Large model-based data prediction method, apparatus and device, and storage medium

Abstract

The invention provides a data prediction method, device and equipment based on a large model and a storage medium, relates to the technical field of computers, in particular to the technical fields of data analysis, data prediction and the like, and can be applied to power consumption prediction, temperature prediction and other scenes. According to the specific implementation scheme, the method comprises the steps of obtaining historical sensing data, auxiliary data and historical time data of a sensor; performing matrix mapping on the historical perception data, the historical time data, the historical auxiliary data and the future auxiliary data to obtain a historical perception matrix, a historical time matrix and a future auxiliary matrix; generating a first fusion matrix according to the historical perception matrix, the historical time matrix and the historical auxiliary matrix; generating a second fusion matrix according to the first fusion matrix and the future auxiliary matrix; and mapping the second fusion matrix into a target time period to obtain prediction perception data of the sensor in the target time period. According to the invention, accurate prediction perception data can be determined, and the power consumption cost is saved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technologies, specifically to technical fields such as data analysis and data prediction, and can be applied to scenarios such as electricity consumption prediction and temperature prediction. In particular, the present disclosure relates to a data prediction method, apparatus, device, and storage medium. Background Art

[0002] In actual production and life, thermal inertia is a common situation in the scenarios of the HVAC control field. Thermal inertia refers to a means of predicting the temperature in an office building for a future period of time using the existing temperature data in a building. Especially in building HVAC, there is a very important need for determining thermal inertia. If it is possible to predict the heat change in a building for a future period of time based on the existing heat change situation (such as indoor temperature) in the building, and formulate corresponding electricity consumption strategies according to the prediction results, electricity consumption costs can be saved.

[0003] In the prior art, future electricity consumption data is predicted by means of manual prediction or model prediction to help users formulate reasonable electricity consumption strategies.

[0004] However, the results predicted manually or by a model are not accurate, and it is difficult to formulate accurate electricity consumption strategies based on the prediction results. Summary of the Invention

[0005] The present disclosure provides a data prediction method, apparatus, device, and storage medium, which can determine accurate predicted perception data, and determine accurate electricity consumption strategies based on the accurate predicted perception data, thereby saving electricity consumption costs.

[0006] According to a first aspect of the present disclosure, there is provided a data prediction method, the method including: obtaining historical perception data of a sensor, auxiliary data, and historical time data corresponding to the historical perception data, where the auxiliary data includes historical auxiliary data within a historical time period corresponding to the historical time data and future auxiliary data within a future target time period, and the auxiliary data is predictable data that can affect the perception data of the sensor; respectively performing matrix mapping on the historical perception data, historical time data, historical auxiliary data, and future auxiliary data to obtain a historical perception matrix corresponding to the historical perception data, a historical time matrix corresponding to the historical time data, a historical auxiliary matrix corresponding to the historical auxiliary data, and a future auxiliary matrix corresponding to the future auxiliary data; generating a first fusion matrix according to the historical perception matrix, historical time matrix, and historical auxiliary matrix; generating a second fusion matrix according to the first fusion matrix and the future auxiliary matrix; and mapping the second fusion matrix to the target time period to obtain predicted perception data of the sensor within the target time period.

[0007] According to a second aspect of the present disclosure, a data prediction device is provided, the device includes: an acquisition unit, a mapping unit, and a processing unit.

[0008] The acquisition unit is configured to acquire historical perception data of a sensor, auxiliary data, and historical time data corresponding to the historical perception data, where the auxiliary data includes historical auxiliary data within a historical time period corresponding to the historical time data and future auxiliary data within a future target time period, and the auxiliary data is predictable data that can affect the perception data of the sensor.

[0009] The mapping unit is configured to perform matrix mapping on the historical perception data, historical time data, historical auxiliary data, and future auxiliary data respectively to obtain a historical perception matrix corresponding to the historical perception data, a historical time matrix corresponding to the historical time data, a historical auxiliary matrix corresponding to the historical auxiliary data, and a future auxiliary matrix corresponding to the future auxiliary data.

[0010] The processing unit is configured to generate a first fusion matrix according to the historical perception matrix, historical time matrix, and historical auxiliary matrix; generate a second fusion matrix according to the first fusion matrix and the future auxiliary matrix; and map the second fusion matrix to the target time period to obtain predicted perception data of the sensor within the target time period.

[0011] According to a third aspect of the present disclosure, an electronic device is provided, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as in the first aspect.

[0012] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, and the computer instructions are used to cause a computer to execute the method according to the first aspect.

[0013] According to a fifth aspect of the present disclosure, a computer program product is provided, including a computer program, and the computer program implements the method according to the first aspect when executed by a processor.

[0014] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become easily understandable through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings are used to better understand the solution and do not constitute a limitation to the present disclosure. Among them:

[0016] Figure 1 is a schematic flowchart of the data prediction method provided by the embodiment of the present disclosure;

[0017] Figure 2 Another flowchart of the data prediction method provided by the embodiments of the present disclosure;

[0018] Figure 3 Another flowchart of the data prediction method provided by the embodiments of the present disclosure.

[0019] Figure 4 Another flowchart of the data prediction method provided by the embodiments of the present disclosure;

[0020] Figure 5 A schematic diagram of the composition of the data prediction device provided by the embodiments of the present disclosure;

[0021] Figure 6 A schematic block diagram of an exemplary electronic device 600 that can be used to implement the embodiments of the present disclosure provided by the embodiments of the present disclosure. Detailed implementation manners

[0022] The following describes exemplary embodiments of the present disclosure with reference to the accompanying drawings. Various details of the embodiments of the present disclosure are included to facilitate understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted below.

[0023] It should be understood that in the embodiments of the present disclosure, the character " / " generally represents an "or" relationship between the associated objects before and after. Terms such as "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features.

[0024] In actual production and life, thermal inertia is a common situation in the field of HVAC control. Thermal inertia refers to a means of predicting the temperature in an office building for a future period of time using the existing temperature data in the building. Especially in building HVAC, there is a very important need in determining thermal inertia. If the heat change in a building in the future can be predicted based on the existing heat change situation in the building (such as indoor temperature), and corresponding electricity consumption strategies can be formulated according to the prediction results, electricity costs can be saved.

[0025] Exemplarily, the application of thermal inertia in building HVAC can be used in temperature prediction. For example, the existing temperature data and historical information in a building can be used to predict the temperature change in the future period of time, and the operation of the HVAC equipment can be adjusted according to the temperature change, thereby improving energy efficiency and saving electricity costs.

[0026] For example, taking the application of thermal inertia in an office building as an example, when predicting the temperature data of the office building, the historical temperature data of the office building will be collected first, and the historical temperature data will be processed and analyzed to predict the temperature change of the office building in the future time period. Furthermore, the electricity consumption of the office building can be reasonably planned according to the predicted data to help the management personnel better manage the electricity consumption of the office building.

[0027] In the prior art, the future electricity consumption data is predicted by means of manual prediction or model prediction to help users formulate reasonable electricity consumption strategies.

