Property equipment control method and device based on digital twinning, equipment, medium

Abstract

The application provides a property equipment control method and device based on digital twinning, equipment and a medium. The method comprises the following steps: constructing a digital twinning model based on entity data of each property equipment, determining a first instance with a preset trigger condition in a plurality of model instances of the digital twinning model, and determining a second instance associated with each first instance; determining a plurality of control strategies of each model instance based on an energy consumption prediction value predicted by an LSTM model; generating a switching strategy and a second running state in each control strategy of the first instance and the second instance based on the trigger condition; and controlling the corresponding property equipment based on any model instance and based on a control information set. According to the technical scheme of the embodiment of the application, the switching strategy and the second running state can be added to the first instance and the associated second instance by using the trigger condition, so that the corresponding property equipment can automatically switch the running state according to the actual situation without manual intervention, thereby improving the efficiency and the automation degree of energy management.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method, apparatus, equipment, and medium for controlling property equipment based on digital twins. Background Technology

[0002] With the development of smart property management, office buildings, shopping malls, residential communities, and industrial parks typically contain numerous energy-related devices. These include energy-consuming equipment such as central air conditioning, lighting systems, and public water and electricity facilities, as well as energy-generating equipment like photovoltaic (PV) systems. The usage demands of property equipment are usually not fixed. For example, maintaining central air conditioning and lighting systems when there are no users will result in unnecessary energy consumption. Conversely, when electricity consumption is high, the electricity generated by PV systems can be used, while when electricity consumption is low, the generated electricity can be fed into the grid. Therefore, how to rationally control the operating status of property equipment has a significant impact on energy conservation.

[0003] In related technologies, artificial intelligence techniques such as Long Short-Term Memory (LSTM) networks can be used to predict the usage demand of property facilities, such as predicting whether users will use a certain area within a specific time period or predicting electricity consumption during a specific time period. However, the flow of people or electricity consumption within a property area changes constantly. The usage demand predicted by artificial intelligence technology can only be used as a reference. Property managers still need to manually control the system based on the actual situation, resulting in low energy management efficiency and a low degree of automation for property facilities. Summary of the Invention

[0004] This invention aims to at least solve one of the technical problems existing in the prior art. To this end, this invention proposes a property equipment control method, device, equipment, and medium based on digital twins, which can dynamically deploy management strategies for each property equipment using a digital twin model of the property area, thereby achieving automatic management of property equipment, reducing manual intervention, and reducing unnecessary energy consumption.

[0005] In a first aspect, embodiments of the present invention provide a property equipment control method based on digital twins, wherein the property management system includes multiple property equipment, and the method includes:

[0006] A digital twin model is constructed based on the physical data of each of the property devices, and a first instance is determined. Each of the property devices corresponds to a model instance in the digital twin model. The first instance is the model instance with preset trigger conditions, which are used to indicate changes in human body signals or preset reference operating parameters.

[0007] Based on the pre-trained LSTM model, the energy consumption prediction value of each model instance on the target date is predicted, and a set of control information is determined based on the energy consumption prediction value. The set of control information includes multiple control strategies, and each control strategy includes a first time period and a first operating state. The first time periods of each control strategy do not overlap.

[0008] Based on any of the first instances, at least one pre-associated second instance is determined. In each of the control strategies corresponding to the first instance and the second instance, a switching strategy and a second operating state are generated based on the triggering conditions. The second operating state is the opposite of the first operating state. The second instance and the property equipment corresponding to each of the first instances are pre-configured with an association relationship.

[0009] Based on any of the model instances, the corresponding property equipment is controlled based on the set of control information.

[0010] According to some embodiments of the present invention, in each corresponding control strategy, a switching strategy and a second operating state are generated based on the triggering conditions, including:

[0011] When the first operating state corresponding to the control strategy is used to indicate energy saving, the switching strategy is used to indicate that the property equipment corresponding to the target instance detects that the value of the human signal or the reference operating parameter is greater than a preset threshold, wherein the target instance is the first instance and / or the associated second instance;

[0012] Alternatively, when the first operating state corresponding to the control strategy is used to indicate energy consumption, the switching strategy is used to indicate that the property equipment corresponding to the target instance loses the human signal or the value of the reference operating parameter is less than or equal to a preset threshold.

[0013] According to some embodiments of the present invention, before predicting the energy consumption forecast value of each model instance on a target date based on a pre-trained LSTM model, the method further includes:

[0014] Obtain the historical energy consumption values ​​of each of the aforementioned property equipment, and construct a model training set based on all the historical energy consumption values, wherein each of the historical energy consumption values ​​corresponds to a historical time period;

[0015] Obtain the preset LSTM model, wherein the LSTM model is a two-layer bidirectional model, the first layer of the LSTM model is an output retention layer, and the second layer of the LSTM model is a sequence compression layer;

[0016] The model training set is input into the retained output layer and the compressed sequence layer respectively for model training, wherein the input data of the retained output layer includes the model training set and the compressed feature sequence output by the compressed sequence layer based on the model training set;

[0017] Multiple instance groups are constructed, and the trained LSTM model is configured into each instance group. When an instance group includes multiple model instances, the property equipment corresponding to each model instance is pre-associated.

