Air conditioner load regulation and control method and device and electronic equipment

Abstract

The invention provides an air conditioner load regulation and control method and device and electronic equipment, and belongs to the technical field of load regulation and control. The method comprises the steps of obtaining an environmental parameter regulation and control instruction; the demand environment parameters, the current air conditioner load parameters and the historical air conditioner load parameters are input into an air conditioner temperature adjusting value prediction model, and an air conditioner temperature adjusting value prediction result output by the air conditioner temperature adjusting value prediction model is obtained; the air conditioner temperature adjusting value prediction result, the working scene type and the meteorological parameters are input into an air conditioner temperature adjusting range prediction model, and an air conditioner temperature adjusting range prediction result of the air conditioner temperature adjusting range prediction model is obtained; and adjusting the temperature of the air conditioner based on the air conditioner temperature adjusting range prediction result to realize air conditioner load regulation and control. The method is used for solving the problem that the method only depends on a single influence factor to predict the air conditioner temperature adjusting value, so that the prediction precision of the temperature adjusting value is low, and the air conditioner load regulation and control effect is not ideal.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of load regulation, in particular to an air conditioner load regulation method, an air conditioner load regulation device and an electronic device. BACKGROUND

[0002] By regulating the load of air conditioners in industrial and commercial scenarios, not only can the pressure on the power grid be relieved and energy costs be reduced, but also the efficiency of the energy system can be improved, which is a key means.

[0003] When the existing method regulates the load of air conditioners, an artificial intelligence model is often trained based on historical load to learn the characteristics of load changes from historical load to predict the future temperature adjustment value of the air conditioner. Then the temperature of the air conditioner is adjusted to the predicted temperature adjustment value to achieve air conditioner load regulation. However, such a method only relies on a single influencing factor to predict the temperature adjustment value of the air conditioner, resulting in low prediction accuracy of the temperature adjustment value, which in turn leads to unsatisfactory air conditioner load regulation effect. SUMMARY

[0004] The purpose of the embodiments of the present application is to provide an air conditioner load regulation method, device and electronic device to solve the problem that the method only relies on a single influencing factor to predict the temperature adjustment value of the air conditioner, resulting in low prediction accuracy of the temperature adjustment value, which in turn leads to unsatisfactory air conditioner load regulation effect.

[0005] To achieve the above purpose, the embodiments of the present application provide an air conditioner load regulation method, comprising: Obtain an environment parameter regulation instruction, wherein the environment parameter regulation instruction carries a demand environment parameter of a user; Input the demand environment parameter, the current load parameter of the air conditioner and the historical load parameter of the air conditioner into an air conditioner temperature adjustment value prediction model to obtain an air conditioner temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model; Input the air conditioner temperature adjustment value prediction result, the work scene type and the meteorological parameter into an air conditioner temperature adjustment range prediction model to obtain an air conditioner temperature adjustment range prediction result of the air conditioner temperature adjustment range prediction model; Adjust the temperature of the air conditioner based on the air conditioner temperature adjustment range prediction result to achieve air conditioner load regulation; The air conditioner temperature adjustment value prediction model is constructed based on a linear model, and the air conditioner temperature adjustment range prediction model is constructed based on a nonlinear model.

[0006] Optionally, the air conditioner temperature adjustment value prediction model is constructed by selecting a multiple linear regression model; and the air conditioner temperature adjustment value prediction model is trained by the following steps: obtaining a plurality of data sets, the data sets comprising sample demand environment parameters, sample air conditioner current load parameters and sample air conditioner historical load parameters; repeating the following steps until a set stop condition is reached: inputting a first data set into the air conditioner temperature adjustment value prediction model to obtain a sample temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model; the first data set is any one of the plurality of data sets; calculating a loss function based on the sample temperature adjustment value prediction result and a true temperature adjustment value label corresponding to the first data set; adjusting a weight coefficient of the air conditioner temperature adjustment value prediction model based on the loss function.

[0007] Optionally, the air conditioner temperature adjustment range prediction model is constructed using a deep learning model; the air conditioner temperature adjustment range prediction model is trained through the following steps: obtaining a plurality of data sets, the data sets comprising sample air conditioner temperature adjustment value prediction results, sample working scenario types and sample meteorological parameters; repeating the following steps until a set stop condition is reached: inputting a first data set into the air conditioner temperature adjustment range prediction model to obtain a sample air conditioner temperature adjustment range prediction result output by the air conditioner temperature adjustment range prediction model; the first data set is any one of the plurality of data sets; calculating a loss function based on the sample air conditioner temperature adjustment range prediction result and a true air conditioner temperature adjustment range label corresponding to the first data set; adjusting network parameters of the air conditioner temperature adjustment range prediction model based on the loss function.

