A hospital energy data evaluation method, device and storage medium

The hospital energy data evaluation method uses IoT sensors and neural networks to accurately collect and correct energy consumption data, addressing inefficiencies in medical building energy management and enhancing energy-saving capabilities.

CN118395146BActive Publication Date: 2024-12-03XIAMEN VANN INTELLIGENT
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410848138.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-06-27
Publication Date
2024-12-03
Estimated Expiration
2044-06-27

AI Technical Summary

Technical Problem

Medical buildings face challenges in accurately collecting and managing energy consumption data due to their complex and variable functions, leading to inefficiencies in energy conservation and emission reduction efforts.

Method used

A hospital energy data evaluation method using Internet of Things sensors to collect data, followed by a preliminary evaluation with Z-tests and graph neural networks to identify and correct abnormalities in energy consumption data for different systems, employing LSTM models for medical equipment and sanitary hot water, and graph neural networks for air conditioning and lighting systems.

Benefits of technology

Improves data quality and enables precise energy consumption control, enhancing energy-saving measures by accurately identifying and correcting anomalies in various hospital systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118395146B_ABST
    Figure CN118395146B_ABST
Patent Text Reader

Abstract

The present invention proposes a hospital energy data evaluation method, device and storage medium, the method comprising: an acquisition step S101, collecting energy data in a hospital within a period of time through an Internet of Things sensor, the energy data including air conditioning system energy consumption data, lighting system energy consumption data, medical equipment energy consumption data and sanitary hot water energy consumption data; a preliminary evaluation step S102, determining the presence of abnormal energy data through a Z test on the air conditioning system energy consumption data, lighting system energy consumption data, medical equipment energy consumption data and sanitary hot water energy consumption data; an evaluation step S103, based on the type of energy data with abnormalities, using a corresponding method for evaluation; a control step S104, performing energy control on the corresponding energy consumption system based on the corrected energy data. The present invention first performs a preliminary evaluation to determine the presence of abnormal energy data, and then uses different neural networks to accurately evaluate and correct according to different data types, so as to achieve the technical effect of energy saving.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of intelligent management of hospital energy consumption and artificial intelligence technology, and specifically to a hospital energy data evaluation method, device and storage medium. Background Art

[0002] Medical buildings are one of the most energy-intensive types of public buildings, with energy consumption levels 1.6 to 2 times that of general public buildings. The energy consumption of medical buildings mainly includes air conditioning, lighting, medical equipment, sanitary hot water, etc., of which the air conditioning system accounts for 40% to 60% of the total energy consumption. The energy consumption characteristics of medical buildings are uninterrupted operation, high quality requirements, and complex and variable functions, which brings huge challenges and potential to energy conservation and emission reduction.

[0003] To carry out accurate energy consumption control, it is necessary to collect accurate energy consumption data of each device in the hospital. If the data is inaccurate, accurate energy consumption control cannot be carried out. Therefore, how to obtain accurate energy consumption data of each system in the hospital is a technical challenge. Summary of the invention

[0004] The present invention proposes the following technical solutions to address one or more technical deficiencies in the above-mentioned prior art.

[0005] A hospital energy data evaluation method, the method comprising:

[0006] An acquisition step is to collect energy data in the hospital over a period of time through an Internet of Things sensor, wherein the energy data includes energy consumption data of an air conditioning system, energy consumption data of a lighting system, energy consumption data of medical equipment, and energy consumption data of sanitary hot water;

[0007] A preliminary evaluation step, determining abnormal energy data by performing a Z test on the air conditioning system energy consumption data, the lighting system energy consumption data, the medical equipment energy consumption data, and the sanitary hot water energy consumption data;

[0008] The evaluation steps are based on the type of abnormal energy data and use corresponding methods to evaluate, including:

[0009] If the energy data with abnormalities is air-conditioning system energy consumption data, construct a first graph neural network to evaluate the collected air-conditioning system energy consumption data, determine that there are abnormal air-conditioning energy consumption data, and correct the abnormal air-conditioning system energy consumption data based on the evaluation result of the first graph neural network;

[0010] If the energy data with abnormalities is lighting system energy consumption data, construct a second graph neural network to evaluate the collected lighting system energy consumption data, determine that there are abnormal lighting system energy consumption data, and correct the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network;

[0011] If the energy data with abnormalities is medical equipment energy consumption data, construct a third neural network to evaluate the collected medical equipment energy consumption data, determine that there are abnormal medical equipment energy consumption data, and correct the abnormal medical equipment energy consumption data;

[0012] If the energy data with abnormalities is sanitary hot water energy consumption data, construct a fourth neural network to evaluate the collected sanitary hot water energy consumption data, determine that there are abnormal sanitary hot water energy consumption data, and correct the abnormal sanitary hot water energy consumption data;

[0013] The control step is to perform energy control on the corresponding energy consumption system based on the corrected energy data.