[0018] According to some embodiments of the present invention, an energy consumption prediction value for each model instance on a target date is predicted based on a pre-trained LSTM model, and a control information set is determined based on the energy consumption prediction value, including:

[0019] Based on any of the model instances, based on the multiple first time periods predicted by the LSTM model based on the historical time period, and the energy consumption prediction values ​​of each of the first time periods predicted based on the historical energy consumption values;

[0020] Based on any of the first time periods, obtain the historical operating status corresponding to each of the historical energy consumption values;

[0021] When the predicted energy consumption value is less than or equal to the historical energy consumption value and the preset energy consumption threshold, and the historical operating status is used to indicate energy consumption, the first operating status is determined to indicate energy saving.

[0022] Alternatively, when the predicted energy consumption value is greater than the historical energy consumption value and the energy consumption threshold, and the historical operating status is used to indicate energy saving, the first operating status is determined to indicate energy consumption.

[0023] Alternatively, the historical operating state can be determined as the first operating state.

[0024] According to some embodiments of the present invention, determining the control information set based on the energy consumption prediction value includes:

[0025] Based on a preset target time, the energy consumption data of each property device on the current date is obtained, wherein the energy consumption data is used to indicate the energy consumption data of the second time period under different operating modes, and the second time periods of the same property device do not overlap with each other;

[0026] Based on any of the instance groups, the assigned LSTM model is incrementally trained using the corresponding device energy consumption data;

[0027] Based on the incrementally trained LSTM model, the control information set of each model instance is predicted on the target date, wherein the target date is the day after the current date.

[0028] According to some embodiments of the present invention, before incrementally training the assigned LSTM model using the corresponding device energy consumption data, the method further includes:

[0029] Arrange the second time period corresponding to each model instance in chronological order, form a time period group based on multiple second time periods in the same order, and traverse the queue formed by the time period groups in sequence.

[0030] When the operating status corresponding to the time period group is used to indicate energy saving, the earliest recorded start time is determined as the target start time of each second time period, and the latest recorded end time is determined as the target end time of each second time period. Wherein, when the target start time determined this time is earlier than the target end time of the previous time period group, the target end time of the previous time period group is modified based on the target start time determined this time.

[0031] Alternatively, when the operating status corresponding to the traversed time period group is used to indicate energy consumption, the latest recorded start time is determined as the target start time of each second time period, and the earliest recorded end time is determined as the target end time of each second time period. Wherein, when the modified target start time is later than the target end time of the previous time period group, the target start time is determined as the target end time of the previous time period group.

[0032] According to some embodiments of the present invention, before incrementally training the assigned LSTM model using the corresponding device energy consumption data, the method further includes:

[0033] Based on any of the model instances, segmented energy consumption data for each of the second time periods are determined;

[0034] Based on any adjusted second time period, when the corresponding operating state is used to indicate energy saving, the newly added time period is interpolated based on the segmented energy consumption data with the smallest value; or, when the corresponding operating state is used to indicate energy consumption, the newly added time period is interpolated based on the average value of the segmented energy consumption data.

[0035] In a second aspect, embodiments of the present invention provide a property equipment control device based on digital twins, including at least one control processor and a memory for communicatively connecting with the at least one control processor; the memory stores instructions executable by the at least one control processor, which, when executed by the at least one control processor, enables the at least one control processor to perform the property equipment control method based on digital twins as described in the first aspect above.

[0036] Thirdly, embodiments of the present invention provide an electronic device including a property equipment control device based on digital twins as described in the second aspect above.

[0037] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer-executable instructions for performing the property equipment control method based on digital twins as described in the first aspect above.

[0038] The property equipment control method based on digital twins according to embodiments of the present invention has at least the following beneficial effects: A digital twin model is constructed based on the entity data of each property equipment, and a first instance is determined, wherein each property equipment corresponds to a model instance in the digital twin model, and the first instance is a model instance with preset trigger conditions, the trigger conditions being used to indicate changes in human body signals or preset reference operating parameters; Energy consumption prediction values ​​for each model instance on a target date are predicted based on a pre-trained LSTM model, and a control information set is determined based on the energy consumption prediction values, wherein the control information set includes multiple control strategies, each control strategy including a first time period and a first operating state, and the first time periods of each control strategy do not overlap; Based on any first instance, at least one pre-associated second instance is determined, and a switching strategy and a second operating state are generated in each control strategy corresponding to the first instance and the second instance based on the trigger conditions, wherein the second operating state is the opposite state of the first operating state, and the second instance and the property equipment corresponding to each of the first instances are pre-configured with an association relationship; Based on any model instance, the corresponding property equipment is controlled based on the control information set. According to the technical solution of the present invention, a model instance of each property equipment can be constructed using digital twin technology. After the energy consumption prediction value is obtained by combining the LSTM model, the same switching strategy and second operating state can be added to the first instance and the associated second instance using trigger conditions. This enables the corresponding property equipment to automatically switch operating states according to the actual situation without manual intervention, thereby improving the efficiency and automation of energy management. Attached Figure Description