[0008] Optionally, after the temperature adjustment of the air conditioner based on the air conditioner temperature adjustment range prediction result to achieve air conditioner load regulation and control, the method further comprises: obtaining a load regulation target value; determining an average load regulation target value of each branch node of the air conditioner based on the load regulation target value; repeating the following steps until a first deviation and a total node power deviation of all branch nodes meet a set condition: calculating candidate load adjustment amounts of all branch nodes of the air conditioner based on an initial load adjustment amount using a gradient descent method; adjusting the load of all branch nodes based on the candidate load adjustment amounts; calculating the first deviation and the total node power deviation of all branch nodes; The first deviation represents a difference between a load balance coefficient of all branch nodes after load adjustment and a historical load balance coefficient before load adjustment; the load balance coefficient represents a deviation degree of a load of each branch node from an average load of all branch nodes of the air conditioner; the total node power deviation represents a sum of node power deviations of all branch nodes after load adjustment; the node power deviation represents a deviation between the candidate load adjustment amount and a load regulation gap value; the load regulation gap value is calculated based on a deviation between a current power value of the branch node and an average load regulation target value; and the candidate load adjustment amount after iteration is taken as a load regulation strategy value of all branch nodes.

[0009] Optionally, the load balance coefficient is calculated by the following steps: calculating a load rate of all branch nodes; the load rate is calculated based on an apparent power of the branch node and a load capacity; calculating an average load rate based on the load rate of all branch nodes; calculating a load balance coefficient of all branch nodes based on the load rate of all branch nodes and the average load rate.

[0010] Optionally, the method further comprises: calculating a regulation coefficient based on the load regulation target value and the load regulation strategy value; the load regulation target value and the regulation coefficient are in a positive correlation relationship; and the load regulation strategy value and the regulation coefficient are in a negative correlation relationship; calculating a regulation evaluation score based on the regulation coefficient and a set evaluation coefficient; the regulation coefficient and the regulation evaluation score are in a positive correlation relationship; and the set evaluation coefficient and the regulation evaluation score are in a negative correlation relationship; determining a regulation effect based on the regulation evaluation score.

[0011] In another aspect, the embodiment of the present application further provides an air conditioner load regulation device, comprising: an acquisition module configured to acquire an environment parameter regulation instruction, wherein the environment parameter regulation instruction carries a demand environment parameter of a user; a first prediction module configured to input the demand environment parameter, a current load parameter of the air conditioner and a historical load parameter of the air conditioner into an air conditioner temperature adjustment value prediction model to obtain an air conditioner temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model; a second prediction module configured to input the air conditioner temperature adjustment value prediction result, a working scene type and a meteorological parameter into an air conditioner temperature adjustment range prediction model to obtain an air conditioner temperature adjustment range prediction result of the air conditioner temperature adjustment range prediction model; An adjusting module is configured to adjust the temperature of the air conditioner based on the air conditioner temperature adjustment range prediction result to achieve air conditioner load regulation. The air conditioner temperature adjustment value prediction model is constructed based on a linear model, and the air conditioner temperature adjustment range prediction model is constructed based on a nonlinear model.

[0012] Optionally, the air conditioner temperature adjustment value prediction model is constructed based on a multiple linear regression model, and the air conditioner temperature adjustment value prediction model is trained by the following steps: A plurality of data sets are obtained, and the data sets include sample demand environment parameters, sample air conditioner current load parameters and sample air conditioner historical load parameters. The following steps are repeatedly performed until a set stop condition is reached: The first data set is input into the air conditioner temperature adjustment value prediction model to obtain a sample temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model; the first data set is any one of the plurality of data sets. A loss function is calculated based on the sample temperature adjustment value prediction result and a real temperature adjustment value label corresponding to the first data set. The weight coefficient of the air conditioner temperature adjustment value prediction model is adjusted based on the loss function.

[0013] Optionally, the air conditioner temperature adjustment range prediction model is constructed based on a deep learning model, and the air conditioner temperature adjustment range prediction model is trained by the following steps: A plurality of data sets are obtained, and the data sets include sample air conditioner temperature adjustment value prediction results, sample working scenario types and sample meteorological parameters. The following steps are repeatedly performed until a set stop condition is reached: The first data set is input into the air conditioner temperature adjustment range prediction model to obtain a sample air conditioner temperature adjustment range prediction result output by the air conditioner temperature adjustment range prediction model; the first data set is any one of the plurality of data sets. A loss function is calculated based on the sample air conditioner temperature adjustment range prediction result and a real air conditioner temperature adjustment range label corresponding to the first data set. The network parameters of the air conditioner temperature adjustment range prediction model are adjusted based on the loss function.