[0014] Furthermore, the third neural network and the fourth neural network are obtained by training the LSTM model using historical medical equipment energy consumption data and sanitary hot water energy consumption data respectively.

[0015] Furthermore, the first graph neural network is constructed in such a way that the M1 air outlets in the air-conditioning system are used as nodes to form a graph, and if two air outlets are in the same closed area, there is an edge between the two air outlets.

[0016] The second graph neural network is constructed in such a way that: a graph is constructed with N1 light sources in the lighting system as nodes, and if two light sources are connected to the same circuit, there is an edge between the two light sources;

[0017] Among them, M1≥3, N1≥3.

[0018] Furthermore, the characteristic value of the i-th node among the M1 nodes in the first graph neural network is for:

[0019] ;

[0020] in, represents the outlet air temperature of the ith air outlet, represents the set temperature of the i-th air outlet, represents the temperature of the liquid medium in the pipe connected to the i-th air outlet;

[0021] The feature value of the jth node among the N1 nodes in the second graph neural network for:

[0022] ;

[0023] in, represents the brightness of the jth light source,

[0024] Among them, 1≤i≤M1, 1≤j≤N1.

[0025] Furthermore, the weights of the edges in the graph in the first graph neural network are:

[0026] ;

[0027] The weights of the edges in the graph in the second graph neural network are:

[0028] .

[0029] Furthermore, the operation of correcting the abnormal air-conditioning system energy consumption data based on the evaluation result of the first graph neural network is:

[0030] The evaluation result of the first graph neural network is the data anomalies of M2 air-conditioning outlets and the corresponding anomaly probabilities. If the anomaly probability is greater than the first threshold, the energy consumption data of other normal air-conditioning outlets in the same closed space as the air-conditioning outlet are averaged as the energy consumption data of the air-conditioning outlet. If the anomaly probability is less than or equal to the first threshold, the energy consumption data of the air-conditioning outlet is corrected to :

[0031] ;

[0032] in, represents the data anomaly probability of the qth air outlet output by the first graph neural network, Indicates the energy consumption data collected for the qth air outlet, It represents the average value of energy consumption data of other normal air-conditioning outlets in the same closed space as the qth air outlet;

[0033] Furthermore, the operation of correcting the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network is as follows: the evaluation result of the second graph neural network is the data abnormality of N2 light sources and the corresponding abnormal probability. If the abnormal probability is greater than the second threshold, the data of the abnormal light source is directly discarded, and a notice is issued that the light source needs maintenance. If the abnormal probability is less than or equal to the second threshold, the energy consumption data of other normal light sources with the same power as the light source are averaged as the energy consumption data of the light source. Wherein, M1>M2, N1>N2.

[0034] The present invention also proposes a hospital energy data evaluation device, which includes:

[0035] An acquisition unit collects energy data in the hospital over a period of time through an Internet of Things sensor, wherein the energy data includes energy consumption data of an air conditioning system, energy consumption data of a lighting system, energy consumption data of medical equipment, and energy consumption data of sanitary hot water;

[0036] A preliminary evaluation unit, for determining abnormal energy data by performing a Z test on the energy consumption data of the air conditioning system, the energy consumption data of the lighting system, the energy consumption data of the medical equipment, and the energy consumption data of the sanitary hot water;

[0037] The evaluation unit uses a corresponding method to evaluate the type of abnormal energy data, including:

[0038] If the energy data with abnormalities is air-conditioning system energy consumption data, construct a first graph neural network to evaluate the collected air-conditioning system energy consumption data, determine that there are abnormal air-conditioning energy consumption data, and correct the abnormal air-conditioning system energy consumption data based on the evaluation result of the first graph neural network;

[0039] If the energy data with abnormalities is lighting system energy consumption data, construct a second graph neural network to evaluate the collected lighting system energy consumption data, determine that there are abnormal lighting system energy consumption data, and correct the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network;

[0040] If the energy data with abnormalities is medical equipment energy consumption data, construct a third neural network to evaluate the collected medical equipment energy consumption data, determine that there are abnormal medical equipment energy consumption data, and correct the abnormal medical equipment energy consumption data;

[0041] If the energy data with abnormalities is sanitary hot water energy consumption data, construct a fourth neural network to evaluate the collected sanitary hot water energy consumption data, determine that there are abnormal sanitary hot water energy consumption data, and correct the abnormal sanitary hot water energy consumption data;

[0042] The control unit performs energy control on the corresponding energy consumption system based on the corrected energy data.