[0028] However, the results predicted manually or by models are not accurate, and it is difficult to formulate accurate electricity consumption strategies based on the prediction results.

[0029] Under this background art, the present disclosure provides a data prediction method, which can determine accurate predicted perception data, and determine accurate electricity consumption strategies according to the accurate predicted perception data, thereby saving electricity costs.

[0030] Exemplarily, the execution subject of this data prediction method can be a computer or a server, or can also be other devices with data processing capabilities. There is no limitation on the execution subject of this method here.

[0031] In some embodiments, the server can be a single server, or can also be a server cluster composed of multiple servers. In some implementation manners, the server cluster can also be a distributed cluster. The present disclosure also has no limitation on the specific implementation manner of the server.

[0032] Figure 1 It is a schematic flowchart of the data prediction method provided by the embodiments of the present disclosure. As Figure 1 shown, this method can include step S101-step S105.

[0033] S101. Obtain the historical perception data of the sensor, the auxiliary data, and the historical time data corresponding to the historical perception data. The auxiliary data includes the historical auxiliary data within the historical time period corresponding to the historical time data and the future auxiliary data within the future target time period. The auxiliary data is predictable data that can affect the perception data of the sensor.

[0034] Exemplarily, the historical perception data, auxiliary data of the sensor, and the historical time data corresponding to the historical perception data can be obtained from a database storing the data, or can also be obtained from a device or server storing the data. Here, the storage location and acquisition method of the historical perception data, auxiliary data of the sensor, and the historical time data corresponding to the historical perception data are not specifically limited. Among them, the future auxiliary data in the auxiliary data can also be obtained and updated in real time from an online server or an online database. The type of the sensor can be a temperature sensor, a time sensor, a device status sensor, etc., and the number of sensors can be at least one. Here, the type and number of sensors are not specifically limited either. The historical perception data, auxiliary data of the sensor, and the historical time data corresponding to the historical perception data can be obtained according to the actual situation. The time length of the historical time period and the time length of the future target time period can be set according to the actual situation and are not specifically limited here. The auxiliary data is predictable data that can affect the perception data of the sensor. For example, the change of future weather can affect the perception data of the sensor, and the weather data is predictable data that can be obtained in the weather forecast.

[0035] For example, taking the location of the sensor in a shopping mall as an example, there can be sensors in two areas in the shopping mall. There can be two sensors in Area 1, namely Temperature Sensor 1 and Chiller Frequency Sensor 1. The historical perception data collected by Temperature Sensor 1 and Chiller Frequency Sensor 1 at historical moment 1 (i.e., historical time data 1) can be 23 and 77 respectively. There can be two sensors in Area 2 (i.e., location data 2), namely Temperature Sensor 2 and Chiller Frequency Sensor 2. The historical perception data collected by Temperature Sensor 2 and Chiller Frequency Sensor 2 at historical moment 1 (i.e., historical time data 1) can be 12 and 12 respectively. The historical auxiliary data of Area 1 and Area 2 at historical moment 1 can both be 27, indicating that the outdoor temperature of Area 1 and Area 2 at historical moment 1 is both 27. The future auxiliary data of Area 1 and Area 2 at future moment 2 within the future target time period can both be 26, indicating that the outdoor temperature of Area 1 and Area 2 at future moment 2 is both 26.

[0036] S102. Perform matrix mapping on the historical perception data, historical time data, historical auxiliary data, and future auxiliary data respectively to obtain a historical perception matrix corresponding to the historical perception data, a historical time matrix corresponding to the historical time data, a historical auxiliary matrix corresponding to the historical auxiliary data, and a future auxiliary matrix corresponding to the future auxiliary data.

[0037] Exemplarily, matrix mapping can be performed on historical perception data, historical time data, historical auxiliary data, and future auxiliary data respectively. Specifically, it can be multiplying the historical perception data by the matrix corresponding to the matrix mapping of the historical perception data, multiplying the historical time data by the matrix corresponding to the matrix mapping of the historical time data, multiplying the historical auxiliary data by the matrix corresponding to the matrix mapping of the historical auxiliary data, and multiplying the future auxiliary data by the matrix corresponding to the matrix mapping of the future auxiliary data, to obtain the historical perception matrix corresponding to the historical perception data, the historical time matrix corresponding to the historical time data, the historical auxiliary matrix corresponding to the historical auxiliary data, and the future auxiliary matrix corresponding to the future auxiliary data respectively. Among them, the matrices corresponding to the matrix mapping of the historical perception data, the matrix mapping of the historical time data, the matrix mapping of the historical auxiliary data, and the matrix mapping of the future auxiliary data can be preset, or can be updated according to actual needs, and no specific restrictions are made here.

[0038] Optionally, matrix mapping on historical perception data, historical time data, historical auxiliary data, and future auxiliary data respectively can also be performed using a mapping network. The parameters of the mapping network can be set according to the actual situation, and through the mapping network, the historical perception matrix corresponding to the historical perception data, the historical time matrix corresponding to the historical time data, the historical auxiliary matrix corresponding to the historical auxiliary data, and the future auxiliary matrix corresponding to the future auxiliary data can be output.

[0039] For example, taking a specific scenario of predicting the temperature data in a shopping mall as an example, the data collected by temperature sensor 1, chiller frequency sensor 1, and cooling pump frequency sensor 1 in area 1 of the shopping mall at 12 o'clock are 23, 77, and 12 respectively. The corresponding historical auxiliary data for area 1 can be 27, and the corresponding future auxiliary data for area 1 can be 30 (the historical auxiliary data and future auxiliary data can be generated respectively from the historical return water temperature of 27° corresponding to area 1 at 12 o'clock and the weather temperature of 30° corresponding to area 1 at 12 o'clock one day later); the data collected by temperature sensor 2, chiller frequency sensor 2, and cooling pump frequency sensor 2 in area 2 at 13 o'clock are 26, 56, and 16 respectively. The corresponding historical auxiliary data for area 2 can be 26, and the corresponding future auxiliary data for area 1 can be 32 (the historical auxiliary data and future auxiliary data can be generated respectively from the historical return water temperature of 26° corresponding to area 1 at 12 o'clock and the weather temperature of 32° corresponding to area 1 at 12 o'clock one day later). Taking this data as an example, using the data collected by temperature sensor 1, chiller frequency sensor 1, and cooling pump frequency sensor 1 in area 1 at 12 o'clock as the first row data of the historical perception matrix, and using the data collected by temperature sensor 2, chiller frequency sensor 2, and cooling pump frequency sensor 2 in area 2 at 13 o'clock as the second row data of the historical perception matrix, the generated historical perception data can be: Taking the historical return water temperature of 27° corresponding to area 1 as the first historical auxiliary data in the first row and the historical return water temperature of 26° corresponding to area 2 as the second historical auxiliary data in the first row; the generated historical auxiliary data can be: [27 26]; taking 12 o'clock corresponding to area 1 as the first historical time data in the first row and 13 o'clock corresponding to area 2 as the second historical time data in the first row; the generated historical time data can be: [12 13]; taking the weather temperature of 30° corresponding to area 1 as the first historical time data in the first row and the weather temperature of 32° corresponding to area 2 as the second historical time data in the first row, the generated future auxiliary data can be: [30 32].

