear interpolation slope, x i represents the i-th data in the time series data, x i-1 Indicates the data before the i-th data in the time series data.

[0084] For example, Figure 2 As shown, compared with the original data of outdoor temperature, the outdoor temperature after cleaning can more clearly reflect the change of outdoor temperature over time. Compared with the original data of outdoor humidity, the outdoor humidity after cleaning can more clearly reflect the change of outdoor humidity over time.

[0085] Step S102: Process the target time series data to obtain target features.

[0086] The target features may include at least a first feature, a second feature, a third feature and a fourth feature.

[0087] The first feature may characterize time information associated with heating demand in the target time series data.

[0088] In this embodiment, the heating demand often changes with the season, weather, and time (such as morning and evening, weekdays and weekends). For example, in the cold winter, the heating demand is usually high; while in the warm spring or autumn, the heating demand is relatively low. In addition, the heating demand will also be different in different time periods of the day (such as morning, noon, and evening) due to changes in outdoor temperature and people's activity habits. Therefore, by analyzing and utilizing time information, the heating demand can be more accurately predicted, thereby optimizing the operation of the heat exchange device.

[0089] The second feature may characterize data associated with heat storage of the target building object in the target time series data.

[0090] In this embodiment, the heat storage of the target building object may be understood as the ability of the target building object to store heat.

[0091] If the target building object has a strong heat storage capacity, the heat exchange device can increase the indoor temperature faster in the early stage of heating because the target building object can store more heat. This helps to reduce the operating time of the heat exchange device, thereby improving its operating efficiency.

[0092] Target buildings with strong heat storage capacity can reduce heat loss during the heating process, because the heat stored inside the target building can be continuously released, thereby reducing the reliance on heat exchange devices. This helps to reduce the overall energy consumption of the heating system.

[0093] Therefore, by analyzing and utilizing data associated with heat storage, we can better understand the heat storage capacity of the target building object, and then optimize the operation strategy of the heat exchange device to improve operating efficiency and reduce energy consumption.

[0094] The third feature may characterize the overall temperature conditions corresponding to key spatial points with different weight factors within the target building object.

[0095] In the heating system, different spatial points within the target building object often have different temperature requirements and weight factors. For example, the temperature requirements in the office area are usually higher, while the temperature requirements in the storage room or corridor are relatively lower. Therefore, by analyzing and utilizing the temperature data of key spatial points (such as workstations, locations near windows, doorways, etc.) and considering the weight factors of key spatial points, the overall temperature conditions within the target building object can be more accurately reflected.

[0096] The fourth feature can characterize the changing trend of the heating quantity in a specified time period.

[0097] The changing trend of heating heat can reflect the operating status and efficiency of the heating system. By analyzing the changing trend of heating heat in a specified time period, we can understand the heating capacity and stability of the heating system, and whether there is overheating or underheating.

[0098] Step S103: Process the target features based on the target model to obtain heating parameters; the heating parameters are used to adjust the operating parameters of the heat exchange device.

[0099] In this embodiment, the heating parameters may include but are not limited to: at least one of the secondary network water supply temperature and the secondary network water supply temperature difference.

[0100] The secondary network water supply temperature can determine the temperature of the hot water received by the heat user (eg, the air conditioning system in the target building object).

[0101] The secondary network water supply temperature difference can be understood as the difference between the secondary network water supply temperature and the return water temperature of the plate heat exchanger. The secondary network water supply temperature difference can reflect the transfer efficiency of the heat exchange device.

[0102] The operating parameters of the heat exchange device may include but are not limited to: the operating frequency and operating time of the water pump and the opening of the regulating valve of the plate heat exchanger.

[0103] In the present application, target features are obtained by obtaining target time series data corresponding to the heat exchange device of the target building object and processing the target time series data, and the target features include at least a first feature, a second feature, a third feature and a fourth feature; the first feature represents the time information associated with the heating demand in the target time series data; the second feature represents the data associated with the heat storage of the target building object in the target time series data; the third feature represents the overall temperature conditions corresponding to the key spatial points with different weight factors in the target building object; the fourth feature represents the change trend of the heating quantity in a specified time period. After obtaining the target features, the target features can be processed based on the target model, so that the target model can comprehensively analyze and predict the heating demand from multiple dimensions such as time, heat storage, temperature of key spatial points and change trend of heating quantity, and then obtain accurate heating parameters. The heating parameters are used to adjust the operating parameters of the heat exchange device, which can avoid the increase of energy consumption caused by excessive heating or insufficient heating, thereby saving energy.

[0104] In this embodiment, the real-time operation information of the heating system, such as the primary network thermal parameters, the temperature, humidity, meteorological parameters of the target building object, the start / stop status and frequency of the primary network water pump, the opening status of the plate heat exchanger and the solenoid valve, and the start / stop status and frequency of the secondary network water pump, can be visualized.

[0105] The heating system can be set in smart mode, automatic mode, and manual mode. The data processing method can be executed in the smart mode, and the edge-side PLC (Programmable Logic Controller) can use heating parameters such as the secondary network water supply temperature and the secondary network water supply temperature difference as control targets to adjust the start and stop of the water pump and the operating frequency, the plate change regulating valve opening, etc. In the automatic mode, the PLC group control logic can be retained, and the system operation status can be adjusted by manually changing the secondary network water supply temperature and the secondary network water supply temperature difference. In manual mode, only the remote control function can be retained, and the equipment start and stop, valve opening, etc. can be directly operated. As a supplement to the smart mode, the automatic mode and manual mode can take over the system control in time in extremely cold weather or when the system status is abnormal, so as to avoid production incidents.

[0106] For example, Figure 3 As shown, operators can select smart mode on the heating system status and control page.

[0107] like Figure 4 As shown, the operator can select the automatic mode on the heating system status and control page. In the automatic mode, the threshold value can be set manually, for example, the set temperature (i.e., the secondary network water supply temperature) is 38.0°C and the set temperature difference is 3.0°C.