[0039] Figure 1This is a schematic diagram illustrating the principle of a property equipment control method based on digital twins provided in one embodiment of the present invention;

[0040] Figure 2 This is a flowchart of a property equipment control method based on digital twins provided in another embodiment of the present invention;

[0041] Figure 3 This is a structural diagram of a property equipment control device based on digital twins provided in another embodiment of the present invention. Detailed Implementation

[0042] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0043] In the description of this invention, it should be understood that the orientation descriptions, such as up, down, front, back, left, right, etc., are based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0044] In the description of this invention, "several" means one or more, "more than" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0045] In the description of this invention, unless otherwise explicitly defined, terms such as "set up," "install," and "connect" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0046] This invention provides a property equipment control method, apparatus, device, and medium based on digital twins. The method includes: constructing a digital twin model based on the entity data of each property equipment; determining a first instance, wherein each property equipment corresponds to a model instance in the digital twin model; the first instance being a model instance with preset trigger conditions, the trigger conditions indicating changes in human body signals or preset reference operating parameters; predicting the energy consumption forecast value of each model instance on a target date based on a pre-trained LSTM model; determining a control information set based on the energy consumption forecast value; wherein the control information set includes multiple control strategies, each control strategy including a first time period and a first operating state, the first time periods of each control strategy not overlapping; determining at least one pre-associated second instance based on any first instance; generating a switching strategy and a second operating state based on the trigger conditions in each control strategy corresponding to the first instance and the second instance; wherein the second operating state is the opposite state of the first operating state; and the second instance and the property equipment corresponding to each of the first instances are pre-configured with an association relationship; and controlling the corresponding property equipment based on any model instance and the control information set. According to the technical solution of the present invention, a model instance of each property equipment can be constructed using digital twin technology. After the energy consumption prediction value is obtained by combining the LSTM model, the same switching strategy and second operating state can be added to the first instance and the associated second instance using trigger conditions. This enables the corresponding property equipment to automatically switch operating states according to the actual situation without manual intervention, thereby improving the efficiency and automation of energy management.

[0047] First, refer to Figure 1 , Figure 1 This is a schematic diagram of the principle provided for an embodiment of the present invention. The property management system of this embodiment includes multiple property devices.

[0048] It should be noted that property equipment can be IoT devices in the property management field, which can be controlled by the control signals of the property management system. Examples include lighting equipment, air conditioning, escalators, elevators, photovoltaic equipment, etc. The property management system can control the operating status of each property device through remote signals.

[0049] The operating state of this embodiment is not limited to on / off; it can also be different operating modes. For example, the brightness of the office building's lighting equipment can be adjusted at different times. During off-hours, basic lighting can be provided at a lower brightness to reduce power consumption, while during working hours, cleaning lighting can be provided at a higher brightness to improve the user experience. Similarly, air conditioners can be switched on and off or their operating temperature adjusted at different times to regulate energy consumption. Escalators can switch to a slower motor speed when not in use to reduce energy consumption. Elevators can turn off their internal lighting when in standby mode. Furthermore, the power generated by photovoltaic equipment can be supplied to the community's electrical equipment during peak electricity consumption periods and output to the grid during off-peak periods.

[0050] The following is based on the appendix Figure 1 The schematic diagram shown further illustrates the technical solution of the embodiment of the present invention.

[0051] Reference Figure 2 , Figure 2 The flowchart illustrates a property equipment control method based on digital twins, as provided in an embodiment of the present invention. This method includes, but is not limited to, the following steps:

[0052] S10, construct a digital twin model based on the physical data of each property equipment, and determine the first instance. Each property equipment corresponds to a model instance in the digital twin model. The first instance is a model instance with preset trigger conditions. The trigger conditions are used to indicate changes in human body signals or preset reference operating parameters.

[0053] It should be noted that building digital twin models based on the physical data of property equipment is an existing technology, and the specific construction method will not be elaborated here. In this embodiment, each property equipment corresponds to a model instance, enabling the model instance to simulate the actual operation of the property equipment.

[0054] It should be noted that in this embodiment, trigger conditions are pre-set for at least one property device. The property device with the set trigger conditions is marked as the first instance in the digital twin model. The trigger conditions characterize the switching requirements of the operating state. The trigger conditions can be associated with the trigger signals of the property device, such as changes in human body signals of devices like lighting equipment, air conditioners, and elevators, or whether the reference operating parameter of the electricity demand of photovoltaic equipment exceeds the switching threshold.

[0055] For example, such as Figure 1As shown, the property equipment includes equipment 1-1 (lighting equipment), equipment 1-2 (air conditioner), equipment 1-3 (escalator), and equipment 2-1 (photovoltaic equipment). Based on the physical data of each property equipment, digital instances are converted, resulting in instance 1-1 corresponding to equipment 1-1, instance 1-2 corresponding to equipment 1-2, instance 1-3 corresponding to equipment 1-3, and instance 2-1 corresponding to equipment 2-1. The lighting equipment can detect whether someone is passing by through infrared. In this embodiment, this trigger signal can be used as a human body signal, thereby identifying equipment 1-1 as the first instance. Similarly, equipment 1-3 can detect whether someone is passing by and can also generate a human body signal, thus identifying it as the first instance. Equipment 2-1 can switch from power supply to grid power supply according to the power demand of the community. In this embodiment, the community power consumption is used as a reference operating parameter. When the community power consumption is lower than the threshold, the switch is triggered, so equipment 2-1 can also be identified as the first instance.