[0043] Furthermore, the third neural network and the fourth neural network are obtained by training the LSTM model using historical medical equipment energy consumption data and sanitary hot water energy consumption data respectively.

[0044] Furthermore, the first graph neural network is constructed in such a way that the M1 air outlets in the air-conditioning system are used as nodes to form a graph, and if two air outlets are in the same closed area, there is an edge between the two air outlets.

[0045] The second graph neural network is constructed in such a way that: a graph is constructed with N1 light sources in the lighting system as nodes, and if two light sources are connected to the same circuit, there is an edge between the two light sources;

[0046] Among them, M1≥3, N1≥3.

[0047] Furthermore, the characteristic value of the i-th node among the M1 nodes in the first graph neural network is for:

[0048] ;

[0049] in, represents the outlet air temperature of the ith air outlet, represents the set temperature of the i-th air outlet, represents the temperature of the liquid medium in the pipe connected to the i-th air outlet;

[0050] The feature value of the jth node among the N1 nodes in the second graph neural network for:

[0051] ;

[0052] in, represents the brightness of the jth light source,

[0053] Among them, 1≤i≤M1, 1≤j≤N1.

[0054] Furthermore, the operation of correcting the abnormal air-conditioning system energy consumption data based on the evaluation result of the first graph neural network is:

[0055] The evaluation result of the first graph neural network is the data anomalies of M2 air-conditioning outlets and the corresponding anomaly probabilities. If the anomaly probability is greater than the first threshold, the energy consumption data of other normal air-conditioning outlets in the same closed space as the air-conditioning outlet are averaged as the energy consumption data of the air-conditioning outlet. If the anomaly probability is less than or equal to the first threshold, the energy consumption data of the air-conditioning outlet is corrected to :

[0056] ;

[0057] in, represents the data anomaly probability of the qth air outlet output by the first graph neural network, Indicates the energy consumption data collected for the qth air outlet, It represents the average value of energy consumption data of other normal air-conditioning outlets in the same closed space as the qth air outlet;

[0058] Furthermore, the operation of correcting the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network is as follows: the evaluation result of the second graph neural network is the data abnormality of N2 light sources and the corresponding abnormal probability. If the abnormal probability is greater than the second threshold, the data of the abnormal light source is directly discarded, and a notice is issued that the light source needs maintenance. If the abnormal probability is less than or equal to the second threshold, the energy consumption data of other normal light sources with the same power as the light source are averaged as the energy consumption data of the light source. Wherein, M1>M2, N1>N2.

[0059] The present invention further provides a computer-readable storage medium, wherein the storage medium stores computer program code, and when the computer program code is executed by a computer, any of the above methods is executed.

[0060] The technical effect of the present invention is: a hospital energy data evaluation method, device and storage medium of the present invention, the method comprising: an acquisition step S101, collecting energy data in a hospital within a period of time through an Internet of Things sensor, the energy data including air-conditioning system energy consumption data, lighting system energy consumption data, medical equipment energy consumption data and sanitary hot water energy consumption data; a preliminary evaluation step S102, determining the presence of abnormal energy data through a Z test on the air-conditioning system energy consumption data, lighting system energy consumption data, medical equipment energy consumption data and sanitary hot water energy consumption data; an evaluation step S103, based on the type of energy data with abnormalities, using a corresponding method to perform an evaluation, specifically comprising: if the energy data with abnormalities is air-conditioning system energy consumption data, constructing a first graph neural network to evaluate the collected air-conditioning system energy consumption data, determining the presence of abnormal air-conditioning energy consumption data, and evaluating the abnormal data based on the evaluation result of the first graph neural network. The energy consumption data of the air-conditioning system is corrected; if the energy data with abnormalities is the energy consumption data of the lighting system, a second graph neural network is constructed to evaluate the collected energy consumption data of the lighting system, determine that there are abnormal energy consumption data of the lighting system, and correct the abnormal energy consumption data of the lighting system based on the evaluation result of the second graph neural network; if the energy data with abnormalities is the energy consumption data of medical equipment, a third neural network is constructed to evaluate the collected energy consumption data of medical equipment, determine that there are abnormal energy consumption data of medical equipment, and correct the abnormal energy consumption data of medical equipment; if the energy data with abnormalities is the energy consumption data of sanitary hot water, a fourth neural network is constructed to evaluate the collected energy consumption data of sanitary hot water, determine that there are abnormal energy consumption data of sanitary hot water, and correct the abnormal energy consumption data of sanitary hot water; control step S104, energy control of the corresponding energy consumption system is performed based on the corrected energy data. In the present invention, energy data in a hospital over a period of time is first collected through Internet of Things sensors, and then a preliminary evaluation is performed on it. After the preliminary evaluation, it is determined that there is abnormal energy data, and the type of abnormal energy data is determined, such as abnormal air-conditioning system energy consumption data, or abnormal lighting system energy consumption data, etc., and then different neural networks are used to accurately evaluate the data according to different data types, and correction is performed. In the present invention, since the air-conditioning system and the lighting system are distributed systems, that is, distributed throughout the hospital, a trained graph neural network is used for evaluation. Medical equipment and sanitary hot water are generally single devices, and the data between the devices are unrelated. Therefore, a trained LSTM model is used for evaluation and correction. That is, the present invention proposes to use different neural networks for accurate evaluation of different data types in the hospital, thereby improving data quality, so that the energy consumption control method of the equipment can be more accurately formulated to achieve the technical effect of energy saving. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Other features, objects and advantages of the present application will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings.