[0108] As another optional embodiment of the present application, a data processing method is provided in Embodiment 2 of the present application. This embodiment is mainly an implementation method of the first feature in Embodiment 1. The first feature may include but is not limited to at least one of the following:

[0109] The first time feature represents the global time information associated with the heating demand in the target time series data.

[0110] In this embodiment, the global time information may include but is not limited to: season, month, date, day of the week, and specific time period of a day (such as morning, noon, evening), etc.

[0111] In this embodiment, the heating demand usually changes with the seasons. For example, in winter, the heating demand is higher, while in summer, the heating demand is lower. By identifying seasonal information, the system can automatically adjust the heating parameters to adapt to different seasons.

[0112] For example, the target building object may include an office building. On weekdays, the heating demand is high because employees need to go to work, while on weekends, the heating demand is relatively low because the flow of people in the office building is reduced. The target model can optimize the heating parameters according to the first time characteristics (for example, the day of the week) to reduce unnecessary energy waste.

[0113] The second time feature characterizes a key time point in the target time series data associated with the internal temperature change of the target building object.

[0114] The key time points associated with the temperature change inside the target building object may include, but are not limited to, temperature increase key time points or temperature decrease key time points.

[0115] In this embodiment, by identifying key time points to predict heating parameters, energy can be used more efficiently.

[0116] For example, in the morning, the indoor temperature may be low, and as employees arrive at the office, the demand for heating is high. Accordingly, the target model can identify the critical time point for heating and increase the heating parameters to ensure indoor comfort.

[0117] In the evening, as employees leave the office, the demand for heating is lower. Accordingly, the target model can identify the critical time for cooling down and reduce the heating parameters to avoid excessive heating and energy waste.

[0118] In this embodiment, the target model processes the first time feature and the second time feature, and can combine the global time information and the key time nodes to obtain more accurate heating parameters. For example, more precise heating parameters can be obtained. For example, in the morning period of weekdays (the key time point for heating up), the target model can recognize that this is the peak period of heating demand in office buildings, and accordingly increase the heating parameters to meet the comfort needs of employees; and in the evening period of weekdays (the key time point for cooling down), the target model can reduce the heating parameters to save energy.

[0119] As another optional embodiment of the present application, a data processing method is provided in Embodiment 3 of the present application. This embodiment is mainly an implementation of step S102 in Embodiment 2. Step S102 may include but is not limited to at least one of the following:

[0120] Step S11, encode the date in the target time series data to obtain a date encoding feature, and add a category feature to the date encoding feature to obtain a first time feature; the category feature represents the holiday to which the date belongs.

[0121] In this embodiment, the date in the target time series data can be converted into a numerical code, and the numerical code can be used as a date code feature. For example, January 1, 2024 can be converted into a numerical code 20230101.

[0122] In this embodiment, the calendar information can be queried to determine whether the date in the target time series data is a holiday. If the date is a holiday, a category feature can be added to the date coding feature. The category feature can be used to characterize the holiday to which the date belongs, such as the Spring Festival, National Day, workday, weekend, etc.

[0123] Step S12: Select key time points from the target time series data, and construct one-hot encoding features for the key time points as second time features.

[0124] In this embodiment, the temperature increase time interval and the temperature decrease time interval may be determined by the target time series data.

[0125] Perform trend analysis on the temperature data in the heating time interval and the cooling time interval, and select representative time points as key time points. For example, the time points corresponding to the peak value, valley value or turning point of temperature change are selected as key time points.

[0126] Construct one-hot encoding features for key time points, which may include but are not limited to:

[0127] For each key time point, create a one-hot encoding vector. The length of this vector is equal to the total number of key time points, and only one element in the vector is 1, and the rest are 0. This 1 element corresponds to the current key time point.

[0128] For example, select three key time points from the target time series data, namely A, B, and C. For key time point A, the one-hot encoding vector is constructed as [1,0,0]. For key time point B, the one-hot encoding vector is constructed as [0,1,0]. For key time point C, the one-hot encoding vector is constructed as [0,0,1].

[0129] In this embodiment, by adding category features (such as holidays, weekdays, weekends, etc.) to the date coding features, the target model can more accurately capture the changing pattern of heating demand with date, and more accurately adjust the heating parameters, thereby avoiding energy waste. In addition, by selecting key time points from the target time series data and constructing a unique hot coding feature for the key time points as the second time feature, the target model can quickly and accurately capture the changes in heating and cooling, and adjust the heating parameters accordingly, so as to more accurately adapt to the changes in heating demand, reduce energy waste, and improve user satisfaction.

[0130] As another optional embodiment of the present application, a data processing method is provided in Embodiment 4 of the present application. This embodiment is mainly an implementation of step S102 in Embodiment 1. Step S102 may include but is not limited to the following steps:

[0131] Step S21: Acquire the temperature value of each key spatial point in each area of ​​at least one area in the target building object from the target time series data.

[0132] In this embodiment, the heating quality evaluation index of the heating system may depend on whether the indoor temperature and the terminal fan outlet temperature are within a reasonable range acceptable to the user. Therefore, in order to accurately evaluate the indoor temperature, key spatial points in each area of ​​at least one area in the target building object may be selected for temperature monitoring.

[0133] Key spatial points can be representative locations in an area, such as the center, a corner, near a window or a door, etc.

[0134] Step S22: Based on the weight factor of each of the key spatial points, the temperature value of each of the key spatial points is processed to obtain the temperature value of the area.

[0135] Due to factors such as room orientation and direct sunlight, the temperature values ​​of different key spatial points may vary. In order to comprehensively consider these differences, different weight factors can be assigned to each key spatial point. The weight factor can represent the importance of the key spatial point in indoor temperature assessment.

[0136] In this embodiment, the temperature value of the area can be determined by the following relationship:

[0137] t i =λ1t i1 +λ2t i2 +…+λ n t in

[0138] t i represents the temperature value of the ith region; t i1 ,t i2 ,…,t in Respectively represent the temperature values ​​of each key spatial point in the i-th region; λ1, λ2, …, λ n Respectively represent the weight factors of each key spatial point in the i-th region.