[0040] Based on this, the historical perception data can be: Performing matrix mapping on the historical perception data, the obtained historical perception matrix can be:

[0041] Also for example, taking the historical auxiliary data as an example, the historical auxiliary data can be: [27 26]. Performing matrix mapping on the historical auxiliary data, the obtained historical auxiliary matrix can be:

[0042] Also for example, taking the historical time data as an example, the historical time data can be: [12 13]. Performing matrix mapping on the historical time data, the obtained historical time matrix can be:

[0043] For another example, taking future auxiliary data and historical time data as an example, the future auxiliary data can be: [30 32]. After performing matrix mapping on the future auxiliary data, the obtained future auxiliary matrix can be:

[0044] S103. Generate a first fusion matrix according to the historical perception matrix, the historical time matrix, and the historical auxiliary matrix.

[0045] Exemplarily, the manner of generating the first fusion matrix according to the historical perception matrix, the historical time matrix, and the historical auxiliary matrix can be obtained by adding the historical perception matrix, the historical time matrix, and the historical auxiliary matrix. Alternatively, the historical perception matrix, the historical time matrix, and the historical auxiliary matrix can be processed according to a fusion network to generate the first fusion matrix.

[0046] For example, taking the manner of generating the first fusion matrix by adding the historical perception matrix, the historical time matrix, and the historical auxiliary matrix as an example, the historical perception matrix can be: The historical time matrix can be: The historical auxiliary matrix can be: By adding the historical perception matrix, the historical time matrix, and the historical auxiliary matrix, the obtained first fusion matrix can be:

[0047] Optionally, generating the first fusion matrix according to the historical perception matrix, the historical time matrix, and the historical auxiliary matrix can be by adding according to the above fusion method, or by using the fusion network method. Here, the method of fusing the historical perception matrix, the historical time matrix, and the historical auxiliary matrix is not limited, as long as it can achieve the fusion of the historical perception matrix, the historical time matrix, and the historical auxiliary matrix.

[0048] S104. Generate a second fusion matrix according to the first fusion matrix and the future auxiliary matrix.

[0049] Exemplarily, the manner of generating the second fusion matrix according to the first fusion matrix and the future auxiliary matrix can be obtained by adding the first fusion matrix and the future auxiliary matrix. Alternatively, the first fusion matrix and the future auxiliary matrix can be processed according to a fusion network to generate the first fusion matrix.

[0050] For example, the first fusion matrix can be: The future auxiliary matrix can be: Taking the addition of the first fusion matrix and the future auxiliary matrix as an example, the generated second fusion matrix can be:

[0051] Optionally, the first fusion matrix and the future auxiliary matrix generate the second fusion matrix, which can be added according to the above fusion method, or, using the fusion network method. The method of fusing the first fusion matrix and the future auxiliary matrix is not limited here, as long as the fusion of the first fusion matrix and the future auxiliary matrix can be achieved.

[0052] S105. Map the second fusion matrix to the target time period to obtain the predicted perception data of the sensor in the target time period.

[0053] Exemplarily, mapping the fusion matrix to the target time period to obtain the predicted perception data of the sensor in the target time period can be obtained by multiplying the fusion matrix by the matrix for mapping the fusion matrix. The matrix for mapping the fusion matrix can be preset or can be updated according to actual needs. The matrix for mapping the fusion matrix is not specifically limited here. The length of the target time period can be set according to the actual situation and is not specifically limited here.

[0054] Optionally, mapping the fusion matrix to the target time period can also be performed through a mapping network. The parameters of the mapping network can be set according to the actual situation. Through the mapping network, the predicted perception data of the sensor in the target time period can be obtained.

[0055] For example, the second fusion matrix can be: Mapping the fusion matrix to the target time period to obtain the predicted perception data of the sensor in the target time period can be:

[0056] The present disclosure obtains historical perception data, auxiliary data, and historical time data corresponding to the historical perception data of a sensor. The auxiliary data includes historical auxiliary data within a historical time period corresponding to the historical time data and future auxiliary data within a future target time period. Matrix mapping is respectively performed on the historical perception data, historical time data, historical auxiliary data, and future auxiliary data to obtain a historical perception matrix corresponding to the historical perception data, a historical time matrix corresponding to the historical time data, a historical auxiliary matrix corresponding to the historical auxiliary data, and a future auxiliary matrix corresponding to the future auxiliary data. The characteristics of the data can be reflected in the form of a matrix through mapping, and the characteristics of the data can be more conveniently reflected through matrix calculations; according to the historical perception matrix, historical time matrix, and historical auxiliary matrix, a first fusion matrix is generated. The first fusion matrix can be used to fuse the historical perception matrix, historical time matrix, and historical auxiliary matrix, making the first fusion matrix more accurate; according to the first fusion matrix and the future auxiliary matrix, a second fusion matrix is generated. The second fusion matrix can be used to fuse the characteristics of the first fusion matrix and the future auxiliary matrix to obtain an accurate second fusion matrix; the second fusion matrix is mapped to the target time period to obtain the predicted perception data of the sensor within the target time period, and accurate predicted perception data can be obtained. Furthermore, relevant strategies can be formulated based on the accurate predicted perception data to reduce resource waste.

[0057] Optionally, the data prediction method can be implemented based on a large model. A large model is a technology that uses a machine learning model with a large number of parameters for data prediction. It utilizes the complex structure and rich parameters of the large model to capture the patterns and correlation information of the historical perception data, auxiliary data, and historical time data corresponding to the historical perception data of the sensor, thereby achieving the ability to accurately predict the predicted perception data. First, the data prediction method based on the large model can use the historical perception data, auxiliary data, and historical time data corresponding to the historical perception data as training data to construct a data prediction model and adjust the parameters of the data prediction model. Through training on the training data, the data prediction model can learn the potential patterns and data characteristics in the training data. After training is completed, the data prediction model can be applied to the prediction of pre-perception data. By inputting new data samples, the data prediction model can analyze the data samples to predict the pre-perception data corresponding to the data samples. Furthermore, relevant strategies can be formulated based on the accurate predicted perception data to reduce resource waste.

[0058] In some embodiments, Figure 1 step S102 in further includes: respectively inputting the historical perception data, historical time data, historical auxiliary data, and future auxiliary data into a mapping network, and performing matrix mapping on the historical perception data, historical time data, historical auxiliary data, and future auxiliary data respectively through the mapping network.

[0059] Exemplarily, the mapping network can be used to perform matrix mapping on historical perception data, historical time data, historical auxiliary data, and future auxiliary data respectively, and output a historical perception matrix corresponding to the historical perception data, a historical time matrix corresponding to the historical time data, a historical auxiliary matrix corresponding to the historical auxiliary data, and a future auxiliary matrix corresponding to the future auxiliary data. The structure of the mapping network can be a feedforward neural network, or a recurrent neural network, or a convolutional neural network, etc. The structure of the mapping network is not specifically limited herein.