[0056] S20: Based on the pre-trained LSTM model, predict the energy consumption of each model instance on the target date, and determine the control information set based on the energy consumption prediction. The control information set includes multiple control strategies, each of which includes a first time period and a first operating state. The first time periods of each control strategy do not overlap.

[0057] It should be noted that the specific model structure and prediction method of the LSTM model are well-known to those skilled in the art and will not be elaborated upon here. This embodiment uses the LSTM model to predict the energy consumption forecast value of each instance model on the target date. The target date can be the day after the current date, that is, using the device data that can be obtained before the current date as input, and using the LSTM model to predict the energy consumption forecast value for the next day. This embodiment uses the time period of operation status switching in historical data for energy consumption prediction. The digital twin model can detect the change in the operation status of each property equipment, thereby constructing multiple first time periods. When the operation status of the property equipment changes, the corresponding start time can be recorded in the digital twin model, and the end time can be recorded when the operation status changes again. The time period composed of the start time and the end time is used as the first time period of this embodiment.

[0058] For example, with Figure 1Taking devices 1-1 and 2-1 as examples, in the current date, the time points of each switch-on or brightness adjustment of device 1-1 are recorded through a digital twin model, thereby constructing time period 1-1, time period 1-2, etc. For example, time period 1-1 is the period of time after device 1-1 is adjusted from the off state to the high-efficiency lighting state (high brightness) and maintains the high-efficiency lighting state, and time period 1-2 is the period of time after device 1-1 switches from the high-efficiency lighting state to the energy-saving lighting state (low brightness) and maintains the energy-saving lighting state. At the same time, the time period 4-1 when device 2-1 supplies power to the community and the time period 4-2 when it switches to supply power to the grid after the community's power consumption decreases are recorded.

[0059] It should be noted that the predicted energy consumption value in this embodiment can be a sequence of multiple predicted values, with each predicted value corresponding to a first time period. The property management system can determine the required first operating state based on the predicted value of each first time period, so that the control strategy records the first time period and the first operating state. The first operating state can be determined based on the control requirements of the first time period.

[0060] For example, with Figure 1 Taking device 1-2 as an example, the first time period includes time periods 2-1, 2-2, and 2-3. During time period 2-1, the user is in the space corresponding to device 1-2, and the energy consumption prediction value is high. Therefore, it can be determined that the current time period is when device 1-2 has just been started and the temperature needs to be adjusted in a short time. Thus, the first operating state can be determined as the start-up state and operated at a high power. During time period 2-2, the user is still in the space corresponding to device 1-2, and the energy consumption prediction value is still high. It can be determined that the temperature adjustment has been completed, and it is only necessary to maintain the current room temperature. In order to save energy, the first operating state can be determined as the start-up state and operated at a low power. During time period 2-3, the user leaves the corresponding space, and the energy consumption prediction value is low. The first operating state can be determined as the off state, or the start-up state can be maintained and the set temperature adjusted to reduce the power consumption of device 1-2.

[0061] S30, based on any first instance, determine at least one pre-associated second instance, generate a switching strategy and a second operating state based on trigger conditions in each control strategy corresponding to the first instance and the second instance, wherein the second operating state is the opposite state of the first operating state, and the property equipment corresponding to the second instance and the first instance are pre-configured with an association relationship.

[0062] It should be noted that the property equipment corresponding to the first instance has preset trigger conditions, and some property equipment is in the same space, so the control can be linked. In this embodiment, the association relationship of multiple property equipment can be set in advance in the property management system. When the model instance corresponding to the property equipment is determined as the first instance, the model instances that are related but not determined as the first instance are determined as the second instance. Thus, the same switching strategy as the first instance is introduced into the second instance. So when the property equipment corresponding to the second instance cannot determine whether it needs to switch the operating state, the trigger condition of the first instance is used as the basis, which effectively improves the energy utilization rate.

[0063] It should be noted that the switching strategy is used to indicate the need for changes in the operating status. For example, the space corresponding to the property equipment may change from occupied to unoccupied, or from unoccupied to occupied. Or, the electricity consumption of the community may be lower than the power generation of the photovoltaic equipment, or the electricity consumption of the community may be greater than the power generation of the photovoltaic equipment. Based on this, the second operating status is the opposite of the first operating status. If the first operating status is on, the second operating status may be off. If the first operating status is supplying power to the community, the second operating status is supplying power to the grid, and so on.

[0064] For example, such as Figure 1 As shown, the control strategy for each model instance includes four pieces of information: the first time period, the first running state, the switching strategy, and the second running state. For example... Figure 1 Examples 1-3 in the example are the first examples, so the control strategy can be determined to include the first time period 3, the running state 3, the switching strategy 1 determined based on the trigger condition 1, and the running state 2.