[0139] Step S23: determining a plurality of temperature attribute values ​​based on the temperature values ​​of each of the regions; the plurality of temperature attribute values ​​characterize the temperature distribution of the at least one region.

[0140] In this embodiment, the multiple temperature attribute values ​​may include but are not limited to:

[0141] Mean, upper quantile, lower quantile, maximum, and minimum values ​​of temperature.

[0142] The temperature mean may represent an average level of temperature in at least one region.

[0143] The upper quantile can characterize the typical value of a portion of data with higher temperature in at least one region.

[0144] The lower quantile can characterize the typical value of a portion of data with lower temperature in at least one region.

[0145] The maximum and minimum values ​​may characterize temperature extremes in at least one region.

[0146] Step S24: Process the multiple temperature attribute values ​​to obtain a third feature.

[0147] In this embodiment, in order to more comprehensively reflect the indoor temperature conditions in the target building object, the multiple temperature attribute values ​​may be processed to obtain an aggregate temperature, and the aggregate temperature is used as the third feature.

[0148] In this embodiment, the polymerization temperature can be determined by the following relationship:

[0149]

[0150] t represents the polymerization temperature, T mean represents the mean temperature, represents the lower quantile, represents the upper quantile, T min and T max Represent the minimum and maximum values ​​respectively, and α and β represent weights. When abnormally high or low temperatures occur, the weights of the maximum and minimum values ​​are increased, making the influence of the extreme values ​​in the aggregation temperature more significant.

[0151] In this embodiment, the third feature comprehensively considers multiple temperature attribute values ​​and can more comprehensively and accurately reflect the indoor temperature situation. Therefore, the target model processes the third feature and can timely perceive the change of the terminal fan outlet temperature, thereby improving the accuracy of the predicted heating parameters. Figure 5 As shown in the figure, there is a strong correlation between the terminal fan outlet temperature and some indoor temperature points and the secondary network water supply temperature of the heating system. This correlation shows that by considering the temperature mean, upper quantile, lower quantile, maximum and minimum values ​​of key spatial points, the target model can more accurately capture the actual operating status of the heating system, and make more accurate heating parameters based on this, so as to timely adjust the operating parameters of the heat exchange device.

[0152] As another optional embodiment of the present application, a data processing method is provided in Embodiment 5 of the present application. This embodiment is mainly an implementation method of step S102 in Embodiment 1. Step S102 may include but is not limited to the following steps:

[0153] Step S31: Based on the target time series data, obtain the current heat characteristics of the heat supply source of the heat exchange device.

[0154] In this embodiment, the heat load can be calculated based on the temperature difference between the primary network water supply and return water in the target time series data, and the calculated heat load is used as the current heat characteristic.

[0155] In this embodiment, the heat load per minute can be calculated as follows:

[0156] E=Cρq(t)(T p,s (t)-T p,r (t))

[0157] Where C is the specific heat of water (J / kg·℃), ρ is the density of water (kg / m 3 ), q(t) represents the primary network water supply flow (m 3 / h), T p,s (t) represents the primary network water supply temperature in the target time series data, T p,r (t) represents the primary network return water temperature in the target time series data.

[0158] In order to more intuitively understand the dynamic characteristics of heat load, a graph showing the change of heat load over time can be drawn, for example Figure 6 As shown, the relationship between the heat load and time can be visualized.

[0159] In this embodiment, the heat load per minute may be accumulated according to the length of the time interval (eg, 15 minutes) to obtain the heat load of each time interval.

[0160] Step S32: determine the heat component based on the current heat characteristics; the thermal component is used to reflect the change of the current heat data in the time domain.

[0161] Since the heat load in each time interval has non-stationary and nonlinear characteristics, the time scale characteristics of the heat load in each time interval can be analyzed based on the empirical mode decomposition (EMD) algorithm.

[0162] In this embodiment, the heat load of the time interval can be decomposed using the following relationship to obtain the IMF component:

[0163]

[0164] Among them, IMF i (t) represents the i-th intrinsic mode function (or, IMF component), r nResidual. The intrinsic mode function can meet certain conditions: within the time range, the function has the same number of local extreme points and zero crossing points (or at most one difference); at any time point, the mean of the envelope of the local maximum (upper envelope) and the local minimum (lower envelope) must be zero.

[0165] In this embodiment, the IMF component may be used as the heat component.

[0166] Step S33: Based on the heat component, determine the predicted heat characteristics of the heat source of the heat exchange device in a specified time interval after the current time interval.

[0167] In this embodiment, the IMF components obtained after EMD decomposition of each time interval can be used as partial features representing the current heat characteristics, and combined with other relevant factors such as weather information, they can be input into the LightGBM model for training. Through training, the LightGBM model can learn the changing laws of heat load and predict the heat characteristics of a specified time interval (e.g., the next time interval) after the current time interval based on these laws.

[0168] Step S34: taking the current heat feature and the predicted heat feature as the fourth feature.

[0169] In this embodiment, the building heating system often has the characteristics of nonlinearity, time lag, thermal inertia and time variation, which makes it very difficult to predict the heating parameters of the next time interval by relying solely on meteorological information and real-time equipment operation data. However, by determining the current heat characteristics and predicted heat characteristics, the current state and future change trend of the heat exchange device's heat source can be reflected. Taking the current heat characteristics and predicted heat characteristics as the fourth characteristics can significantly improve the accuracy and reliability of the target model's prediction of heating parameters.

[0170] As another optional embodiment of the present application, a data processing method is provided in Embodiment 6 of the present application. This embodiment is mainly an implementation method of the target features in Embodiment 1. The target features may include but are not limited to:

[0171] The first feature, the second feature, the third feature, the fourth feature and the fifth feature.

[0172] For understanding of the first feature, the second feature, the third feature, and the fourth feature, please refer to the relevant introduction in Example 1, which will not be repeated here.

[0173] The fifth feature can be used to characterize the rate of change of heat consumption of the heat supply source of the heat exchange device.