[0060] For example, as in the example of step S102, the data collected by the temperature sensor 1, the chiller frequency sensor 1, and the cooling pump frequency sensor 1 in area 1 at 12 o'clock is used as the first row of data of the historical perception matrix, and the data collected by the temperature sensor 2, the chiller frequency sensor 2, and the cooling pump frequency sensor 2 in area 2 at 13 o'clock is used as the second row of data of the historical perception matrix. The generated historical perception data can be: Taking the historical return water temperature corresponding to area 1 as 27° as the first historical auxiliary data in the first row and the historical return water temperature corresponding to area 2 as 26° as the second historical auxiliary data in the first row; the generated historical auxiliary data can be: [27 26]; taking 12 o'clock corresponding to area 1 as the first historical time data in the first row and 13 o'clock corresponding to area 2 as the second historical time data in the first row; the generated historical time data can be: [12 13]; taking the weather temperature corresponding to area 1 as 30° as the first historical time data in the first row and the weather temperature corresponding to area 2 as 32° as the second historical time data in the first row, the generated future auxiliary data can be: [30 32]. Through the mapping network, the historical perception matrix corresponding to the generated historical perception data can be: The historical time matrix corresponding to the historical time data can be: The historical auxiliary matrix corresponding to the historical auxiliary data can be: The future auxiliary matrix corresponding to the future auxiliary data can be:

[0061] In this embodiment, by performing matrix mapping on historical perception data, historical time data, historical auxiliary data, and future auxiliary data respectively through the mapping network, it can help to discover the correlation and mutual influence among historical perception data, historical time data, historical auxiliary data, and future auxiliary data, and more conveniently reflect the characteristics of the data through matrix calculation.

[0062] In some embodiments, Figure 1 step S103 further includes: inputting the historical perception matrix, the historical time matrix, and the historical auxiliary matrix into a first fusion network, and outputting a first fusion matrix through the first fusion network.

[0063] Exemplarily, the structure of the first fusion network can be a convolutional neural network or a recurrent neural network. There is no specific limitation on the form of the first fusion network here, as long as it can fuse the historical perception matrix, the historical time matrix, and the historical auxiliary matrix to output the first fusion matrix.

[0064] For example, the historical perception matrix corresponding to the historical perception data can be: The historical time matrix corresponding to the historical time data can be: The historical auxiliary matrix corresponding to the historical auxiliary data can be: Inputting the historical perception matrix, the historical time matrix, and the historical auxiliary matrix into the first fusion network, the first fusion matrix output by the first fusion network can be:

[0065] In this embodiment, by fusing the historical perception matrix, the historical time matrix, and the historical auxiliary matrix through the first fusion matrix to obtain the first fusion matrix, the accuracy of the first fusion matrix can be improved.

[0066] In some embodiments, Figure 1 step S104 in further includes: inputting the first fusion matrix and the future auxiliary matrix into a second fusion network, and outputting a second fusion matrix through the second fusion network.

[0067] Exemplarily, the structure of the second fusion network can be a convolutional neural network or a recurrent neural network. There is no specific limitation on the form of the second fusion network here, as long as it can fuse the first fusion matrix and the future auxiliary matrix to output the second fusion matrix.

[0068] For example, the first fusion matrix can be: The future auxiliary matrix can be: Inputting the first fusion matrix and the future auxiliary matrix into the second fusion network, the second fusion matrix output by the second fusion network can be:

[0069] In this embodiment, by fusing the first fusion matrix and the future auxiliary matrix through the second fusion network, the accuracy of the second fusion matrix can be improved.

[0070] In some embodiments, the time length of the target time period is less than or equal to the time length of the historical time period.

[0071] Exemplarily, the time length of the target time period can be less than or equal to the time length of the historical time period, which can make full use of the data within the historical time period to obtain accurate predicted perception data of the sensor within the target time period.

[0072] For example, taking the time length of the target time period as one month, the time length of the historical time period can be one month or two months. The specific time length of the historical time period can be set according to actual needs.

[0073] In this embodiment, the time length of the target time period is less than or equal to the time length of the historical time period, which can make full use of the data within the historical time period, improve the accuracy and reliability of the prediction of the predicted perception data, and can also improve the prediction efficiency and accelerate the acquisition of the predicted perception data.

[0074] In some embodiments, Figure 1 Before step S102, the data prediction method further includes: performing data cleaning on the historical perception data, historical time data, historical auxiliary data, and future auxiliary data.

[0075] Data cleaning may refer to filtering out outliers and filling in missing values for the historical perception data, historical time data, historical auxiliary data, and future auxiliary data. The specific methods and means of data cleaning are not specifically limited herein.

[0076] In this embodiment, by performing data cleaning on the historical perception data, historical time data, historical auxiliary data, and future auxiliary data, the accuracy and availability of the historical perception data, historical time data, historical auxiliary data, and future auxiliary data can be improved, thereby improving the accuracy and reliability of the prediction of the predicted perception data.

[0077] In some embodiments, the historical time data includes at least two time nodes. Figure 1 Before step S102, the data prediction method further includes: for the target time node, expanding the historical perception data of the target time node according to the historical perception data corresponding to the time node adjacent to the target time node.

[0078] Exemplarily, the historical perception data corresponding to the time node adjacent to the target time node can be obtained by adding the historical perception data corresponding to the time node adjacent to the target time node to the historical perception data of the target time node, or by taking the average value, or by expanding the dimension of the historical perception data of the target time node according to the historical perception data corresponding to the time node adjacent to the target time node. The specific method is not specifically limited herein.

[0079] Optionally, when the historical perception data of the time node adjacent to the target time node does not exist, a default expansion parameter can be set to expand the historical perception data of the target time node.

[0080] For example, taking a specific scenario of predicting mall temperature data as an example, the data collected by temperature sensor 1, chiller frequency sensor 1, and cooling pump frequency sensor 1 in area 1 of the mall at 12 o'clock are 23, 77, and 12 respectively, the data collected at 13 o'clock are 26, 56, and 16 respectively, and the data collected at 14 o'clock are 25, 70, and 15 respectively. Taking the way of expanding the historical perception data of the target time node as an example, where the historical perception data corresponding to the time node adjacent to the target time node is added to the historical perception data of the target time node, the historical perception data of the expanded target time node can be expanded to 75, 189, 44 at 12 o'clock, can be expanded to 74, 203, 53 at 13 o'clock, and can be expanded to 77, 182, 47 at 13 o'clock.

[0081] In this embodiment, for the target time node, according to the historical perception data corresponding to the time node adjacent to the target time node, the historical perception data of the target time node is expanded, so that the historical perception data of the target time node can maintain a strong correlation with the historical perception data of the adjacent time node. Furthermore, the elements in the generated historical perception matrix will have a certain connection with other data, improving the accuracy of the historical perception matrix.

[0082] In some embodiments, the time nodes adjacent to the target time node include the time nodes before the target time node, and / or the time nodes after the target time node.