[0014] Optionally, the device further includes: The computing module is configured to obtain a load regulation target value, determine an average load regulation target value of each branch node of the air conditioner based on the load regulation target value, and repeatedly perform the following steps until a first deviation and a total node power deviation of the nodes meet a set condition: calculate candidate load adjustment amounts of all branch nodes of the air conditioner based on an initial load adjustment amount by using a gradient descent method; perform load adjustment on all branch nodes based on the candidate load adjustment amounts; and calculate the first deviation and the total node power deviation of the nodes. The first deviation represents a difference between a load balance coefficient of all branch nodes after load adjustment and a historical load balance coefficient before load adjustment; the load balance coefficient represents a deviation degree of the load of each branch node from an average load of all branch nodes of the air conditioner; the total node power deviation represents a sum of node power deviations of all branch nodes after load adjustment; the node power deviation represents a deviation between the candidate load adjustment amount and a load regulation gap value; and the load regulation gap value is calculated based on a deviation between a current power value of the branch node and an average load regulation target value.

[0015] Optionally, the load balance coefficient is calculated by the following steps: Calculate a load rate of all branch nodes; the load rate is calculated based on an apparent power of the branch node and a load capacity; Calculate an average load rate based on the load rates of all branch nodes; Calculate the load balance coefficient of all branch nodes based on the load rates of all branch nodes and the average load rate.

[0016] Optionally, the device further comprises: A regulation effect determination module is configured to calculate a regulation coefficient based on the load regulation target value and the load regulation strategy value; the load regulation target value and the regulation coefficient are in a positive correlation relationship; the load regulation strategy value and the regulation coefficient are in a negative correlation relationship; calculate a regulation evaluation score based on the regulation coefficient and a set evaluation coefficient; the regulation coefficient and the regulation evaluation score are in a positive correlation relationship; the set evaluation coefficient and the regulation evaluation score are in a negative correlation relationship; and determine a regulation effect based on the regulation evaluation score.

[0017] In another aspect, the present application further provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor implements the air conditioner load regulation method when executing the program.

[0018] In another aspect, the present application also provides a machine readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the air conditioner load regulation method.

[0019] In another aspect, the present application also provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the air conditioner load regulation method.

[0020] Through the above technical solution, the air conditioner temperature adjustment range is predicted by combining the linear model and the nonlinear model, considering various influencing factors such as demand environment parameters, current air conditioner load parameters, historical air conditioner load parameters, working scene types, and meteorological parameters, and the air conditioner is temperature adjusted based on the air conditioner temperature adjustment range prediction result, so as to realize air conditioner load regulation. The air conditioner temperature adjustment range is predicted by comprehensively considering multiple models and multiple influencing factors, so as to improve the prediction accuracy of the air conditioner temperature adjustment value, and further optimize the air conditioner load regulation effect.

[0021] Other features and advantages of the present application will be described in detail in the following specific implementation part. BRIEF DESCRIPTION OF DRAWINGS

[0022] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, and are used together with the following specific implementation to explain the present application, but do not constitute a limitation on the present application. In the drawings: Figure 1 is one of the flowcharts of the air conditioner load regulation method provided by the present application; Figure 2 is the second flowchart of the air conditioner load regulation method provided by the present application; Figure 3 is a schematic diagram of the load model library provided by the present application; Figure 4 is the third flowchart of the air conditioner load regulation method provided by the present application; Figure 5 is the fourth flowchart of the air conditioner load regulation method provided by the present application; Figure 6 is a schematic diagram of the regulation strategy component provided by the present application; Figure 7 is the fifth flowchart of the air conditioner load regulation method provided by the present application; Figure 8 is a structural schematic diagram of the air conditioner load regulation device provided by the present application; Figure 9 is a structural schematic diagram of the electronic device provided by the present application. DETAILED DESCRIPTION

[0023] The specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are merely intended to illustrate and explain the embodiments of the present application, and are not intended to limit the embodiments of the present application.

[0024] Method embodiments Please refer to Figure 1 The embodiments of the present application provide an air conditioner load regulation method, comprising: Step 100, obtaining an environment parameter regulation instruction, wherein the environment parameter regulation instruction carries a user's demand environment parameter.

[0025] An electronic device (for example, which can be a load management device) obtains an environment parameter regulation instruction, wherein the environment parameter regulation instruction carries a user's demand environment parameter. The demand environment parameter can include a user's work demand temperature or temperature range, work demand humidity or humidity range. For example, in a commercial scenario, for example, in a large commercial supermarket, different stores have different work demand temperatures or work demand humidities, so the demand environment parameter can include a user's work demand temperature range and work demand humidity range. The air conditioner of the embodiments of the present application can be a large air conditioning device such as a multi-split variable frequency air conditioning system (i.e., a central air conditioner) applied in an industrial or commercial scenario.