[0174] In this embodiment, the calorimeter data collected by the primary network can be obtained from the target time series data.

[0175] Next, the calorie meter data may be differentially calculated to obtain a calorie consumption difference value. For example, the calorie consumption difference value of 15 minutes before and after (or other predetermined time intervals) may be calculated to reflect the change in calorie consumption in a short period of time.

[0176] The calorie consumption difference obtained by differential calculation can be further used to calculate the calorie consumption change rate per unit time, and the calorie consumption change rate is used as the fifth feature.

[0177] The heat consumption change rate can reflect the heat consumption speed of the heating source and the fluctuation of heat consumption in different time periods.

[0178] In this embodiment, the heat consumption change rate is used as the fifth feature, which can provide richer information to the target model. The target model can more accurately capture the heat supply of the heating source based on the fifth feature, thereby more accurately predicting the heating parameters.

[0179] As another optional embodiment of the present application, a data processing method is provided in Embodiment 7 of the present application. This embodiment is mainly an implementation method for training the target model in Embodiment 1. The target model can be trained in the following manner but is not limited to:

[0180] Step S41, obtaining sample time series data and target heating parameters corresponding to the heat exchange device of the target building object.

[0181] In this embodiment, the sample time series data can be understood by referring to the introduction of the target time series data in Embodiment 1, which will not be repeated here.

[0182] In this embodiment, considering the lag of the heating system, in order to more effectively capture the dynamic changes of the system and reduce data redundancy, the frequency of the sample time series data (frequency is once per minute) can be downsampled to increase the information entropy of the sample time series data, that is, to reduce the amount of data without losing key information, so that subsequent analysis and modeling are more efficient. For example, the collection frequency of sample time series data can be reduced from once per minute to once every 15 minutes. This step

[0183] Also, due to the complexity of the Internet of Things and network environment, the sample time series may contain expired data, and the timestamps between different measurement points may not be aligned. To address these issues, interpolation data processing methods can be used. For example, for expired data, linear or nonlinear interpolation is performed based on adjacent valid data points to estimate the missing data values. For the problem of timestamp misalignment, interpolation or resampling technology is used to align the data of all measurement points on a unified time scale.

[0184] In this embodiment, the target function parameter may include: secondary network water supply temperature setting values ​​at different time points. The secondary network water supply temperature setting values ​​may be pre-set based on expert experience or actual needs of the heating system.

[0185] The change of the secondary network water supply temperature set value over time can form a curve. In the curve of the secondary network water supply temperature set value over time, the places with large fluctuations are usually considered to be caused by overshoot. These abnormal fluctuations will have a negative impact on subsequent model training and prediction. In order to eliminate these abnormal fluctuations, filtering and smoothing technology can be used to smooth the curve to reduce the fluctuation amplitude while retaining the overall trend of the curve. After filtering and smoothing, the abnormal fluctuations in the secondary network water supply temperature set value curve are eliminated, for example, Figure 7 As shown, the curve becomes smoother.

[0186] Step S42: Process the sample time series data to obtain sample features.

[0187] In this embodiment, the target time series data can be processed by referring to the method of processing the target time series data in the aforementioned embodiment to obtain the target features, and the sample time series data is processed, which will not be described in detail here.

[0188] Step S43: Process the sample features based on multiple first learning modules to obtain first heating parameters determined by each of the first learning modules.

[0189] These learning modules are built based on different machine learning algorithms and are designed to extract information related to heating parameters from sample features. Each learning module analyzes and learns the input features independently to output its own first heating parameters. These parameters are the model's initial prediction or estimate of heating demand, providing the basis for fusion and optimization in subsequent steps.

[0190] Multiple first learning modules can be selected by:

[0191] The control of heat exchange devices is affected by multiple factors such as equipment mechanism, meteorological conditions, building characteristics, office rhythm, etc., and human factors have a greater impact. The heating targets and operation modes under different conditions and periods vary greatly, and it is difficult to model the scenarios through traditional operations research algorithms. Therefore, it is considered to convert the scheduling optimization of heat exchange devices into an imitation learning scheme, learn expert experience through early open-loop operation data, establish the relationship between heat load, indoor temperature, meteorological data, and heat load, realize the dynamic adjustment of secondary network water supply temperature and temperature difference according to working conditions, and reduce the heat consumption of the heat exchange station system.

[0192] The historical data corresponding to the heat exchange device is usually relatively small. Generally, only one to two weeks of data can be obtained at the beginning of the project. Therefore, it is difficult to extract control experience through deep learning methods. Ensemble learning based on additive models can effectively solve this problem. The key to ensemble learning lies in the selection of the base model. The higher the accuracy and the greater the difference of the selected model, the better the generalization and accuracy of the final model. Therefore, this embodiment uses automated machine learning technology to design and use a common regression model as the imitation learning base model. After setting the initialization parameters, it is trained on the training set and the determination coefficient (R 2 ), mean square error (MSE), root mean square error (RMSE) and other indicators are used to evaluate the performance of each base model.

[0193] By comparing the performance indicators of different models, the top n base models with different types and better performance can be screened out. For example, the determination coefficient indicators of each base model can be found in Table 1. As shown in Table 1, CatBoost Regressor, Random Forest Regressor and Bayesian Ridge can be screened out.

[0194] Table 1

[0195]

[0196]

[0197]

[0198] In this embodiment, the first n base models of different types and with better performance can be screened out and used as the first learning modules respectively.

[0199] Step S44: adjust the parameters of each of the first learning modules based on the first losses corresponding to each of the first learning modules.

[0200] The first loss may represent a difference between the first heating parameter and the target heating parameter.

[0201] Step S45, fixing the parameters of each of the first learning modules, inputting the first heating parameters determined by each of the first learning modules into the second learning module, and obtaining the second heating parameters determined by the second learning module.

[0202] Step S46: Adjust the parameters of the second learning module based on the second loss.

[0203] The second loss represents a difference between the second heating parameter and the target heating parameter.