[0065] S40 controls the corresponding property equipment based on any model instance and the set of control information.

[0066] It should be noted that after determining the control strategy for each model instance, this embodiment directly controls each property device based on the control information set on the target date, enabling the property devices to operate under reasonable energy consumption control, effectively reducing energy consumption and improving energy management efficiency. After each model instance determines the control signal based on the control strategy, it is remotely sent to the corresponding property device to complete the control.

[0067] In another embodiment, step S30 specifically includes, but is not limited to, the following steps:

[0068] S31, when the first operating state corresponding to the control strategy is used to indicate energy saving, the switching strategy is used to indicate that the property equipment corresponding to the target instance detects a human signal or the value of the reference operating parameter is greater than a preset threshold, wherein the target instance is the first instance and / or the associated second instance;

[0069] S32, when the first operating state corresponding to the control strategy is used to indicate energy consumption, the switching strategy is used to indicate that the property equipment corresponding to the target instance loses human body signal or the value of the reference operating parameter is less than or equal to a preset threshold.

[0070] It should be noted that, when determining the switching strategy, this embodiment determines the actual state represented by the first operating state, and thus adopts different judgment conditions as the switching strategy under different states.

[0071] It should be noted that energy saving in this embodiment can be achieved by turning off the corresponding property equipment or switching to a low-power mode such as energy-saving mode, for example, reducing the brightness of the lighting system or increasing the temperature of the air conditioner when cooling. When the first operating state is used to indicate energy saving, its opposite state is to consume energy in exchange for a better user experience. Therefore, the switching strategy can be to detect a human signal or a reference operating parameter value that is greater than a preset threshold. For example, the lighting system detects someone entering when the brightness is low, or the photovoltaic equipment detects that the reference operating parameter of the community's electricity consumption is greater than a preset threshold when it is transmitting power to the grid.

[0072] It should be noted that the energy consumption in this embodiment may be achieved by setting property equipment to operate at high power to improve user experience. For example, air conditioners may operate at a lower target temperature to achieve rapid cooling, lighting systems may use high brightness to improve lighting effects, or photovoltaic equipment may supply electricity to the community to provide electricity for the community. The switching strategy may be based on the loss of human body signal or the value of a reference operating parameter being less than or equal to a preset threshold. For example, a user leaving the space of the property equipment causes the property equipment to lose its human body signal, or the reference operating parameter of the community's electricity consumption is less than or equal to a threshold.

[0073] In another embodiment, before performing step S20, the following steps are included, but are not limited to:

[0074] S211, Obtain the historical energy consumption values ​​of each property equipment, and build a model training set based on all historical energy consumption values, where each historical energy consumption value corresponds to a historical time period;

[0075] S212, Obtain the preset LSTM model, wherein the LSTM model is a two-layer bidirectional model, the first layer of the LSTM model is the output retention layer, and the second layer of the LSTM model is the sequence compression layer;

[0076] S213, The model training set is input into the retained output layer and the compressed sequence layer respectively for model training. The input data of the retained output layer includes the model training set and the compressed feature sequence output by the compressed sequence layer based on the model training set.

[0077] S214, construct multiple instance groups, and configure the trained LSTM model to each instance group respectively. When an instance group includes multiple model instances, the property equipment corresponding to each model instance has a pre-defined relationship.

[0078] It should be noted that in order to train the LSTM model, this embodiment can obtain the historical energy consumption values ​​of the property equipment after constructing the digital twin model, and use the historical energy consumption values ​​as the model training set to initially train the LSTM model, for example, training for 50 rounds to enable the LSTM model to have preliminary prediction capabilities.

[0079] It should be noted that the LSTM model in this embodiment is a two-layer bidirectional mode. The first layer is the output retention layer, and the second layer is the compressed sequence layer. The feature sequence is extracted from the model training set data by the compressed sequence layer and then compressed. The resulting compressed feature sequence is then fed back into the output retention layer for training. The two-layer bidirectional LSTM model has stronger modeling capabilities. Furthermore, the number of neurons in the output retention layer in this embodiment is twice that of the compressed sequence layer, enabling the deep structure of the output retention layer to learn higher-order features.

[0080] It should be noted that after the initial training of the LSTM model is completed, this embodiment will construct an instance group from multiple model instances that have pre-defined relationships, for example... Figure 1 As shown, devices 1-1, 1-2, and 1-3 are pre-associated, meaning that instances 1-1, 1-2, and 1-3 belong to the same instance group. A separate LSTM model is configured for each instance group, enabling the LSTM model to make accurate energy consumption predictions based on each instance group.

[0081] In another embodiment, step S20 specifically includes, but is not limited to, the following steps:

[0082] S221, based on any model instance, based on the LSTM model to predict multiple first time periods based on historical time periods, and based on the energy consumption prediction values ​​of each first time period based on historical energy consumption values.

[0083] S222, based on any first time period, obtain the historical operating status corresponding to each historical energy consumption value;

[0084] S223, when the predicted energy consumption value is less than or equal to the historical energy consumption value and the preset energy consumption threshold, and the historical operating status is used to indicate energy consumption, the first operating status is determined to indicate energy saving.