[0026] Step 200, inputting the demand environment parameter, air conditioner current load parameter and air conditioner historical load parameter into an air conditioner temperature regulation value prediction model to obtain an air conditioner temperature regulation value prediction result output by the air conditioner temperature regulation value prediction model.

[0027] The electronic device further obtains an air conditioner current load parameter and an air conditioner historical load parameter. The air conditioner historical load parameter is historical data corresponding to the air conditioner current load parameter. The air conditioner current load parameter can include an air conditioner type (for example, fixed frequency, variable frequency), a regulation type (flexible regulation, rigid regulation), an air conditioner energy efficiency ratio, a refrigeration (heating) amount, a refrigeration (heating) rated power, etc., which can be obtained according to user interaction information. In addition to the historical data corresponding to the air conditioner current load parameter, the air conditioner historical load parameter can also include an air conditioner operation time interval, user comfort, device regulation potential, etc. The user comfort and device regulation potential can be quantified by defining a data interval. For example, the user comfort can be defined as 0-3 for poor, 4-7 for medium, and 8-10 for excellent.

[0028] On the basis of obtaining the demand environment parameters, the air conditioner current load parameters and the air conditioner historical load parameters, the electronic device can perform data preprocessing. For example, the electronic device can perform data cleaning on the demand environment parameters, the air conditioner current load parameters and the air conditioner historical load parameters. The data cleaning can specifically include deleting noise data, filling missing data, etc. The missing data can be filled by mean, median or mode.

[0029] The electronic device inputs the demand environment parameters, the air conditioner current load parameters and the air conditioner historical load parameters after data preprocessing into the air conditioner temperature adjustment value prediction model to obtain an air conditioner temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model. The air conditioner temperature adjustment value prediction model is constructed based on a linear model. The air conditioner temperature adjustment value prediction model can be constructed based on a multiple linear regression model, a ridge regression model, a polynomial regression model or the like. Thus, the air conditioner temperature adjustment value prediction is realized through multiple air conditioner load influence data to obtain the air conditioner temperature adjustment value prediction result. The air conditioner temperature adjustment value prediction result refers to an air conditioner temperature adjustment target value.

[0030] It should be noted that the electronic device can input the demand environment parameters, the air conditioner current load parameters and the air conditioner historical load parameters changing over time into the air conditioner temperature adjustment value prediction model to obtain time-continuous air conditioner temperature adjustment value prediction results.

[0031] Step 300, inputting the air conditioner temperature adjustment value prediction result, the work scene type and the meteorological parameter into the air conditioner temperature adjustment range prediction model to obtain an air conditioner temperature adjustment range prediction result of the air conditioner temperature adjustment range prediction model.

[0032] Please refer to Figure 2 The electronic device performs data normalization on the air conditioner temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model to obtain a feature value corresponding to the air conditioner temperature adjustment value prediction result. A feature library is obtained by accumulating multiple feature values to prepare for the prediction of the air conditioner temperature adjustment range.

[0033] The electronic device inputs the feature value corresponding to the air conditioner temperature adjustment value prediction result, the work scene type and the meteorological parameter into the air conditioner temperature adjustment range prediction model to obtain an air conditioner temperature adjustment range prediction result of the air conditioner temperature adjustment range prediction model. The work scene type includes a user type and a region type. The user type includes an industrial user and a commercial user. The region type includes a north China region and a southwest region. Similarly, the user type and the region type are also quantified by different numerical values. Taking the user type as an example, for example, an industrial user can be defined as 1 and a commercial user can be defined as 2. The meteorological parameter can include temperature, humidity, air pressure and the like of the environment.

[0034] The air conditioner temperature adjustment range prediction model is constructed based on a nonlinear model. For example, the air conditioner temperature adjustment range prediction model can be constructed based on a Long Short-Term Memory (LSTM), a Gated Recurrent Unit (GRU), or a FEDformer (Frequency Enhanced Decomposed Transformer) model. Thus, the embodiment of the present application can predict based on multiple input data of the air conditioner temperature adjustment value prediction result, the working scene type, and the meteorological parameter to obtain the air conditioner temperature adjustment range prediction result. Thus, according to the calculation and fitting of various air conditioner load influencing factors, the embodiment of the present application converts the demand physical quantity into a characteristic value, and can calculate the air conditioner temperature adjustment range by inputting the working scene type and the meteorological parameter, and combining the characteristic value.

[0035] Step 400: adjusting the temperature of the air conditioner based on the air conditioner temperature adjustment range prediction result to realize air conditioner load regulation.