[0062] Figure 1 is a flow chart of a hospital energy data evaluation method according to an embodiment of the present invention.

[0063] Figure 2 4 is a structural diagram of a hospital energy data evaluation device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0064] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings.

[0065] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0066] Figure 1 A hospital energy data evaluation method of the present invention is shown, the method comprising:

[0067] Acquisition step S101, collecting energy data in the hospital over a period of time through an Internet of Things sensor, the energy data including air conditioning system energy consumption data, lighting system energy consumption data, medical equipment energy consumption data and sanitary hot water energy consumption data;

[0068] A preliminary evaluation step S102, determining whether there is abnormal energy data by performing a Z test on the energy consumption data of the air conditioning system, the energy consumption data of the lighting system, the energy consumption data of the medical equipment and the sanitary hot water;

[0069] Evaluation step S103, based on the type of abnormal energy data, uses a corresponding method to perform evaluation, specifically including:

[0070] If the energy data with abnormalities is air-conditioning system energy consumption data, construct a first graph neural network to evaluate the collected air-conditioning system energy consumption data, determine that there are abnormal air-conditioning energy consumption data, and correct the abnormal air-conditioning system energy consumption data based on the evaluation result of the first graph neural network;

[0071] If the energy data with abnormalities is lighting system energy consumption data, construct a second graph neural network to evaluate the collected lighting system energy consumption data, determine that there are abnormal lighting system energy consumption data, and correct the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network;

[0072] If the energy data with abnormalities is medical equipment energy consumption data, construct a third neural network to evaluate the collected medical equipment energy consumption data, determine that there are abnormal medical equipment energy consumption data, and correct the abnormal medical equipment energy consumption data;

[0073] If the energy data with abnormalities is sanitary hot water energy consumption data, construct a fourth neural network to evaluate the collected sanitary hot water energy consumption data, determine that there are abnormal sanitary hot water energy consumption data, and correct the abnormal sanitary hot water energy consumption data;

[0074] Control step S104: performing energy control on the corresponding energy consumption system based on the corrected energy data.

[0075] In the present invention, energy data in a hospital within a period of time is first collected through an Internet of Things sensor, and then a preliminary evaluation is performed on it. After the preliminary evaluation, it is determined that there is abnormal energy data, and the type of abnormal energy data is determined, such as abnormal air-conditioning system energy consumption data, or abnormal lighting system energy consumption data, etc., and then different neural networks are used to accurately evaluate the data according to different data types, and correction is performed. In the present invention, since the air-conditioning system and the lighting system are distributed systems, that is, distributed throughout the hospital, a trained graph neural network is used for evaluation. Medical equipment and sanitary hot water are generally single devices, and the data between the devices are unrelated. Therefore, a trained LSTM model is used for evaluation and correction. That is, the present invention proposes to use different neural networks for accurate evaluation of different data types in the hospital, thereby improving data quality, so that the energy consumption control method of the equipment can be more accurately formulated to achieve energy-saving technical effects. This is a manifestation of the important inventive concept of the present invention.

[0076] In one embodiment, the third neural network and the fourth neural network are obtained by training the LSTM model using historical medical equipment energy consumption data and sanitary hot water energy consumption data respectively. The LSTM model is a long short-term memory neural network and has good prediction performance for single time series data. In the present invention, since the energy consumption between medical equipment and sanitary hot water is not correlated, LSTM is used for evaluation, which is one of the specific manifestations of the inventive concept of the present invention.

[0077] In one embodiment, the first graph neural network is constructed in such a way that M1 air outlets in the air-conditioning system are used as nodes to form a graph, and if two air outlets are in the same closed area, there is an edge between the two air outlets.