[0083] Exemplarily, the time node adjacent to the target time node can be a node that only includes the time nodes before the target time node, can also be a node that only includes the time nodes after the target time node, or can also include both the time nodes before the target time node and the time nodes after the target time node.

[0084] For example, the data collected by temperature sensor 1, chiller frequency sensor 1, and cooling pump frequency sensor 1 in area 1 of the mall at 12 o'clock are 23, 77, and 12 respectively, the data collected at 13 o'clock are 26, 56, and 16 respectively, and the data collected at 14 o'clock are 25, 70, and 15 respectively. When the target time node is 13 o'clock, the time nodes adjacent to the target time node can be 12 o'clock, can also be 14 o'clock, or can also include both 12 o'clock and 14 o'clock.

[0085] In this embodiment, the time node adjacent to the target time node is a node that only includes time nodes before the target time node, or a node that only includes time nodes after the target time node, or includes time nodes before the target time node and also includes time nodes after the target time node, which can improve the diversity of historical perception data of the target time node, and at the same time make the historical perception data of the target time node and the adjacent time nodes maintain connectivity, improving the stability of the historical perception data of the target time node.

[0086] In some embodiments, there are N time nodes adjacent to the target time node, and the historical perception data corresponding to the target time node is an M-dimensional vector. The data prediction method further includes: expanding the M-dimensional vector into an M + N-dimensional vector.

[0087] Exemplarily, the number of time nodes adjacent to the target time node can be a number set according to actual needs. Specifically, the user can set it according to actual requirements, and no specific limitation is made here. It can be all time nodes adjacent to the target time node, or some time nodes among all time nodes adjacent to the target time node.

[0088] For example, the historical perception data collected by temperature sensor 1, chiller frequency sensor 1, and cooling pump frequency sensor 1 in area 1 of a shopping mall at 12 o'clock are 23, 77, and 12 respectively, at 13 o'clock are 26, 56, and 16 respectively, and at 14 o'clock are 25, 70, and 15 respectively. When N is 1 and the target time node is 13, the 1-dimensional vector 26 collected by temperature sensor 1 can be expanded into a 2-dimensional vector [23 26], the 1-dimensional vector 56 collected by chiller frequency sensor 1 can be expanded into a 2-dimensional vector [77 56], and expanding the M-dimensional vector into an M + N-dimensional vector is the expansion method as illustrated in the above example.

[0089] Optionally, the way of expanding the M-dimensional vector into an M + N-dimensional vector can also be based on an expansion model, without specific limitation, as long as it can achieve the above-mentioned expansion of the data of the target time node according to the historical perception data of the time nodes adjacent to the target time node.

[0090] In this embodiment, by expanding the M-dimensional vector into an M + N-dimensional vector, the representation ability of the data of the target time node can be increased to provide more information and features, and then more information and features can be used to support more complex analysis and prediction.

[0091] Figure 2 It is another process schematic diagram of the data prediction method provided by the embodiments of the present disclosure. As Figure 2 shown, in some embodiments, Figure 1Before step S102, the data prediction method further includes steps S201 - S202.

[0092] S201. Obtain the normalization parameters corresponding to the historical perception data.

[0093] S202. Normalize the historical perception data according to the normalization parameters.

[0094] Exemplarily, the normalization parameters corresponding to the historical perception data can be obtained based on the historical perception data (such as calculating the mean, or the maximum value, or the variance, etc. of the historical perception data), or the user can preset the normalization parameters corresponding to the historical perception data. Here, the acquisition method of the normalization parameters corresponding to the historical perception data is not specifically limited. Normalizing the historical perception data according to the normalization parameters can be a process of performing operations on the historical perception data according to the normalization parameters, so that the absolute value of the historical perception data is between 0 and 1 (including 0 or 1).

[0095] Optionally, the method of normalizing the historical perception data according to the normalization parameters can be implemented by performing operations on the historical perception data according to the normalization parameters, or a normalization model can be used to process the historical perception data according to the normalization parameters. Here, it is not specifically limited, and the corresponding normalization method can be selected according to actual needs.

[0096] Optionally, the number of normalization parameters can be at least one, and here the number of normalization parameters is not specifically limited.

[0097] For example, the historical perception data collected by temperature sensor 1, chiller frequency sensor 1, and cooling pump frequency sensor 1 in area 1 of a shopping mall at 13:00 are 26, 56, and 16 respectively, and the historical perception data collected at 14:00 are 25, 70, and 15 respectively. The normalization parameters corresponding to the obtained historical perception data can be 70 (taking the maximum value of the historical perception data). After normalizing the historical perception data according to the normalization parameters (calculating the ratio of the historical perception data and the normalization parameters), the normalized historical perception data obtained can be: the historical perception data collected by temperature sensor 1, chiller frequency sensor 1, and cooling pump frequency sensor 1 in area 1 of the shopping mall at 13:00 are 0.37, 0.8, and 0.23 respectively, and the historical perception data collected at 14:00 are 0.36, 1, and 0.21 respectively.

[0098] In this embodiment, by obtaining the normalization parameters corresponding to the historical perception data and normalizing the historical perception data according to the normalization parameters, the processing of the historical perception data can be accelerated, the range consistency of the historical perception data can be maintained, the distribution of the historical perception data can be quickly obtained, and the historical perception data can be better explained.

[0099] Figure 3 Another flowchart diagram of the data prediction method provided by the embodiments of the present disclosure. As Figure 3 shown, in some embodiments, Figure 2 step S201 in Figure 2 may further include step S301-step S302,

[0100] S301. Determine the first normalization parameter according to the mean value of the historical perception data corresponding to the historical time data.

[0101] S302. Determine the second normalization parameter according to the variance of the historical perception data corresponding to the historical time data.

[0102] Exemplarily, the number of normalization parameters may be two, including the first normalization parameter and the second normalization parameter.

[0103] For example, the historical perception data collected by temperature sensor 1, chiller frequency sensor 1, and cooling pump frequency sensor 1 in area 1 of a shopping mall at 13:00 are 26, 56, and 16 respectively, and the historical perception data collected at 14:00 are 25, 70, and 15 respectively. According to the mean value of the historical perception data corresponding to the historical time data, the first normalization parameter can be 34.7 (the mean value of the historical perception data corresponding to the historical time data), and the determined second normalization parameter can be 11245.35 (the variance of the historical perception data corresponding to the historical time data).

[0104] S303. Determine the intermediate historical perception data according to the difference between the historical perception data corresponding to the historical time data and the first normalization parameter.

[0105] S304. Determine the normalized historical perception data according to the ratio of the intermediate historical perception data corresponding to the historical time data and the second normalization parameter.

[0106] For example, according to the difference between the historical perception data corresponding to the historical time data and the first normalization parameter, the intermediate historical perception data can be respectively: -8.67, 21.33, -18.67, -9.67, 35.33, -19.67. According to the ratio of the intermediate historical perception data corresponding to the historical time data and the second normalization parameter, the normalized historical perception data can be respectively: 0.091, 0.221, 0.193, 0.100, 0.293, 0.197.