[0036] The electronic device can randomly take a value within the temperature adjustment range in the air conditioner temperature adjustment range prediction result to obtain a target temperature adjustment value, and adjust the temperature of the air conditioner to the target temperature adjustment value. For example, the air conditioner temperature adjustment range prediction result is 20-22 degrees Celsius, and the electronic device randomly takes a value within 20-22 degrees Celsius, for example, 21.5 degrees Celsius. Then the electronic device adjusts the temperature of the air conditioner to 21.5 degrees Celsius. In another embodiment, the electronic device can calculate the average value within the temperature adjustment range in the air conditioner temperature adjustment range prediction result to obtain a target temperature adjustment value, and adjust the temperature of the air conditioner to the target temperature adjustment value. For example, the air conditioner temperature adjustment range prediction result is 20-22 degrees Celsius, and the electronic device calculates the average value of 20-22 degrees Celsius to obtain 21 degrees Celsius. Then the electronic device adjusts the temperature of the air conditioner to 21 degrees Celsius. The embodiment of the present application adjusts the temperature of the air conditioner based on the air conditioner temperature adjustment range prediction result to realize air conditioner load regulation, which can make the air conditioner reach the target power in a shorter time, better and faster realize the regulation target, and optimize the air conditioner load regulation effect.

[0037] Please refer to Figure 3In this embodiment of the invention, the algorithm library primarily performs calculations and fitting based on various factors influencing air conditioning load, transforming demand physical quantities into feature values. It can also calculate the air conditioning temperature adjustment range by combining these feature values ​​with parameter inputs such as work scenario type and meteorological parameters. The feature library is a collection of feature values ​​corresponding to the predicted air conditioning temperature adjustment values ​​obtained through training on basic data. The application interface provides functional interfaces for external calls, enabling data reading and writing, data initialization, and other related functions. The algorithm library, load management system, and application interface together form a load model library.

[0038] This invention combines linear and nonlinear models to predict the air conditioning temperature adjustment range by fully considering various influencing factors such as environmental parameters, current air conditioning load parameters, historical air conditioning load parameters, work scenario type, and meteorological parameters. Based on the predicted temperature adjustment range, the air conditioner temperature is adjusted to achieve air conditioning load regulation. This invention integrates multiple models and multiple influencing factors to predict the air conditioning temperature adjustment range, thereby improving the prediction accuracy of the air conditioning temperature adjustment value and optimizing the air conditioning load regulation effect.

[0039] In other aspects of this invention, the air conditioning temperature regulation value prediction model is constructed using a multiple linear regression model; the air conditioning temperature regulation value prediction model is trained through the following steps: Acquire multiple datasets, which include sample demand environment parameters, sample air conditioning current load parameters, and sample air conditioning historical load parameters; Repeat the following steps until the set stopping condition is met: input the first dataset into the air conditioner temperature regulation value prediction model to obtain the sample temperature regulation value prediction result output by the air conditioner temperature regulation value prediction model; the first dataset is any one of the multiple datasets; calculate the loss function based on the sample temperature regulation value prediction result and the real temperature regulation value label corresponding to the first dataset; adjust the weight coefficients of the air conditioner temperature regulation value prediction model based on the loss function.

[0040] In one embodiment, a multiple linear regression model can be expressed by the following formula: Y = β 0+ β 1 X 1+ β 2 X 2+…+ β p X p + ε ; in ，β 0 、β 1、β 2 、…、β p a weight parameter representing a multiple linear regression model; ε a intercept factor. X 1、 X 2、 …、X p a sample demand environment parameter, a sample air conditioner current load parameter, and a sample air conditioner historical load parameter. Y a sample temperature adjustment value prediction result. The sample temperature adjustment value prediction result is obtained by inputting the sample demand environment parameter, the sample air conditioner current load parameter, and the sample air conditioner historical load parameter into the multiple linear regression model. Then, a loss function (for example, mean square error) is calculated based on the sample temperature adjustment value prediction result and a true temperature adjustment value label corresponding to the first data set. Finally, the weight coefficient of the air conditioner temperature adjustment value prediction model is adjusted based on the loss function. The above steps are repeatedly executed until a set number of iterations is reached. Thus, a trained air conditioner temperature adjustment value prediction model is obtained.

[0041] The multiple linear regression model is used as a flexible load baseline optimization model for air conditioner temperature adjustment value prediction in the embodiments of the present application. That is, the multiple linear regression model is used as a curve fitting algorithm in the embodiments of the present application. The multiple linear regression model has strong explainability and high computational efficiency, which improves the computational efficiency of air conditioner temperature adjustment value prediction and provides a basis for predicting the air conditioner temperature adjustment range.