[0078] The second graph neural network is constructed in such a way that: a graph is constructed with N1 light sources in the lighting system as nodes, and if two light sources are connected to the same circuit, there is an edge between the two light sources;

[0079] Among them, M1≥3, N1≥3.

[0080] In one embodiment, the feature value of the i-th node among the M1 nodes in the first graph neural network for:

[0081] ;

[0082] in, represents the outlet air temperature of the ith air outlet, represents the set temperature of the i-th air outlet, represents the temperature of the liquid medium in the pipe connected to the i-th air outlet;

[0083] The feature value of the jth node among the N1 nodes in the second graph neural network for:

[0084] ;

[0085] in, represents the brightness of the jth light source,

[0086] Among them, 1≤i≤M1, 1≤j≤N1.

[0087] In one embodiment, the weights of the edges in the graph in the first graph neural network are:

[0088] ;

[0089] The weights of the edges in the graph in the second graph neural network are:

[0090] .

[0091] In the present invention, a method for constructing a first graph neural network and a second graph neural network for evaluating the energy consumption data of the air-conditioning system and the energy consumption data of the lighting system, a method for determining the eigenvalues ​​of the nodes, and a method for determining the weights of the edges are proposed. The present invention determines the relevant eigenvalues ​​and weight calculations based on the actual physical parameters of the air-conditioning system and the lighting system. Since the air-conditioning system and the lighting system are different, different methods for calculating the eigenvalues ​​and edge weights are proposed for different systems. Through the two graph neural networks constructed by the present invention, it is possible to quickly and accurately locate which node of the air-conditioning system and the lighting system has problems with the energy consumption data. Compared with traditional methods, the efficiency of discovering abnormal data is greatly improved, thereby improving the fault detection rate. This is a manifestation of the important inventive concept of the present invention.

[0092] In one embodiment, the operation of correcting abnormal air conditioning system energy consumption data based on the evaluation result of the first graph neural network is:

[0093] The evaluation result of the first graph neural network is the data anomalies of M2 air-conditioning outlets and the corresponding anomaly probabilities. If the anomaly probability is greater than the first threshold, the energy consumption data of other normal air-conditioning outlets in the same closed space as the air-conditioning outlet are averaged as the energy consumption data of the air-conditioning outlet. If the anomaly probability is less than or equal to the first threshold, the energy consumption data of the air-conditioning outlet is corrected to :

[0094] ;

[0095] in, represents the data anomaly probability of the qth air outlet output by the first graph neural network, Indicates the energy consumption data collected for the qth air outlet, It represents the average value of energy consumption data of other normal air-conditioning outlets in the same closed space as the qth air outlet;

[0096] In one embodiment, the operation of correcting the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network is as follows: the evaluation result of the second graph neural network is the data abnormality of N2 light sources and the corresponding abnormal probability. If the abnormal probability is greater than the second threshold, the data of the abnormal light source is directly discarded, and a notice is issued that the light source needs maintenance. If the abnormal probability is less than or equal to the second threshold, the energy consumption data of other normal light sources with the same power as the light source are averaged as the energy consumption data of the light source. Wherein, M1>M2, N1>N2.

[0097] In the present invention, correction methods are proposed for the abnormal data of the air-conditioning system and the lighting system respectively. Different correction methods are adopted according to the different data characteristics of the two. The air-conditioning is generally calibrated based on the energy consumption data of other normal air outlets in the same enclosed space, and the lighting system requires the energy consumption data of other normal light sources with the same power for calibration. If the data is abnormal to the point where it cannot be used, it may be that the light source is damaged, and an alarm message is sent. This is another inventive concept of the present invention.

[0098] Figure 2 A hospital energy data evaluation device of the present invention is shown, the device comprising:

[0099] An acquisition unit 201 collects energy data in a hospital over a period of time through an Internet of Things sensor, wherein the energy data includes energy consumption data of an air conditioning system, energy consumption data of a lighting system, energy consumption data of medical equipment, and energy consumption data of sanitary hot water;

[0100] A preliminary evaluation unit 202 determines whether there is abnormal energy data by performing a Z test on the air conditioning system energy consumption data, the lighting system energy consumption data, the medical equipment energy consumption data, and the sanitary hot water energy consumption data;

[0101] The evaluation unit 203 performs evaluation in a corresponding manner based on the type of abnormal energy data, specifically including:

[0102] If the energy data with abnormalities is air-conditioning system energy consumption data, construct a first graph neural network to evaluate the collected air-conditioning system energy consumption data, determine that there are abnormal air-conditioning energy consumption data, and correct the abnormal air-conditioning system energy consumption data based on the evaluation result of the first graph neural network;