[0085] S224, when the predicted energy consumption value is greater than the historical energy consumption value and the energy consumption threshold, and the historical operating status is used to indicate energy saving, the first operating status is determined to indicate energy consumption.

[0086] S225, the historical operating status is determined as the first operating status.

[0087] It should be noted that, in determining the first time period in this embodiment, according to the description of the above embodiment, different historical time periods can be determined based on the changes in the operating status of the property equipment. This embodiment does not directly use the historical time periods, but uses the historical time periods as the prediction object to predict multiple first time periods, so that the first time period can be predicted by combining the usage status of the property equipment over multiple days, resulting in higher accuracy.

[0088] It should be noted that after determining the first time period, the historical operating status can be obtained from the historical energy consumption value, such as the energy-saving status and high-efficiency lighting status described in the above embodiment. This embodiment has a preset energy consumption threshold. When the energy consumption threshold is higher, the energy-consuming mode needs to be executed; otherwise, the energy-saving mode is executed. The energy consumption threshold can be electricity consumption, water consumption, gas consumption, etc., or it can be the community's electricity consumption configured for photovoltaic equipment.

[0089] It should be noted that in steps S223 to S225, this embodiment takes the first time period of a model instance as an example, and the first running state can be determined in the same way for each time period.

[0090] It should be noted that in step S223, when the predicted energy consumption value is less than or equal to the historical energy consumption value and the energy consumption threshold, and the historical operating state is consuming energy, the energy consumption threshold can be used to determine that a change in the operating state needs to be made. If the predicted energy consumption value is less than the historical energy consumption value, then the energy-saving operating state is taken as the first operating state.

[0091] It should be noted that in step S224, the opposite of step S223, if the historical operating state is energy saving, then the energy consumption prediction value is greater than the energy consumption threshold to determine that a state change is required, and the first operating state is determined to be energy consuming.

[0092] It should be noted that, except for the cases in steps S223 and S224, the remaining cases can be determined to be that the predicted energy consumption value and the historical energy consumption value are both above or below the energy consumption threshold, so the historical operating status can be directly used.

[0093] In another embodiment, step S20 specifically includes, but is not limited to, the following steps:

[0094] S231, based on a preset target time, obtain the energy consumption data of each property equipment on the current date, wherein the energy consumption data is used to indicate the energy consumption data of the second time period under different operating modes, and the different second time periods of the same property equipment do not overlap.

[0095] S232, Based on any instance group, incrementally train the assigned LSTM model using the corresponding device energy consumption data;

[0096] S233, based on the incrementally trained LSTM model, predicts the set of control information for each model instance on the target date, where the target date is the day after the current date.

[0097] It should be noted that in this embodiment, when making predictions, the energy consumption data of each property device on the current date is obtained, and the running time period of the energy consumption data is used as the second time period. The first time period is the running time period predicted by the LSTM model, and the second time period is the actual running time period of the property device in different states on the current date. The first time period may overlap with the second time period, or they may not overlap. For example, if the first running state predicted in the first time period is energy saving, and a user enters during this process, triggering the switching strategy, the property device will switch to the second running state. This makes the actual running time of the first running state less than the predicted first time period. Therefore, in this embodiment, the actual running time period is used as the second time period, which is obtained statistically based on the digital twin model during actual operation.

[0098] It should be noted that, according to the above embodiment, each instance group is assigned an LSTM model. This embodiment further performs incremental training on the LSTM model within each instance group once a day, using device energy consumption data from the current date. This incremental training allows the LSTM model of each instance group to gradually approach the actual operating conditions of the space corresponding to that instance group. For example… Figure 1 The example group shown consists of Example 1-1, Example 1-2, and Example 1-3, which corresponds to the shop area. With daily incremental training, the control strategies of each property equipment become more dependent on the actual operation of the shop area, thereby improving the prediction accuracy of the LSTM model.

[0099] In another embodiment, before performing step S28, the following steps are included, but are not limited to:

[0100] S271, arrange the second time period corresponding to each model instance in chronological order, form a time period group based on multiple second time periods in the same order, and traverse the queue formed by the time period groups in sequence.

[0101] S272, when the operating status corresponding to the time period group is used to indicate energy saving, the earliest recorded start time is determined as the target start time of each second time period, and the latest recorded end time is determined as the target end time of each second time period. Wherein, when the target start time determined this time is earlier than the target end time of the previous time period group, the target end time of the previous time period group is modified based on the target start time determined this time.

[0102] S273, when the operating status corresponding to the traversed time period group is used to indicate energy consumption, the latest recorded start time is determined as the target start time of each second time period, and the earliest recorded end time is determined as the target end time of each second time period. Wherein, when the modified target start time is later than the target end time of the previous time period group, the target end time of the previous time period group is determined based on the target start time.

[0103] It should be noted that after determining the second time period for each model instance, the control logic of each property device within the same instance group is the same, such as switching to energy-saving mode or high-efficiency mode at the same time. Therefore, this embodiment needs to align the second time period for each model instance to ensure that the predicted first time period is the same.