[0042] In other aspects of the embodiments of the present application, the air conditioner temperature adjustment range prediction model is constructed using a deep learning model. The air conditioner temperature adjustment range prediction model is trained by the following steps: a plurality of data sets are obtained, the data sets including a sample air conditioner temperature adjustment value prediction result, a sample working scenario type, and a sample meteorological parameter; the following steps are repeatedly executed until a set stop condition is reached: a first data set is input into the air conditioner temperature adjustment range prediction model to obtain a sample air conditioner temperature adjustment range prediction result output by the air conditioner temperature adjustment range prediction model; the first data set is any one of the plurality of data sets; a loss function is calculated based on the sample air conditioner temperature adjustment range prediction result and a true air conditioner temperature adjustment range label corresponding to the first data set; and network parameters of the air conditioner temperature adjustment range prediction model are adjusted based on the loss function.

[0043] In an embodiment, the deep learning model can be constructed using a FEDformer algorithm. The embodiment of the present application inputs the sample air conditioner temperature adjustment value prediction result, the sample working scene type, and the sample meteorological parameter into the deep learning model to obtain a sample air conditioner temperature adjustment range prediction result. Then, a loss function (such as a cross-entropy loss function) is calculated based on the sample temperature adjustment value prediction result and a true air conditioner temperature adjustment range label corresponding to the first data set. Finally, the weight coefficient of the air conditioner temperature adjustment range prediction model is adjusted based on the loss function. The above steps are repeatedly executed until a set number of iterations is reached. Thus, a trained air conditioner temperature adjustment range prediction model is obtained.

[0044] Therefore, the embodiment of the present application combines the deep learning model and the flexible load baseline optimization model as an updated load baseline, fully considers the variability caused by time, weather, and other environmental factors, improves the prediction accuracy at abnormal points, further improves the overall prediction accuracy of the flexible load baseline, improves the prediction accuracy of the air conditioner temperature adjustment value, and further optimizes the air conditioner load regulation effect.

[0045] In other aspects of the embodiment of the present application, after step 400, temperature adjustment is performed on the air conditioner based on the air conditioner temperature adjustment range prediction result to achieve air conditioner load regulation, which further includes: Obtaining a load regulation target value.

[0046] Determining an average load regulation target value of each branch node of the air conditioner based on the load regulation target value.

[0047] Repeating the following steps until the first deviation and the node power total deviation of the nodes meet a set condition: calculating candidate load adjustment amounts of all branch nodes of the air conditioner based on the initial load adjustment amount using a gradient descent method; performing load adjustment on all branch nodes based on the candidate load adjustment amounts; and calculating the first deviation and the node power total deviation of all branch nodes.

[0048] The first deviation represents the difference between the load balance coefficient of all branch nodes after load adjustment and the historical load balance coefficient before load adjustment; the load balance coefficient represents the deviation degree of the load of each branch node from the average load of all branch nodes of the air conditioner; the node power total deviation represents the sum of the node power deviations of all branch nodes after load adjustment; the node power deviation represents the deviation between the candidate load adjustment amount and the load regulation gap value; the load regulation gap value is calculated based on the deviation between the current power value of the branch node and the average load regulation target value; and the candidate load adjustment amount after iteration is taken as the load regulation strategy value of all branch nodes.

[0049] Since the air conditioner of the embodiment of the present application selects a multi-connected variable frequency air conditioning system. Then the multi-connected variable frequency air conditioning system is composed of multiple branch nodes. The core function of the branch node in the multi-connected variable frequency air conditioning system in the embodiment of the present application is to split the main refrigerant pipeline (liquid pipe and gas pipe) from the outdoor unit to multiple indoor units as needed, to realize accurate distribution of refrigerant in the parallel circuit. Therefore, the multiple branch nodes of the multi-connected variable frequency air conditioning system of the embodiment of the present application can be a divergent pipe as a refrigerant pipeline splitter, and an electronic expansion valve installed at the inlet of each indoor unit. After the electronic device obtains the load regulation target value, the load regulation target value needs to be allocated to all load-controllable branch nodes of the multi-connected variable frequency air conditioning system. Each branch node needs to be analyzed according to the load parameter in the current controllable range, and finally the load regulation strategy value of each branch node is obtained.

[0050] Please refer to Figure 4 , the electronic device first obtains the load regulation target value, and then determines the average load regulation target value of each branch node of the air conditioner based on the load regulation target value. For example, the load regulation target value is 100w, and the average load regulation target value of each branch node is 100 / 5=20w. It should be noted that the embodiment of the present application will preset a mapping relationship between different air conditioner temperature adjustment ranges and load regulation target values. Therefore, after obtaining the air conditioner temperature adjustment range prediction result, the electronic device can obtain the load regulation target value corresponding to the air conditioner temperature adjustment range prediction result based on the above mapping relationship. Each branch node has its own load adjustment range, and the average load regulation target value cannot be greater than the maximum adjustment amount of the load adjustment range.