[0107] In this embodiment, the first normalization parameter is determined according to the mean of the historical perception data corresponding to the historical time data, the second normalization parameter is determined according to the variance of the historical perception data corresponding to the historical time data, and the intermediate historical perception data is determined according to the difference between the historical perception data corresponding to the historical time data and the first normalization parameter, which can eliminate the offset of the historical perception data and retain the distribution characteristics of the historical perception data; the normalized historical perception data is determined according to the ratio of the intermediate historical perception data corresponding to the historical time data to the second normalization parameter, which can eliminate the scale difference of the historical perception data and reduce the influence caused by the scale difference between different historical perception data, ensuring the comparability between different historical perception data.

[0108] In some embodiments, Figure 1 Before step S104 of the data prediction method, the method further includes: performing denormalization on the second fusion matrix according to the normalization parameter.

[0109] Exemplarily, as the normalization parameter described in step S302, the number of normalization parameters can be at least one. Performing denormalization on the second fusion matrix according to the normalization parameter can be achieved by performing operations on the elements in the second fusion matrix according to the normalization parameter.

[0110] Optionally, the manner of performing denormalization on the second fusion matrix according to the normalization parameter can be achieved by performing operations on the second fusion matrix according to the normalization parameter, or a denormalization model can be used to perform processing on the historical perception data according to the normalization parameter. There is no specific limitation here, and the corresponding denormalization manner can be selected according to actual needs.

[0111] For example, the second fusion matrix can be: The normalization parameters can include two, which are 34.7 and 11245.35 respectively. After performing denormalization on the second fusion matrix according to the normalization parameters, the denormalized second fusion matrix obtained can be:

[0112] In this embodiment, by performing denormalization on the second fusion matrix according to the normalization parameter, the values of the elements in the second fusion matrix can be restored to the same scale and dimension as the historical perception data, improving the interpretability, applicability, and displayability of the elements in the second fusion matrix.

[0113] Figure 4 Another flowchart of the data prediction method provided by the embodiments of the present disclosure. As Figure 4 shown, in some embodiments, the data prediction method further includes step S401 - step S402.

[0114] S401. Determine an intermediate second fusion matrix based on the product of the elements in the second fusion matrix and the second normalization parameter.

[0115] S402. Determine the denormalized second fusion matrix based on the sum of the elements in the intermediate second fusion matrix and the first normalization parameter.

[0116] For example, the second fusion matrix can be: The normalization parameters can include two, which are 34.7 and 11245.35 respectively. The intermediate second fusion matrix determined based on the product of the elements in the second fusion matrix and the second normalization parameter can be: The denormalized second fusion matrix determined based on the sum of the elements in the intermediate second fusion matrix and the first normalization parameter can be:

[0117] In this embodiment, determining the intermediate second fusion matrix based on the product of the elements in the second fusion matrix and the second normalization parameter can eliminate the deviation in the element distribution of the second fusion matrix after normalization, making the element distribution in the second fusion matrix closer to the historical perception data; determining the denormalized second fusion matrix based on the sum of the elements in the intermediate second fusion matrix and the first normalization parameter can make the element distribution in the denormalized second fusion matrix have the same dimension as the historical perception data and retain the statistical characteristics of the elements in the second fusion matrix.

[0118] In some embodiments, Figure 1 before step S104, the data prediction method further includes: updating the first fusion matrix through a residual network.

[0119] Exemplarily, a residual network can be used to process the first fusion matrix to obtain the processed first fusion matrix, and the processed first fusion matrix can be superimposed and fused with the first fusion matrix input into the residual network to achieve the update of the first fusion matrix.

[0120] Optionally, the residual network can process the first fusion matrix N times, where N is a positive integer greater than or equal to 1. And the residual network can superimpose the matrix obtained by processing the first fusion matrix for the Nth time on the matrix input into the residual network for the Nth time to update the first fusion matrix.

[0121] Optionally, after updating the first fusion matrix, a second fusion matrix can be generated based on the updated first fusion matrix and the future auxiliary matrix.

[0122] For example, the first fusion matrix can be: The matrix obtained by processing the first fusion matrix once using a residual network can be: And the matrix obtained by processing the first fusion matrix once is superimposed on the first fusion matrix (that is, the two matrices are added), and the updated first fusion matrix obtained can be:

[0123] In this embodiment, the first fusion matrix is updated through a residual network, which can superimpose new feature information on the first fusion matrix while retaining the information of the original first fusion matrix, ensuring both the original information of the first fusion matrix and adding new features to the first fusion matrix, improving the accuracy and reliability of the elements in the first fusion matrix.

[0124] In some embodiments, the types of sensors include at least one of the following: temperature sensors, humidity sensors, flow sensors, and power sensors.

[0125] Exemplarily, taking the temperature sensor as an example, the temperature sensor can be an indoor temperature sensor, a water temperature sensor, etc. The historical perception data collected is the measured temperature or temperature change data. The indoor temperature sensor measures the indoor temperature as 25°C; taking the humidity sensor as an example, the humidity sensor can be an indoor humidity sensor, a soil humidity sensor, etc. The historical perception data collected is the data of the moisture content in the air measured. The indoor humidity sensor shows that the indoor relative humidity is 50%; taking the flow sensor as an example, the flow sensor can be a water flow sensor, an air flow sensor, etc. The historical perception data collected can be the data of the flow rate of liquid or gas measured. The water flow sensor shows that the water flow rate is 10 liters per minute; taking the power sensor as an example, the power sensor can be an indoor power sensor, a total power sensor, etc. The power data collected by the indoor power sensor can be 1035°.

[0126] In this embodiment, the types of sensors include at least one of the following: temperature sensors, humidity sensors, flow sensors, and power sensors, which can include historical perception data collected by multiple sensors, improving the diversity of historical perception data.

[0127] In some embodiments, the historical auxiliary data includes at least one of the following types: the historical return water temperature corresponding to the time node in the historical time data, the number of people corresponding to the time node in the historical time data, the outlet water temperature corresponding to the time node in the historical time data, the sensor power corresponding to the time node in the historical time data; the future auxiliary data includes at least one of the following types: the weather data corresponding to the time node in the target time period, the sensor status corresponding to the time node in the target time period, the indoor temperature corresponding to the time node in the target time period, the outdoor temperature corresponding to the time node in the target time period, the wind speed corresponding to the time node in the target time period, the irradiance corresponding to the time node in the target time period.

[0128] Exemplarily, taking the historical return water temperature corresponding to the time node in the historical time data as the historical auxiliary data, the historical return water temperature corresponding to the time node in the historical time data can affect the value of the historical perception data. For example, the historical return water temperature collected at 13 o'clock in the historical time data can be 56 degrees Celsius. Taking the number of people corresponding to the time node in the historical time data as the historical auxiliary data, the number of people corresponding to the time node in the historical time data can affect the value of the historical perception data. For example, the net number of people collected at 13 o'clock in the historical time data can be 522.