[0204] The trained second learning module can outperform the single model in multiple key performance indicators. For example, according to the final model indicators shown in Table 2, the second learning module has a better determination coefficient (R 2 ), which shows that the correlation between the model prediction value and the actual value is very high; at the same time, the mean square error (MSE) and root mean square error (RMSE) of the second learning module are also relatively low, which further proves the accuracy of the model prediction. In addition, due to the characteristics of ensemble learning, the second learning module also has good generalization ability and can maintain stable prediction performance on unseen data.

[0205] Table 2

[0206] MAE MSE RMSE <![CDATA[R 2 ]]> RMSLE MAPE 0.3896 1.5994 1.1879 0.876 0.0414 0.0136

[0207] In this embodiment, the second learning module may include but is not limited to: XGBRegressor (XGBoost regression model). Figure 8 As shown, based on CatBoost Regressor, Random Forest Regressor and Bayesian Ridge, the sample features are processed respectively to obtain first heating parameters determined by CatBoost Regressor, Random Forest Regressor and Bayesian Ridge; based on the first losses corresponding to CatBoost Regressor, RandomForest Regressor and Bayesian Ridge respectively, the parameters of CatBoost Regressor, Random Forest Regressor and Bayesian Ridge are adjusted, and after the training of CatBoost Regressor, RandomForest Regressor and Bayesian Ridge is completed, the parameters of CatBoost Regressor, RandomForest Regressor and Bayesian Ridge are fixed, and the first heating parameters determined by CatBoost Regressor, Random ForestRegressor and Bayesian Ridge are input into XGBRegressor to obtain the second heating parameters determined by XGBRegressor; based on the second loss, the parameters of XGBRegressor are adjusted.

[0208] In this embodiment, the performance of the target model can be comprehensively evaluated on the test set, and the test results can be visualized. Fig. 9As shown in the figure, the blue line clearly depicts the prediction results of the target model, while the red line represents the actual secondary network water supply temperature on that day. By comparing these two lines, it can be intuitively seen that the error between the prediction results and the actual secondary network water supply temperature is very small, which can prove the high accuracy and reliability of the target model.

[0209] Further observation Fig. 9 , it can be found that the target model can adjust quickly during the critical heating and cooling time periods. Whether the temperature rises sharply or drops rapidly, the target model can quickly capture this change and make timely and accurate prediction adjustments.

[0210] In this embodiment, the residual of the target model can also be visualized on the test set, such as Fig.10 As shown, the absolute value of the residual is controlled at about 1 degree, indicating that the target generalization ability is strong and can meet the temperature control requirements of the heating system.

[0211] As another optional embodiment of the present application, a data processing method provided in Embodiment 8 of the present application may include but is not limited to the following steps:

[0212] Step S201: Obtain target time series data corresponding to a heat exchange device of a target building object.

[0213] Step S202, processing the target time series data to obtain target features; the target features include at least a first feature, a second feature, a third feature and a fourth feature; the first feature represents the time information associated with the heating demand in the target time series data; the second feature represents the data associated with the heat storage of the target building object in the target time series data; the third feature represents the overall temperature conditions corresponding to the key spatial points with different weight factors in the target building object; the fourth feature represents the change trend of the heating quantity in a specified time period;

[0214] Step S203: Process the target features based on the target model to obtain heating parameters; the heating parameters are used to adjust the operating parameters of the heat exchange device; the target model includes multiple first learning modules and second learning modules.

[0215] The training process of multiple first learning modules and second learning modules can be found in the relevant introduction in Example 7, which will not be repeated here.

[0216] Step S204: Based on the target characteristics and the heating parameters, the first learning modules and the second learning modules are evaluated to obtain index values ​​of the first learning modules and the second learning modules.

[0217] As weather conditions and personnel mobility change, the data distribution of the collection points may drift, which may cause the deviation between the predicted value of the target model and the actual optimal secondary network water supply temperature to gradually increase. In order to capture this data drift in time and evaluate the adaptability of the target model to the current working conditions, the actual data of the day (i.e., target features) can be used as input to make predictions through the trained target model and calculate the determination coefficient (R 2 ), mean square error (MSE), mean absolute error (MAE) and other indicator values.

[0218] The indicator value can be used to characterize the accuracy of each of the first learning module and the second learning module.

[0219] Step S205: Determine that the indicator value does not meet the set indicator threshold, and retrain the first learning module and the second learning module.

[0220] When the index value does not meet the set index threshold, a warning may be issued, indicating that the current target model may no longer be suitable for the current working conditions. At this time, more historical operating data of the heat exchange device may be collected to retrain each of the first learning module and the second learning module to improve their adaptability and accuracy. The retraining process can refer to the relevant introduction of steps S41-S46 in Example 7, which will not be repeated here.

[0221] In addition, this embodiment can also provide a model monitoring page, such as Fig.11 As shown in the figure, this page can display the version information of the target model, various evaluation indicators (such as determination coefficient, mean square error, etc.) on the current data set, and the operation status of the target model. This information is very valuable for operators because it can help them accurately grasp the fitness of the model and take appropriate measures when necessary, such as manually triggering model updates.

[0222] When the automatic model update cannot meet the existing working conditions, the operator can manually click the page button in the model monitoring page to manually trigger the model update, which will automatically obtain historical operation data for model training and hot update.

[0223] Historical operation data may include but are not limited to TSP (actual secondary network water supply temperature), T108 (air conditioning manifold temperature), T109 (air conditioning collector temperature), Tdsp (actual secondary network water supply temperature difference), F102 (air conditioning water supply flow), WindSpeed ​​(wind speed), WindPower (wind force), WindDirection (wind direction), DewTemperature (dew point temperature), Illumination (irradiance), TotalRadiation (total radiation), WindDirectionDegree (wind direction), AirHumidity (air humidity), AirTemperature (air temperature), Noise (noise), PreTotlE (primary network accumulated heat), PreP (primary network instantaneous heat), PreF (primary network instantaneous flow), PreTotalF (primary network accumulated flow), PreSupplyT (primary network water supply temperature), PreReturnT (primary network return water temperature), PreDeltT (primary network supply and return water temperature difference), P101 (target building object water supply pressure), etc.