[0103] If the energy data with abnormalities is lighting system energy consumption data, construct a second graph neural network to evaluate the collected lighting system energy consumption data, determine that there are abnormal lighting system energy consumption data, and correct the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network;

[0104] If the energy data with abnormalities is medical equipment energy consumption data, construct a third neural network to evaluate the collected medical equipment energy consumption data, determine that there are abnormal medical equipment energy consumption data, and correct the abnormal medical equipment energy consumption data;

[0105] If the energy data with abnormalities is sanitary hot water energy consumption data, construct a fourth neural network to evaluate the collected sanitary hot water energy consumption data, determine that there are abnormal sanitary hot water energy consumption data, and correct the abnormal sanitary hot water energy consumption data;

[0106] The control unit 204 performs energy control on the corresponding energy consumption system based on the corrected energy data.

[0107] In the present invention, energy data in a hospital within a period of time is first collected through an Internet of Things sensor, and then a preliminary evaluation is performed on it. After the preliminary evaluation, it is determined that there is abnormal energy data, and the type of abnormal energy data is determined, such as abnormal air-conditioning system energy consumption data, or abnormal lighting system energy consumption data, etc., and then different neural networks are used to accurately evaluate the data according to different data types, and correction is performed. In the present invention, since the air-conditioning system and the lighting system are distributed systems, that is, distributed throughout the hospital, a trained graph neural network is used for evaluation. Medical equipment and sanitary hot water are generally single devices, and the data between the devices are unrelated. Therefore, a trained LSTM model is used for evaluation and correction. That is, the present invention proposes to use different neural networks for accurate evaluation of different data types in the hospital, thereby improving data quality, so that the energy consumption control method of the equipment can be more accurately formulated to achieve energy-saving technical effects. This is a manifestation of the important inventive concept of the present invention.

[0108] In one embodiment, the third neural network and the fourth neural network are obtained by training the LSTM model using historical medical equipment energy consumption data and sanitary hot water energy consumption data respectively. The LSTM model is a long short-term memory neural network and has good prediction performance for single time series data. In the present invention, since the energy consumption between medical equipment and sanitary hot water is not correlated, LSTM is used for evaluation, which is one of the specific manifestations of the inventive concept of the present invention.

[0109] In one embodiment, the first graph neural network is constructed in such a way that M1 air outlets in the air-conditioning system are used as nodes to form a graph, and if two air outlets are in the same closed area, there is an edge between the two air outlets.

[0110] The second graph neural network is constructed in such a way that: a graph is constructed with N1 light sources in the lighting system as nodes, and if two light sources are connected to the same circuit, there is an edge between the two light sources;

[0111] Among them, M1≥3, N1≥3.

[0112] In one embodiment, the feature value of the i-th node among the M1 nodes in the first graph neural network for:

[0113] ;

[0114] in, represents the outlet air temperature of the ith air outlet, represents the set temperature of the i-th air outlet, represents the temperature of the liquid medium in the pipe connected to the i-th air outlet;

[0115] The feature value of the jth node among the N1 nodes in the second graph neural network for:

[0116] ;

[0117] in, represents the brightness of the jth light source,

[0118] Among them, 1≤i≤M1, 1≤j≤N1.

[0119] In one embodiment, the weights of the edges in the graph in the first graph neural network are:

[0120] ;

[0121] The weights of the edges in the graph in the second graph neural network are:

[0122] .

[0123] In the present invention, a method for constructing a first graph neural network and a second graph neural network for evaluating the energy consumption data of the air-conditioning system and the energy consumption data of the lighting system, a method for determining the eigenvalues ​​of the nodes, and a method for determining the weights of the edges are proposed. The present invention determines the relevant eigenvalues ​​and weight calculations based on the actual physical parameters of the air-conditioning system and the lighting system. Since the air-conditioning system and the lighting system are different, different methods for calculating the eigenvalues ​​and edge weights are proposed for different systems. Through the two graph neural networks constructed by the present invention, it is possible to quickly and accurately locate which node of the air-conditioning system and the lighting system has problems with the energy consumption data. Compared with traditional methods, the efficiency of discovering abnormal data is greatly improved, thereby improving the fault detection rate. This is a manifestation of the important inventive concept of the present invention.