[0129] For another example, taking the weather data corresponding to the time node in the target time period as the future auxiliary data, the weather data corresponding to the time node in the target time period can affect the value in the historical perception data. For example, the weather data corresponding to 13 o'clock in the future is 30 degrees Celsius; taking the sensor state corresponding to the time node in the target time period as the future auxiliary data, for example, the switch state of the air conditioner corresponding to 13 o'clock in the future is on, and the identification information of the corresponding switch state can be 1.

[0130] The historical auxiliary data in this embodiment includes at least one of the following types: the historical return water temperature corresponding to the time node in the historical time data, the number of people corresponding to the time node in the historical time data, the outlet water temperature corresponding to the time node in the historical time data, the sensor power corresponding to the time node in the historical time data; the future auxiliary data includes at least one of the following types: the weather data corresponding to the time node in the target time period, the sensor state corresponding to the time node in the target time period, the indoor temperature corresponding to the time node in the target time period, the outdoor temperature corresponding to the time node in the target time period, the wind speed corresponding to the time node in the target time period, the irradiance corresponding to the time node in the target time period, which can provide auxiliary data in terms of history and future, help users perform data analysis and decision-making, and thus enhance the reference and support in the planning process and improve the reliability and effectiveness of decision-making.

[0131] In an exemplary embodiment, the present disclosure also provides a data prediction device, which can be used to implement the data prediction method in the foregoing embodiment. Figure 5 It is a schematic diagram of the composition of the data prediction device provided by the embodiments of the present disclosure. As Figure 5 shown, the device may include: an acquisition unit 501, a mapping unit 502, and a processing unit 503.

[0132] The acquisition unit 501 is configured to acquire the historical perception data of the sensor, the auxiliary data, and the historical time data corresponding to the historical perception data. The auxiliary data includes the historical auxiliary data in the historical time period corresponding to the historical time data and the future auxiliary data in the future target time period. The auxiliary data is predictable data that can affect the perception data of the sensor.

[0133] A mapping unit 502 is configured to perform matrix mapping on historical perception data, historical time data, historical auxiliary data, and future auxiliary data respectively, to obtain a historical perception matrix corresponding to the historical perception data, a historical time matrix corresponding to the historical time data, a historical auxiliary matrix corresponding to the historical auxiliary data, and a future auxiliary matrix corresponding to the future auxiliary data.

[0134] A processing unit 503 is configured to generate a first fusion matrix according to the historical perception matrix, the historical time matrix, and the historical auxiliary matrix; generate a second fusion matrix according to the first fusion matrix and the future auxiliary matrix; and map the second fusion matrix within a target time period to obtain predicted perception data of the sensor within the target time period.

[0135] Optionally, the mapping unit 502 is specifically configured to: input the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data into a mapping network respectively, and perform matrix mapping on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data respectively through the mapping network.

[0136] Optionally, the processing unit 503 is specifically configured to: input the historical perception matrix, the historical time matrix, and the historical auxiliary matrix into a first fusion network, and output a first fusion matrix through the first fusion network.

[0137] Optionally, the processing unit 503 is specifically configured to: input the first fusion matrix and the future auxiliary matrix into a second fusion network, and output a second fusion matrix through the second fusion network.

[0138] Optionally, the time length of the target time period is less than or equal to the time length of the historical time period.

[0139] Optionally, before performing matrix mapping on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data respectively, the acquisition unit 501 is further configured to: perform data cleaning on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data.

[0140] Optionally, the historical time data includes at least two time nodes. Before performing matrix mapping on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data respectively, the acquisition unit 501 is further configured to: for a target time node, expand the historical perception data of the target time node according to the historical perception data corresponding to the time node adjacent to the target time node.

[0141] Optionally, the time nodes adjacent to the target time node include the time nodes before the target time node, and / or, the time nodes after the target time node.

[0142] Optionally, there are N time nodes adjacent to the target time node, and the historical perception data corresponding to the target time node is an M-dimensional vector. The obtaining unit 501 is specifically configured to: expand the M-dimensional vector into an M+N-dimensional vector.

[0143] Optionally, before performing matrix mapping on the historical perception data, historical time data, historical auxiliary data, and future auxiliary data respectively, the obtaining unit 501 is further configured to: obtain the normalization parameter corresponding to the historical perception data; normalize the historical perception data according to the normalization parameter.

[0144] Optionally, the obtaining unit 501 is specifically configured to: determine the first normalization parameter according to the mean value of the historical perception data corresponding to the historical time data; determine the second normalization parameter according to the variance of the historical perception data corresponding to the historical time data; determine the intermediate historical perception data according to the difference between the historical perception data corresponding to the historical time data and the first normalization parameter; determine the normalized historical perception data according to the ratio of the intermediate historical perception data corresponding to the historical time data and the second normalization parameter.

[0145] Optionally, before mapping the second fusion matrix to the target time period, the mapping unit 502 is further configured to: denormalize the second fusion matrix according to the normalization parameter.

[0146] Optionally, the mapping unit 502 is specifically configured to: determine the intermediate second fusion matrix according to the product of the elements in the second fusion matrix and the second normalization parameter; determine the denormalized second fusion matrix according to the sum of the elements in the intermediate second fusion matrix and the first normalization parameter.

[0147] Optionally, before generating the second fusion matrix according to the first fusion matrix and the future auxiliary matrix, the mapping unit 502 is further configured to: update the first fusion matrix through a residual network.

[0148] Optionally, the types of the sensors include at least one of the following: temperature sensor, humidity sensor, flow sensor, power sensor.

[0149] Optionally, the historical auxiliary data includes at least one of the following types: historical return water temperature corresponding to the time node in the historical time data, human flow corresponding to the time node in the historical time data, outlet water temperature corresponding to the time node in the historical time data, sensor power corresponding to the time node in the historical time data; the future auxiliary data includes at least one of the following types: weather data corresponding to the time node in the target time period, sensor status corresponding to the time node in the target time period, indoor temperature corresponding to the time node in the target time period, outdoor temperature corresponding to the time node in the target time period, wind speed corresponding to the time node in the target time period, irradiance corresponding to the time node in the target time period.

[0150] In the technical solutions of the present disclosure, the acquisition, storage, and application of the user's personal information involved all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0151] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0152] In an exemplary embodiment, the electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method as described in the above embodiments.

[0153] In an exemplary embodiment, the readable storage medium may be a non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the method as described in the above embodiments.

[0154] In an exemplary embodiment, the computer program product includes a computer program that, when executed by a processor, implements the method as described in the above embodiments.

[0155] Figure 6 FIG. shows a schematic block diagram of an exemplary electronic device 600 that can be used to implement the embodiments of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as, for example, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as, for example, a personal digital processor, a cellular phone, a smart phone, a wearable device, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely exemplary and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0156] As Figure 6 shown, the electronic device 600 includes a computing unit 601 that can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 can also be stored. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0157] Multiple components in the electronic device 600 are connected to the I / O interface 605, including: an input unit 606, such as a keyboard, a mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a magnetic disk, an optical disc, etc.; and a communication unit 609, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 609 allows the electronic device 600 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0158] The computing unit 601 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 executes the various methods and processes described above, such as the data prediction method. For example, in some embodiments, the data prediction method can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as the storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the data prediction method described above can be executed. Alternatively, in other embodiments, the computing unit 601 can be configured to execute the data prediction method by any other suitable means (e.g., by means of firmware).