[0224] At the same time, in order to enhance the readability of data and the understanding of operators, historical operation data (including the environmental factors and operation parameters mentioned above) and key indicators in the model training process (such as actual secondary network water supply temperature, actual secondary network water supply temperature difference, wind speed, wind force, etc.) can be visualized. Fig.12 As shown in the figure, these visualization charts can help operators understand the performance and data distribution of the target model more intuitively.

[0225] After the model training is completed, the data of the day can be used again for simulation testing to verify whether the model's fitting coefficients and prediction results meet the online standards. 2 The closer the value is to 1, the better the model fit is. 2 A low value of may indicate that the model is less adaptable to the current working conditions.

[0226] The simulation results can be visualized. Fig.13 As shown, the simulation results can help operators understand the simulation results of the target model more intuitively.

[0227] As another optional embodiment of the present application, a data processing method provided in Embodiment 9 of the present application may include but is not limited to the following steps:

[0228] Step S301: Obtain target time series data corresponding to a heat exchange device of a target building object.

[0229] Step S302: Process the target time series data to obtain target features; the target features include at least a first feature, a second feature, a third feature and a fourth feature; the first feature represents the time information associated with the heating demand in the target time series data; the second feature represents the data associated with the heat storage of the target building object in the target time series data; the third feature represents the overall temperature conditions corresponding to key spatial points with different weight factors in the target building object; the fourth feature represents the changing trend of the heating heat in a specified time period.

[0230] Step S303: Process the target features based on the target model to obtain heating parameters; the heating parameters are used to adjust the operating parameters of the heat exchange device.

[0231] The detailed process of steps S301-S303 can refer to the relevant introduction of steps S101-S103 in Example 1, which will not be repeated here.

[0232] Step S304: Perform a first output constraint on the heating parameter so that the regulation frequency of the heating parameter by the heat exchange device satisfies a set regulation frequency threshold.

[0233] In this embodiment, the set adjustment frequency range (e.g., once every 25 to 40 minutes) can be determined by analyzing data of typical projects, taking into account factors such as model scheduling frequency, model output disturbance, system response time, and secondary network pipeline heating lag, and a regulation frequency can be selected from the set adjustment frequency range as the set adjustment frequency threshold.

[0234] In this embodiment, by performing the first output constraint on the heating parameters, it can be ensured that the frequency of adjusting the heating parameters by the heat exchange device will not be too fast, thereby avoiding a decrease in heating quality due to over-adjustment of the heating amount.

[0235] Step S305: Perform a second output constraint on the heating parameter, so that the change process of the heat exchange device from the current heating parameter to the heating parameter is relatively stable.

[0236] Since there is a certain range between the heating temperature of the municipal pipe network and the heating temperature of the building secondary network (such as the heating temperature of the municipal pipe network is 50-90 degrees Celsius, and the heating temperature of the building secondary network is 28-35 degrees Celsius), the heating system can respond quickly to the heating demand. However, excessive heating temperature adjustment will cause the end users to feel alternating cold and warm, thus affecting the heating quality. Therefore, the heating parameters (such as the secondary network water supply temperature and the secondary network water supply temperature difference) can be smoothed, and the temperature can be increased or decreased in a step-by-step manner to ensure the stability of the heating process.

[0237] Step S306: Perform a third output constraint on the heating parameter so that the heating parameter obtained by the heat exchange device is within a set upper and lower limit range.

[0238] Setting the upper and lower limits may include but is not limited to: setting at least one of the upper and lower limits of the secondary network water supply temperature and the upper and lower limits of the temperature difference.

[0239] In this embodiment, by setting reasonable upper and lower limits, the heating quality and efficiency of the heat exchange device can be guaranteed, and energy waste or insufficient heating caused by overheating or overcooling can be avoided.

[0240] Step S307: performing a fourth output constraint on the heating parameters so that the heat obtained by the target building object meets the overall heat demand and the heating flow and heating temperature of the target monitoring area in the target building object meet the set area demand.

[0241] Due to factors such as building design, pipeline layout or flow distribution, the heating effect in the set areas (such as key terminal areas) of the target building object may be more easily affected. In order to ensure that the heating needs of these areas are met and avoid user experience problems caused by insufficient or overheating, the heating flow and heating temperature of these areas need to be strictly monitored and adjusted.

[0242] In this embodiment, by performing the fourth output constraint on the heating parameters, it can be ensured that the set area in the target building object receives sufficient attention during the heating process, thereby improving the overall heating quality and user experience.

[0243] In this embodiment, one or more steps of steps S304-S307 may be selected for execution as needed, and this is not limited in this application.

[0244] As another optional embodiment of the present application, a data processing method is provided in Embodiment 10 of the present application. In this embodiment, the data processing method can be applied to an edge computing server. The method may include but is not limited to the following steps:

[0245] Step S401: Obtain target time series data corresponding to a heat exchange device of a target building object.

[0246] Step S402: Process the target time series data to obtain target features.

[0247] The target features include at least a first feature, a second feature, a third feature and a fourth feature.

[0248] The first feature represents time information associated with heating demand in the target time series data.

[0249] The second feature represents data associated with heat storage of the target building object in the target time series data.

[0250] The third feature characterizes the overall temperature conditions corresponding to the key spatial points with different weight factors in the target building object.

[0251] The fourth feature represents the changing trend of the heating quantity in a specified time period;

[0252] Step S403: Process the target features based on the target model to obtain heating parameters; the heating parameters are used to adjust the operating parameters of the heat exchange device.

[0253] The detailed process of steps S401-S403 can refer to the relevant introduction of steps S101-S103 in Example 1, which will not be repeated here.