[0124] In one embodiment, the operation of correcting abnormal air conditioning system energy consumption data based on the evaluation result of the first graph neural network is:

[0125] The evaluation result of the first graph neural network is the data anomalies of M2 air-conditioning outlets and the corresponding anomaly probabilities. If the anomaly probability is greater than the first threshold, the energy consumption data of other normal air-conditioning outlets in the same closed space as the air-conditioning outlet are averaged as the energy consumption data of the air-conditioning outlet. If the anomaly probability is less than or equal to the first threshold, the energy consumption data of the air-conditioning outlet is corrected to :

[0126] ;

[0127] in, represents the data anomaly probability of the qth air outlet output by the first graph neural network, Indicates the energy consumption data collected for the qth air outlet, It represents the average value of energy consumption data of other normal air-conditioning outlets in the same closed space as the qth air outlet;

[0128] In one embodiment, the operation of correcting the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network is as follows: the evaluation result of the second graph neural network is the data abnormality of N2 light sources and the corresponding abnormal probability. If the abnormal probability is greater than the second threshold, the data of the abnormal light source is directly discarded, and a notice is issued that the light source needs maintenance. If the abnormal probability is less than or equal to the second threshold, the energy consumption data of other normal light sources with the same power as the light source are averaged as the energy consumption data of the light source. Wherein, M1>M2, N1>N2.

[0129] In the present invention, correction methods are proposed for the abnormal data of the air-conditioning system and the lighting system respectively. Different correction methods are adopted according to the different data characteristics of the two. The air-conditioning is generally calibrated based on the energy consumption data of other normal air outlets in the same enclosed space, and the lighting system requires the energy consumption data of other normal light sources with the same power for calibration. If the data is abnormal to the point where it cannot be used, it may be that the light source is damaged, and an alarm message is sent. This is another inventive concept of the present invention.

[0130] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0131] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a storage medium, such as ROM / RAM, a disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the devices described in each embodiment of the present application or some parts of the embodiments.

[0132] Finally, it should be noted that the above embodiments are only intended to illustrate rather than limit the technical solutions of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the present invention can still be modified or replaced by equivalents. Any modification or partial replacement that does not depart from the spirit and scope of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A hospital energy data evaluation method, characterized in that: The method includes: An acquisition step is to collect energy data in the hospital over a period of time through an Internet of Things sensor, wherein the energy data includes energy consumption data of an air conditioning system, energy consumption data of a lighting system, energy consumption data of medical equipment, and energy consumption data of sanitary hot water; A preliminary evaluation step, determining abnormal energy data by performing a Z test on the air conditioning system energy consumption data, the lighting system energy consumption data, the medical equipment energy consumption data, and the sanitary hot water energy consumption data; The evaluation steps are based on the type of abnormal energy data and use corresponding methods to evaluate, including: If the energy data with abnormalities is air-conditioning system energy consumption data, construct a first graph neural network to evaluate the collected air-conditioning system energy consumption data, determine that there are abnormal air-conditioning energy consumption data, and correct the abnormal air-conditioning system energy consumption data based on the evaluation result of the first graph neural network; If the energy data with abnormalities is lighting system energy consumption data, construct a second graph neural network to evaluate the collected lighting system energy consumption data, determine that there are abnormal lighting system energy consumption data, and correct the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network; If the energy data with abnormalities is medical equipment energy consumption data, construct a third neural network to evaluate the collected medical equipment energy consumption data, determine that there are abnormal medical equipment energy consumption data, and correct the abnormal medical equipment energy consumption data; If the energy data with abnormalities is sanitary hot water energy consumption data, construct a fourth neural network to evaluate the collected sanitary hot water energy consumption data, determine that there are abnormal sanitary hot water energy consumption data, and correct the abnormal sanitary hot water energy consumption data; The control step is to perform energy control on the corresponding energy consumption system based on the corrected energy data.

2. The method according to claim 1, characterized in that The third neural network and the fourth neural network are obtained by training the LSTM model using historical medical equipment energy consumption data and sanitary hot water energy consumption data respectively.

3. The method according to claim 2, characterized in that The first graph neural network is constructed in such a way that: the M1 air outlets in the air conditioning system are used as nodes to form a graph, and if two air outlets are in the same closed area, there is an edge between the two air outlets. The second graph neural network is constructed in such a way that: a graph is constructed with N1 light sources in the lighting system as nodes, and if two light sources are connected to the same circuit, there is an edge between the two light sources; Among them, M1≥3, N1≥3.

4. The method according to claim 3, characterized in that The characteristic value Aircond of the i-th node among the M1 nodes in the first graph neural network i for: Among them, f i represents the outlet air temperature of the ith outlet, s i represents the set temperature of the i-th air outlet, c i represents the temperature of the liquid medium in the pipe connected to the i-th air outlet; The feature value Light of the jth node among the N1 nodes in the second graph neural network j for: Among them, lum j represents the brightness of the jth light source, Among them, 1≤i≤M1, 1≤j≤N1.