[0104] It should be noted that in this embodiment, multiple second time periods in the same order are grouped into the same time period group. That is, the first second time period of multiple model instances is a time period group, or the second second time period of multiple model instances is a time period group, and so on. The alignment operation is performed by traversing each time period group.

[0105] It should be noted that since the equipment energy consumption data is for the current day, the states corresponding to each second time period are known. Since the sequence numbers of the second time periods in the same time period group are the same, the corresponding operating states are also the same. When the operating state of the time period group is energy saving.

[0106] It should be noted that in step S272, in order to reduce energy consumption, this embodiment compares values ​​within multiple second time periods, using the earliest start time as the target start time and the latest end time as the target end time, making the synchronized second time period the longest. Furthermore, since the property equipment corresponding to the target end time is still in an energy-saving state, other property equipment also has the need to be in an energy-saving state. Adjusting the second time period will not cause the property equipment to fail to meet actual needs. It is worth noting that, because the target start time is advanced, if it is earlier than the target end time of the previous time period group, since the second time period is determined based on state changes, the operating state corresponding to the previous time period group is energy-consuming. Therefore, this embodiment corrects the target end time of the previous time period group based on the current target end time, further increasing the running time of the energy-saving mode.

[0107] It should be noted that in step S273, when the operating state corresponding to the time period group is energy consumption, the second time period needs to be shortened as much as possible to reduce energy consumption. Therefore, the latest recorded start time is determined as the target start time of each second time period, and the earliest recorded end time is determined as the target end time of each second time period, so that the property equipment can reduce the operating time of energy consumption. Similarly, since the latest start time is used to determine the target start time, at least one start time of the second time period is delayed, resulting in a gap time between it and the next second time period. Therefore, this embodiment corrects the target end time of the previous time period group based on the target start time of the current time period group, so that the second time period of the previous time period group is extended, and the energy-saving mode can be operated for a longer period of time.

[0108] In another embodiment, before performing step S28, the following steps are included, but are not limited to:

[0109] S274, Based on any model instance, determine the segmented energy consumption data for each second time period;

[0110] S275, based on any adjusted second time period, when the corresponding operating status is used to indicate energy saving, interpolate the newly added time period based on the segmented energy consumption data with the smallest value, or when the corresponding operating status is used to indicate energy consumption, interpolate the newly added time period based on the average value of the segmented energy consumption data.

[0111] It should be noted that in this embodiment, if the second time period is extended after the adjustment of the second time period is completed, the energy consumption data will be unusable because the operating state corresponding to the energy consumption data of the previous second time period is different from that of the current second time period. For example, the current second time period corresponds to the energy saving mode, while the previous one corresponds to the high-efficiency mode. If the energy consumption data of the high-efficiency mode is used for prediction, the prediction will be inaccurate. Therefore, in this embodiment, after extending the second time period, the energy consumption data of the extended time period is deleted, and the data is increased by interpolation of segmented energy consumption data to ensure the accuracy of the prediction.

[0112] like Figure 3 As shown, Figure 3 This is a structural diagram of a property equipment control device based on digital twins according to an embodiment of the present invention. The present invention also provides a property equipment control device based on digital twins, comprising:

[0113] The processor 401 can be implemented using a general-purpose central processing unit (CPU), microprocessor, application specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0114] The memory 402 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 402 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 402 and is called and executed by the processor 401 to execute the digital twin-based property equipment control method of the embodiments of this application.

[0115] Input / output interface 403 is used to implement information input and output;

[0116] The communication interface 404 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0117] Bus 405 transmits information between various components of the device (e.g., processor 401, memory 402, input / output interface 403, and communication interface 404);

[0118] The processor 401, memory 402, input / output interface 403 and communication interface 404 are connected to each other within the device via bus 405.

[0119] This application also provides an electronic device, including the property equipment control device based on digital twin as described above.

[0120] This application embodiment also provides a storage medium, which is a computer-readable storage medium, storing a computer program that, when executed by a processor, implements the above-described property equipment control method based on digital twins.

[0121] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof. The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separate, and may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0122] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically include computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0123] The above provides a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of the present invention.

Claims

1. A property equipment control method based on digital twin, characterized in that, Applied to a property management system, which includes multiple property devices, the method includes: A digital twin model is constructed based on the physical data of each of the property devices, and a first instance is determined. Each of the property devices corresponds to a model instance in the digital twin model. The first instance is the model instance with preset trigger conditions, which are used to indicate changes in human body signals or preset reference operating parameters. Based on the pre-trained LSTM model, the energy consumption prediction value of each model instance on the target date is predicted, and a set of control information is determined based on the energy consumption prediction value. The set of control information includes multiple control strategies, and each control strategy includes a first time period and a first operating state. The first time periods of each control strategy do not overlap. Based on any of the first instances, at least one pre-associated second instance is determined. In each of the control strategies corresponding to the first instance and the second instance, a switching strategy and a second operating state are generated based on the triggering conditions. The second operating state is the opposite state of the first operating state. The second instance is pre-configured with the property equipment corresponding to each of the first instances. The second instance is a model instance that has not been determined as the first instance. Based on any of the model instances, the corresponding property equipment is controlled based on the set of control information.