[0254] Step S404: Based on the operating status of the edge computing server, determine whether to send a periodic signal to the control device of the heat exchange device, so that when the control device receives the periodic signal, it adjusts the operating parameters of the heat exchange device based on the heating parameters. When the periodic signal is not received, it is prohibited to use the heating parameters to adjust the operating parameters of the heat exchange device.

[0255] In an actual production environment, there is a problem of edge computing server service downtime or network fluctuations leading to data uplink and downlink failures, so that the target model cannot generate effective output. Therefore, the control device of the heat exchange device can be designed (such as a communication detection mechanism between the edge-side PLC and the edge computing server. When the edge computing server operates normally and the model output is normal, a periodic signal (such as a pulse signal) is sent to the PLC, and the PLC only accepts the target model output when the pulse is normal. When no pulse signal is received or a DC signal is received, the PLC no longer executes the target model output, switches to automatic mode to run the group control logic and issues an alarm to remind the operating personnel to deal with the problem in time, which can avoid potential erroneous operations.

[0256] In this embodiment, combined with Fig.14 , the process of scheduling and optimizing the heating system based on the edge computing server is explained. Fig.14 As shown, the edge computing server can verify and clean the continuously collected data of various parameters of the thermal exchange device to obtain the target time series data.

[0257] Feature engineering: The edge computing server can process the target time series data to obtain the target features.

[0258] Imitation learning: The edge computing server can process the target features based on the target model obtained through model learning training to obtain heating parameters.

[0259] Expert experience constraints: The edge computing server can constrain the heating parameters (e.g., the first output constraint, the second output constraint, the third output constraint, and the fourth output constraint).

[0260] PLC instruction filtering: When the edge computing server runs normally and the model output is normal, a periodic signal (such as a pulse signal) is sent to the PLC, and the PLC only accepts the target model output when the pulse is normal. When no pulse signal is received or a DC signal is received, the PLC no longer executes the target model output, switches to automatic mode to run the group control logic and issues an alarm to remind the operator to deal with the problem in time.

[0261] Next, the data processing device provided by the present application is introduced. The data processing device introduced below and the data processing method introduced above can be referenced to each other.

[0262] Please refer to Fig.15 The data processing device includes: an acquisition module 100, a first processing module 200 and a second processing module 300.

[0263] The acquisition module 100 is used to obtain target time series data corresponding to the heat exchange device of the target building object.

[0264] The first processing module 200 is used to process the target time series data to obtain target features; the target features include at least a first feature, a second feature, a third feature and a fourth feature; the first feature represents the time information associated with the heating demand in the target time series data; the second feature represents the data associated with the heat storage of the target building object in the target time series data; the third feature represents the overall temperature conditions corresponding to key spatial points with different weight factors in the target building object; the fourth feature represents the changing trend of the heating heat in a specified time period.

[0265] The second processing module 300 is used to process the target feature based on the target model to obtain heating parameters; the heating parameters are used to adjust the operating parameters of the heat exchange device.

[0266] In this embodiment, the first feature may include at least one of the following:

[0267] The first time feature represents the global time information associated with the heating demand in the target time series data;

[0268] The second time feature characterizes a key time point in the target time series data associated with the internal temperature change of the target building object.

[0269] The process in which the first processing module 200 processes the target time series data to obtain target features may specifically include at least one of the following:

[0270] Encode the date in the target time series data to obtain a date encoding feature, and add a category feature to the date encoding feature to obtain a first time feature; the category feature represents the holiday to which the date belongs;

[0271] A key time point is selected from the target time series data, and a one-hot encoding feature is constructed for the key time point as a second time feature.

[0272] In this embodiment, the process in which the first processing module 200 processes the target time series data to obtain target features may specifically include:

[0273] Acquire the temperature value of each key spatial point in each area of ​​at least one area in the target building object from the target time series data;

[0274] Based on the weight factors of the key spatial points, the temperature values ​​of the key spatial points are processed to obtain the temperature value of the area;

[0275] Based on the temperature values ​​of each of the regions, determining a plurality of temperature attribute values; the plurality of temperature attribute values ​​characterizing the temperature distribution of the at least one region;

[0276] The multiple temperature attribute values ​​are processed to obtain a third feature.

[0277] The first processing module 200 processes the target time series data to obtain target features, which may specifically include:

[0278] Based on the target time series data, obtaining the current heat characteristics of the heat supply source of the heat exchange device;

[0279] Based on the current heat characteristics, determining a heat component; the thermal component is used to reflect the change of the current heat data in the time domain;

[0280] Determining, based on the heat component, a predicted heat characteristic of a heat supply source of the heat exchange device in a specified time interval after a current time interval;

[0281] The current heat feature and the predicted heat feature are used as the fourth feature.

[0282] The target features may also include:

[0283] The fifth feature is used to characterize the rate of change of heat consumption of the heat supply source of the heat exchange device.

[0284] The data processing device may further include: a training module, configured to:

[0285] Obtain sample time series data and target heating parameters corresponding to the heat exchange device of the target building object;

[0286] Processing the sample time series data to obtain sample features;

[0287] Processing the sample features based on a plurality of first learning modules to obtain first heating parameters determined by each of the first learning modules;

[0288] Based on the first losses corresponding to the first learning modules, adjusting the parameters of the first learning modules; the first losses represent the difference between the first heating parameters and the target heating parameters;

[0289] Fixing the parameters of each of the first learning modules, inputting the first heating parameters determined by each of the first learning modules into the second learning module, and obtaining the second heating parameters determined by the second learning module;

[0290] Based on a second loss, the parameters of the second learning module are adjusted; the second loss represents the difference between the second heating parameter and the target heating parameter.

[0291] The data processing device may further include:

[0292] Update modules for:

[0293] Based on the target feature and the heating parameter, each of the first learning modules and the second learning modules is evaluated to obtain an index value of each of the first learning modules and the second learning modules; the index value is used to characterize the accuracy of each of the first learning modules and the second learning modules;

[0294] It is determined that the indicator value does not meet the set indicator threshold, and the first learning module and the second learning module are retrained.