[0051] The following describes the calculation process of the load balancing coefficient in the embodiment of the present application. First, the load rate of all branch nodes is calculated; then the load balancing coefficient is calculated based on the load rate of all branch nodes. In one embodiment, the load balancing coefficient is calculated by the following steps: calculating the load rate of all branch nodes; the load rate is calculated based on the apparent power and load capacity of the branch node; calculating the average load rate based on the load rate of all branch nodes; calculating the load balancing coefficient of all branch nodes based on the load rate of all branch nodes and the average load rate.

[0052] Wherein the load rate of the branch node is calculated by the following formula: S i = ;λ i =S i / C i ; Wherein, P i represents the active power of the i-th branch node; Q i represents the reactive power of the i-th branch node; S irepresents the apparent power of the i-th branch node; C i represents the load capacity of the i-th branch node, wherein the load capacity represents the maximum carrying power of the load of the branch node; λ i represents the load rate of the i-th branch node, wherein the load rate reflects the load tension degree of the branch node. i = 1, 2, 3, …, n. i represents the i-th branch node, and n represents the total number of branch nodes.

[0053] The load balancing coefficient is calculated by the following formula: λ= λ i ; ; K std ; Wherein, λ represents the average load rate. K std represents the load balancing coefficient, wherein the value of the load balancing coefficient is closer to 1, and the balancing value is better. The calculation of the load balancing coefficient in the embodiment of the application reflects the deviation degree of the load rate of each branch node from the average load rate, and mainly applies the standard deviation calculation method for calculation. In other embodiments, the load balancing coefficient can also be calculated by applying the variance calculation method.

[0054] The embodiment of the application matches power based on the load regulation target value, adopts an infinite approximation calculation method, and obtains the load regulation strategy value matched by each branch node. The infinite approximation calculation method adopts an iterative optimization algorithm, adjusts the load power of the adjustable branch node step by step, so as to achieve the target of the overall load regulation target value, and the specific steps are as follows: I. Initial load adjustment amount setting Let the initial load adjustment amount be , calculate the initial load balancing coefficient K std （0） and the initial node power deviation E i (0) (t).

[0055] Wherein, i represents the i-th branch node, and t is a time point.

[0056] II. Iterative optimization process Adjust the adjustable load by iteration, and satisfy the power balance and adjustment constraints.

[0057] Iteration step (kth time): a) Calculate the kth node power deviation ; D i(t) is the load regulation gap value of the node i, which is obtained by reversing the average load regulation target value, for example, the current power value of the node i is 15w, and the average load regulation target value is 20w, so the load regulation gap value of the node i is 20-15=5w. is the kth candidate load adjustment value of the ith branch node. represents the kth node power deviation of the ith branch node.

[0058] b) Update strategy of the candidate load adjustment value Gradient-based adjustment: ; Wherein, γ represents an iteration step (which needs to be dynamically adjusted to avoid oscillation); represents the sensitivity of the equilibrium coefficient to the load adjustment value; ω i is the priority of the ith load, for example, the priority of the important load is high, and the load adjustment value is limited; k represents the iteration number; is the k+1th candidate load adjustment value of the ith branch node.

[0059] c) Convergence judgment When the following conditions are met, the iteration is terminated: And ( , is a preset precision threshold, such as 10 -3 ); Wherein, represents the above-mentioned first deviation, the difference between the load equilibrium coefficient of all branch nodes after the first deviation load adjustment and the historical load equilibrium coefficient before the load adjustment; represents the total node power deviation, which represents the sum of the node power deviations of all branch nodes after the load adjustment. It should be noted that the candidate load adjustment value after the iteration is used as the load regulation strategy value of all branch nodes.

[0060] The embodiment of the application calculates the load regulation strategy value of all branch nodes by using the infinite approximation calculation method based on the load regulation target value. Therefore, the embodiment of the application distributes the load regulation target value to all branch nodes based on the load regulation target value, each branch node needs to be analyzed according to the load parameter in the current controllable range, and finally the load regulation strategy value of each branch node is obtained.

[0061] In other aspects of the embodiments of the present application, after obtaining the load regulation strategy value of all branch nodes, the air conditioner load regulation method of the embodiments of the present application further comprises: calculating a regulation coefficient based on the load regulation target value and the load regulation strategy value; the load regulation target value and the regulation coefficient are in a positive correlation relationship; the load regulation strategy value and the regulation coefficient are in a negative correlation relationship; calculating a regulation evaluation score based on the regulation coefficient and a set evaluation coefficient; the regulation coefficient and the regulation evaluation score are in a positive correlation relationship; the set evaluation coefficient and the regulation evaluation score are in a negative correlation relationship; determining a regulation effect based on the regulation evaluation score.