5. The method according to claim 4, characterized in that The weights of the edges in the graph in the first graph neural network are: The weights of the edges in the graph in the second graph neural network are: The operation of correcting the abnormal air conditioning system energy consumption data based on the evaluation result of the first graph neural network is: The evaluation result of the first graph neural network is the data anomalies of M2 air-conditioning outlets and the corresponding anomaly probabilities. If the anomaly probability is greater than the first threshold, the energy consumption data of other normal air-conditioning outlets in the same closed space as the air-conditioning outlet are averaged as the energy consumption data of the air-conditioning outlet. If the anomaly probability is less than or equal to the first threshold, the energy consumption data of the air-conditioning outlet is corrected to f′. q : f′ q =err q *f q +(1-err q )*Aver q ; Among them, err q represents the data anomaly probability of the qth air outlet output by the first graph neural network, f q Indicates the energy consumption data collected for the qth air outlet, Aver q It represents the average value of energy consumption data of other normal air-conditioning outlets in the same closed space as the qth air outlet; The operation of correcting the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network is: the evaluation result of the second graph neural network is the data abnormality of N2 light sources and the corresponding abnormal probability. If the abnormal probability is greater than the second threshold, the data of the abnormal light source is directly discarded, and a notification is issued that the light source needs maintenance. If the abnormal probability is less than or equal to the second threshold, the energy consumption data of other normal light sources with the same power as the light source are averaged as the energy consumption data of the light source.

6. A hospital energy data evaluation device, characterized in that: The device includes: An acquisition unit collects energy data in the hospital over a period of time through an Internet of Things sensor, wherein the energy data includes energy consumption data of an air conditioning system, energy consumption data of a lighting system, energy consumption data of medical equipment, and energy consumption data of sanitary hot water; A preliminary evaluation unit, for determining abnormal energy data by performing a Z test on the energy consumption data of the air conditioning system, the energy consumption data of the lighting system, the energy consumption data of the medical equipment, and the energy consumption data of the sanitary hot water; The evaluation unit uses a corresponding method to evaluate the type of abnormal energy data, including: If the energy data with abnormalities is air-conditioning system energy consumption data, construct a first graph neural network to evaluate the collected air-conditioning system energy consumption data, determine that there are abnormal air-conditioning energy consumption data, and correct the abnormal air-conditioning system energy consumption data based on the evaluation result of the first graph neural network; If the energy data with abnormalities is lighting system energy consumption data, construct a second graph neural network to evaluate the collected lighting system energy consumption data, determine that there are abnormal lighting system energy consumption data, and correct the abnormal lighting system energy consumption data based on the evaluation result of the second graph neural network; If the energy data with abnormalities is medical equipment energy consumption data, construct a third neural network to evaluate the collected medical equipment energy consumption data, determine that there are abnormal medical equipment energy consumption data, and correct the abnormal medical equipment energy consumption data; If the energy data with abnormalities is sanitary hot water energy consumption data, construct a fourth neural network to evaluate the collected sanitary hot water energy consumption data, determine that there are abnormal sanitary hot water energy consumption data, and correct the abnormal sanitary hot water energy consumption data; The control unit performs energy control on the corresponding energy consumption system based on the corrected energy data.

7. The device according to claim 6, characterized in that The third neural network and the fourth neural network are obtained by training the LSTM model using historical medical equipment energy consumption data and sanitary hot water energy consumption data respectively.

8. The device according to claim 7, characterized in that The first graph neural network is constructed in such a way that: the M1 air outlets in the air conditioning system are used as nodes to form a graph, and if two air outlets are in the same closed area, there is an edge between the two air outlets. The second graph neural network is constructed in such a way that: a graph is constructed with N1 light sources in the lighting system as nodes, and if two light sources are connected to the same circuit, there is an edge between the two light sources; Among them, M1≥3, N1≥3.

9. The device according to claim 8, characterized in that The characteristic value Aircond of the i-th node among the M1 nodes in the first graph neural network i for: Among them, f i represents the outlet air temperature of the ith outlet, s i represents the set temperature of the i-th air outlet, c i represents the temperature of the liquid medium in the pipe connected to the i-th air outlet; The feature value Light of the jth node among the N1 nodes in the second graph neural network j for: Among them, lum j represents the brightness of the jth light source, Among them, 1≤i≤M1, 1≤j≤N1.

10. A computer-readable storage medium, wherein a computer program code is stored on the storage medium, and when the computer program code is executed by a computer, the method according to any one of claims 1 to 5 is executed.