[0295] The data processing device may further include:

[0296] A constraint module is used to perform at least one of the following:

[0297] Performing a first output constraint on the heating parameter so that the regulation frequency of the heating parameter by the heat exchange device satisfies a set regulation frequency threshold;

[0298] Performing a second output constraint on the heating parameter so that a change process of the heat exchange device from the current heating parameter to the heating parameter is relatively stable;

[0299] Performing a third output constraint on the heating parameter so that the heating parameter obtained by the heat exchange device is within a set upper and lower limit range;

[0300] A fourth output constraint is performed on the heating parameters so that the heat obtained by the target building object meets the overall heat demand and the heating flow and heating temperature of the target monitoring area in the target building object meet the set area demand.

[0301] In another embodiment of the present application, an electronic device is provided, comprising: a processor and a memory connected to the processor, wherein:

[0302] The memory is used to store computer programs;

[0303] The processor is used to execute the computer program so that the electronic device can implement the data processing method described in any one of the above embodiments 1-9.

[0304] In another embodiment of the present application, a storage medium is provided, which carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement the data processing method described in any one of the above embodiments 1-9.

[0305] It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

[0306] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0307] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0308] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from a website site, a computer, a training device, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device, or data center. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integrations. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)), etc.

Claims

1. A data processing method, comprising: Obtain target time series data corresponding to the heat exchange device of the target building object; Processing the target time series data to obtain target features; The target features include at least a first feature, a second feature, a third feature and a fourth feature; the first feature represents the time information associated with the heating demand in the target time series data; the second feature represents the data associated with the heat storage of the target building object in the target time series data; the third feature represents the overall temperature conditions corresponding to the key spatial points with different weight factors in the target building object; the fourth feature represents the change trend of the heating quantity in a specified time period; The target characteristics are processed based on the target model to obtain heating parameters; the heating parameters are used to adjust the operating parameters of the heat exchange device.

2. The data processing method according to claim 1, wherein the first feature comprises at least one of the following: The first time feature represents the global time information associated with the heating demand in the target time series data; The second time feature characterizes a key time point in the target time series data associated with the internal temperature change of the target building object.

3. According to the data processing method of claim 2, the target time series data is processed to obtain target features, including at least one of the following: Encode the date in the target time series data to obtain a date encoding feature, and add a category feature to the date encoding feature to obtain a first time feature; the category feature represents the holiday to which the date belongs; A key time point is selected from the target time series data, and a one-hot encoding feature is constructed for the key time point as a second time feature.

4. The data processing method according to claim 1, processing the target time series data to obtain target features, comprising: Acquire the temperature value of each key spatial point in each area of ​​at least one area in the target building object from the target time series data; Based on the weight factors of the key spatial points, the temperature values ​​of the key spatial points are processed to obtain the temperature value of the area; Based on the temperature values ​​of each of the regions, determining a plurality of temperature attribute values; the plurality of temperature attribute values ​​characterizing the temperature distribution of the at least one region; The multiple temperature attribute values ​​are processed to obtain a third feature.

5. The data processing method according to claim 1, processing the target time series data to obtain target features, comprising: Based on the target time series data, obtaining the current heat characteristics of the heat supply source of the heat exchange device; determining a heat component based on the current heat characteristic; The thermal component is used to reflect the change of the current heat data in the time domain; Determining, based on the heat component, a predicted heat characteristic of a heat supply source of the heat exchange device in a specified time interval after a current time interval; The current heat feature and the predicted heat feature are used as the fourth feature.

6. The data processing method according to claim 1, wherein the target feature further comprises: The fifth feature is used to characterize the rate of change of heat consumption of the heat supply source of the heat exchange device.

7. The data processing method according to claim 1, wherein the target model is trained by: Obtain sample time series data and target heating parameters corresponding to the heat exchange device of the target building object; Processing the sample time series data to obtain sample features; Processing the sample features based on a plurality of first learning modules to obtain first heating parameters determined by each of the first learning modules; Based on the first losses corresponding to the first learning modules, adjusting the parameters of the first learning modules; the first losses represent the difference between the first heating parameters and the target heating parameters; Fixing the parameters of each of the first learning modules, inputting the first heating parameters determined by each of the first learning modules into the second learning module, and obtaining the second heating parameters determined by the second learning module; Based on a second loss, the parameters of the second learning module are adjusted; the second loss represents the difference between the second heating parameter and the target heating parameter.

8. The data processing method according to claim 7, further comprising: Based on the target feature and the heating parameter, each of the first learning modules and the second learning modules is evaluated to obtain an index value of each of the first learning modules and the second learning modules; the index value is used to characterize the accuracy of each of the first learning modules and the second learning modules; It is determined that the indicator value does not meet the set indicator threshold, and the first learning module and the second learning module are retrained.

9. The data processing method according to claim 1, further comprising at least one of the following: Performing a first output constraint on the heating parameter so that the regulation frequency of the heating parameter by the heat exchange device satisfies a set regulation frequency threshold; Performing a second output constraint on the heating parameter so that a change process of the heat exchange device from the current heating parameter to the heating parameter is relatively stable; Performing a third output constraint on the heating parameter so that the heating parameter obtained by the heat exchange device is within a set upper and lower limit range; A fourth output constraint is performed on the heating parameters so that the heat obtained by the target building object meets the overall heat demand and the heating flow and heating temperature of the target monitoring area in the target building object meet the set area demand.

10. A data processing device, comprising: An acquisition module, used to obtain target time series data corresponding to a heat exchange device of a target building object; The first processing module is used to process the target time series data to obtain target features; the target features include at least a first feature, a second feature, a third feature and a fourth feature; the first feature represents the time information associated with the heating demand in the target time series data; the second feature represents the data associated with the heat storage of the target building object in the target time series data; the third feature represents the overall temperature conditions corresponding to the key spatial points with different weight factors in the target building object; the fourth feature represents the change trend of the heating quantity in a specified time period; The second processing module is used to process the target characteristics based on the target model to obtain heating parameters; the heating parameters are used to adjust the operating parameters of the heat exchange device.

Images