[0159] The various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs, which can be executed and / or interpreted on a programmable system including at least one programmable processor, the programmable processor can be a special or general programmable processor, can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0160] The program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that the program codes, when executed by the processor or controller, cause the functions / operations specified in the flowchart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0161] In the context of the present disclosure, a machine-readable medium may be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0162] In order to provide interaction with a user, the systems and techniques described herein may be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0163] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.

[0164] A computer system can include a client and a server. The client and the server are generally far from each other and typically interact through a communication network. The client-server relationship is generated by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, can also be a server of a distributed system, or a server incorporating a blockchain.

[0165] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in this disclosure can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and no limitations are imposed herein.

[0166] The above specific embodiments do not constitute a limitation on the protection scope of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements, and improvements made within the spirit and principles of this disclosure shall be included within the protection scope of this disclosure.

Claims

1. A data prediction method, the method comprises: obtaining historical perception data of a sensor, auxiliary data, and historical time data corresponding to the historical perception data, where the auxiliary data includes historical auxiliary data within a historical time period corresponding to the historical time data and future auxiliary data within a future target time period, and the auxiliary data is predictable data that can affect the perception data of the sensor; performing matrix mapping on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data respectively to obtain a historical perception matrix corresponding to the historical perception data, a historical time matrix corresponding to the historical time data, a historical auxiliary matrix corresponding to the historical auxiliary data, and a future auxiliary matrix corresponding to the future auxiliary data; generating a first fusion matrix according to the historical perception matrix, the historical time matrix, and the historical auxiliary matrix; generating a second fusion matrix according to the first fusion matrix and the future auxiliary matrix; mapping the second fusion matrix into the target time period to obtain predicted perception data of the sensor within the target time period.

2. The method according to claim 1, wherein the performing matrix mapping on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data respectively comprises: inputting the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data into a mapping network respectively, and performing matrix mapping on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data respectively through the mapping network.

3. The method according to claim 1, wherein the generating a first fusion matrix according to the historical perception matrix, the historical time matrix, and the historical auxiliary matrix comprises: inputting the historical perception matrix, the historical time matrix, and the historical auxiliary matrix into a first fusion network, and outputting the first fusion matrix through the first fusion network.

4. The method according to claim 1, wherein the generating a second fusion matrix according to the first fusion matrix and the future auxiliary matrix comprises: inputting the first fusion matrix and the future auxiliary matrix into a second fusion network, and outputting the second fusion matrix through the second fusion network.

5. The method according to any one of claims 1 - 4, wherein the time length of the target time period is less than or equal to the time length of the historical time period.

6. The method according to any one of claims 1 - 5, before the performing matrix mapping on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data respectively, the method further comprises: performing data cleaning on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data.

7. The method according to any one of claims 1-6, wherein the historical time data includes at least two time nodes. Before respectively performing matrix mapping on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data, the method further comprises: For a target time node, expanding the historical perception data of the target time node according to the historical perception data corresponding to the time node adjacent to the target time node.

8. The method according to claim 7, wherein the time node adjacent to the target time node includes the time node before the target time node, and / or the time node after the target time node.

9. The method according to claim 7 or 8, wherein there are N time nodes adjacent to the target time node, and the historical perception data corresponding to the target time node is an M-dimensional vector. Expanding the historical perception data of the target time node comprises: Expanding the M-dimensional vector into an M+N-dimensional vector.

10. The method according to any one of claims 1-9, wherein before respectively performing matrix mapping on the historical perception data, the historical time data, the historical auxiliary data, and the future auxiliary data, the method further comprises: Obtaining the normalization parameter corresponding to the historical perception data; Normalizing the historical perception data according to the normalization parameter.

11. The method according to claim 10, wherein obtaining the normalization parameter corresponding to the historical perception data comprises: Determining a first normalization parameter according to the mean value of the historical perception data corresponding to the historical time data; Determining a second normalization parameter according to the variance of the historical perception data corresponding to the historical time data; Normalizing the historical sensor data according to the normalization parameter, comprising: Determining intermediate historical perception data according to the difference between the historical perception data corresponding to the historical time data and the first normalization parameter; Determining the normalized historical perception data according to the ratio of the intermediate historical perception data corresponding to the historical time data and the second normalization parameter.

12. The method according to claim 10 or 11, before mapping the second fusion matrix into the target time period, the method further comprises: Denormalizing the second fusion matrix according to the normalization parameter.

13. The method according to claim 12, wherein denormalizing the second fusion matrix according to the normalization parameter comprises: Determining an intermediate second fusion matrix according to the product of the elements in the second fusion matrix and the second normalization parameter; Determining the denormalized second fusion matrix according to the sum of the elements in the intermediate second fusion matrix and the first normalization parameter.

14. The method according to any one of claims 1-13, before generating the second fusion matrix according to the first fusion matrix and the future auxiliary matrix, the method further comprises: Updating the first fusion matrix through a residual network.

15. The method according to any one of claims 1-14, wherein the types of the sensors include at least one of the following: temperature sensor, humidity sensor, flow sensor, and power sensor.

16. The method according to any one of claims 1-15, wherein the historical auxiliary data includes at least one of the following types: the historical return water temperature corresponding to the time node in the historical time data, the pedestrian flow corresponding to the time node in the historical time data, the outlet water temperature corresponding to the time node in the historical time data, and the sensor power corresponding to the time node in the historical time data; the future auxiliary data includes at least one of the following types: the weather data corresponding to the time node in the target time period, the sensor status corresponding to the time node in the target time period, the indoor temperature corresponding to the time node in the target time period, the outdoor temperature corresponding to the time node in the target time period, the wind speed corresponding to the time node in the target time period, and the irradiance corresponding to the time node in the target time period.

17. A data prediction device, the device comprising: an acquisition unit configured to acquire the historical sensed data of the sensor, the auxiliary data, and the historical time data corresponding to the historical sensed data, wherein the auxiliary data includes the historical auxiliary data within the historical time period corresponding to the historical time data and the future auxiliary data within the future target time period, and the auxiliary data is predictable data that can affect the sensed data of the sensor; a mapping unit configured to perform matrix mapping on the historical sensed data, the historical time data, the historical auxiliary data, and the future auxiliary data respectively to obtain a historical sensed matrix corresponding to the historical sensed data, a historical time matrix corresponding to the historical time data, a historical auxiliary matrix corresponding to the historical auxiliary data, and a future auxiliary matrix corresponding to the future auxiliary data; a processing unit configured to generate a first fusion matrix according to the historical sensed matrix, the historical time matrix, and the historical auxiliary matrix; generate a second fusion matrix according to the first fusion matrix and the future auxiliary matrix; map the second fusion matrix to the target time period to obtain the predicted sensed data of the sensor in the target time period.

18. An electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method according to any one of claims 1-16.

19. A non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the method according to any one of claims 1-16.

20. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-16.

Images