2. The property equipment control method based on digital twin according to claim 1, characterized in that, In each of the corresponding control strategies, a switching strategy and a second operating state are generated based on the triggering conditions, including: When the first operating state corresponding to the control strategy is used to indicate energy saving, the switching strategy is used to indicate that the property equipment corresponding to the target instance detects that the value of the human signal or the reference operating parameter is greater than a preset threshold, wherein the target instance is the first instance and / or the associated second instance; Alternatively, when the first operating state corresponding to the control strategy is used to indicate energy consumption, the switching strategy is used to indicate that the property equipment corresponding to the target instance loses the human signal or the value of the reference operating parameter is less than or equal to a preset threshold.

3. The property equipment control method based on digital twin according to claim 1, characterized in that, Before predicting the energy consumption of each model instance on the target date based on the pre-trained LSTM model, the method further includes: Obtain the historical energy consumption values ​​of each of the aforementioned property equipment, and construct a model training set based on all the historical energy consumption values, wherein each of the historical energy consumption values ​​corresponds to a historical time period; Obtain the preset LSTM model, wherein the LSTM model is a two-layer bidirectional model, the first layer of the LSTM model is an output retention layer, and the second layer of the LSTM model is a sequence compression layer; The model training set is input into the retained output layer and the compressed sequence layer respectively for model training, wherein the input data of the retained output layer includes the model training set and the compressed feature sequence output by the compressed sequence layer based on the model training set; Multiple instance groups are constructed, and the trained LSTM model is configured into each instance group. When an instance group includes multiple model instances, the property equipment corresponding to each model instance is pre-associated.

4. The property equipment control method based on digital twin according to claim 3, characterized in that, Based on a pre-trained LSTM model, the energy consumption prediction value for each model instance on the target date is calculated, and based on the energy consumption prediction value, a set of control information is determined, including: Based on any of the model instances, based on the multiple first time periods predicted by the LSTM model based on the historical time period, and the energy consumption prediction values ​​of each of the first time periods predicted based on the historical energy consumption values; Based on any of the first time periods, obtain the historical operating status corresponding to each of the historical energy consumption values; When the predicted energy consumption value is less than or equal to the historical energy consumption value and the preset energy consumption threshold, and the historical operating status is used to indicate energy consumption, the first operating status is determined to indicate energy saving. Alternatively, when the predicted energy consumption value is greater than the historical energy consumption value and the energy consumption threshold, and the historical operating status is used to indicate energy saving, the first operating status is determined to indicate energy consumption. Alternatively, the historical operating state can be determined as the first operating state.

5. The property equipment control method based on digital twin according to claim 3, characterized in that, The determination of the control information set based on the energy consumption prediction value includes: Based on a preset target time, the energy consumption data of each property device on the current date is obtained, wherein the energy consumption data is used to indicate the energy consumption data of the second time period under different operating modes, and the second time periods of the same property device do not overlap with each other; Based on any of the instance groups, the assigned LSTM model is incrementally trained using the corresponding device energy consumption data; Based on the incrementally trained LSTM model, the control information set of each model instance on the target date is predicted, wherein the target date is the day after the current date.

6. The property equipment control method based on digital twin according to claim 5, characterized in that, Before incrementally training the assigned LSTM model using the corresponding device energy consumption data, the method further includes: Arrange the second time period corresponding to each model instance in chronological order, form a time period group based on multiple second time periods in the same order, and traverse the queue formed by the time period groups in sequence. When the operating status corresponding to the time period group is used to indicate energy saving, the earliest recorded start time is determined as the target start time of each second time period, and the latest recorded end time is determined as the target end time of each second time period. Wherein, when the target start time determined this time is earlier than the target end time of the previous time period group, the target end time of the previous time period group is modified based on the target start time determined this time. Alternatively, when the operating status corresponding to the traversed time period group is used to indicate energy consumption, the latest recorded start time is determined as the target start time of each second time period, and the earliest recorded end time is determined as the target end time of each second time period. Wherein, when the modified target start time is later than the target end time of the previous time period group, the target start time is determined as the target end time of the previous time period group.

7. The property equipment control method based on digital twin according to claim 5, characterized in that, Before incrementally training the assigned LSTM model using the corresponding device energy consumption data, the method further includes: Based on any of the model instances, segmented energy consumption data for each of the second time periods are determined; Based on any adjusted second time period, when the corresponding operating state is used to indicate energy saving, the newly added time period is interpolated based on the segmented energy consumption data with the smallest value; or, when the corresponding operating state is used to indicate energy consumption, the newly added time period is interpolated based on the average value of the segmented energy consumption data.

8. A property equipment control device based on digital twin, characterized in that, It includes at least one control processor and a memory for communicatively connecting to the at least one control processor; the memory stores instructions executable by the at least one control processor to enable the at least one control processor to perform the digital twin-based property equipment control method as described in any one of claims 1 to 7.

9. An electronic device, characterized in that, Includes the property equipment control device based on digital twin as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions for causing a computer to perform the property equipment control method based on digital twins as described in any one of claims 1 to 7.

Images