[0062] Please refer to Figure 5 , the embodiments of the present application compare the load regulation target value and the load regulation strategy value to obtain the regulation effect (evaluation level) of the current scheme. Specifically, the electronic device first calculates a regulation coefficient = (Vadj / Vdest)*100% based on the load regulation target value and the load regulation strategy value; wherein Vadj represents the load regulation target value; Vdest represents the load regulation strategy value. The electronic device then calculates a regulation evaluation score = (1+regulation coefficient-set evaluation coefficient)*100% based on the regulation coefficient and a set evaluation coefficient. Finally, the regulation effect is determined based on the regulation evaluation score. The embodiments of the present application can divide the regulation effect into excellent, good, medium, and poor effects according to the requirements of the business system. For example, the regulation evaluation score is 100% or more for excellent, the regulation evaluation score is 90% or more for good, the regulation evaluation score is 80% or more for medium, and the regulation evaluation score is 80% or less for poor. Therefore, the embodiments of the present application can calculate the regulation evaluation score corresponding to the load regulation strategy value of each branch node to obtain the regulation effect of each branch node.

[0063] Please refer to Figure 6 , the algorithm library of the embodiments of the present application further comprises a scheme matching algorithm and a scheme evaluation algorithm. The embodiments of the present application use the scheme matching algorithm to perform power matching based on the load regulation target value, and use an infinite approximation calculation method to obtain the load regulation strategy value matched by each branch node. The embodiments of the present application also use the scheme evaluation algorithm to compare the regulation scheme power value with the target power value to obtain the evaluation level of the branch node. The application interface is a function interface that provides external calls, and completes the functions of scheduling strategy and strategy evaluation through interface calls. The present application comprises a regulation strategy component through the algorithm library and the application interface.

[0064] The air conditioner flexible load regulation strategy of the embodiment of the present application is formulated, the regulation period and the regulation total task are confirmed through the air conditioner load baseline, the real-time regulation strategy is confirmed based on the air conditioner flexible load characteristic database and the target dynamic programming algorithm, and the load regulation effect is evaluated by using the regulation evaluation mechanism. For the load regulation target value of the air conditioner load, the dynamic programming method of the air conditioner operation is designed based on the real-time collected air conditioner load data and the start-stop state, and it is required to complete the regulation task as much as possible in the state of controlling a small number of branch nodes participating in the regulation. The model outputs the regulation command (start or stop) of each branch node in the regulation period by inputting the real-time data of the air conditioner power. The regulation interaction effect evaluation mechanism and the calculation method are formulated, the load regulation strategy value matched by the participating branch node is compared with the issued load regulation target value, and the completion degree and the regulation effect of the regulation task are evaluated.

[0065] The embodiment of the present application can also formulate a "cloud-edge-end" collaborative management and control air conditioner load management architecture to realize flexible regulation of the air conditioner flexible load. The cloud management and control platform is responsible for management of the characteristic database, updating of the algorithm model and formulation of the load regulation target value; the edge side device is responsible for formulation of the specific regulation strategy and evaluation of the regulation effect; and the load management device is responsible for load and environment data collection and regulation task execution.

[0066] The embodiment of the present application performs comprehensive evaluation on the regulation effect from the response completion degree, the interaction speed and the regulation stability by formulating the regulation interaction effect evaluation mechanism and the calculation method, and performs localization continuous optimization on the algorithm according to the evaluation result. The flexible load analysis is carried out locally in the load management device, the real-time data used for the analysis is better, the algorithm adaptability is better, and the error of the universal algorithm is excluded. The model is introduced to complete establishment of the flexible load baseline, continuous model learning and optimization can be completed, and the complexity of the main control MCU system and the development difficulty of the program are reduced by relying on the algorithm of the model.

[0067] In summary, please refer to Figure 7 The embodiment of the present application establishes, on one hand, the multi-type flexible load fine model (air conditioner temperature regulation value prediction model and air conditioner temperature regulation range prediction model) with different inherent parameters and the air conditioner load typical characteristic database containing the user portrait in the complex environment. On the other hand, the load baseline and the adjustable potential can be accurately predicted by using the big data analysis technology, the regulation interaction strategy is formulated by combining the load identification technology, the accurate and flexible regulation of the end side flexible load is realized, the comprehensive evaluation mechanism of the regulation effect is established, and the overall algorithm is continuously updated based on the statistical regulation effect and the continuously updated load data.

[0068] Device embodiment Please refer to Figure 8In another aspect, the embodiments of the present application also provide an air conditioner load regulation device, comprising: The acquisition module 801 is configured to acquire an environmental parameter regulation instruction, wherein the environmental parameter regulation instruction carries a demand environmental parameter of a user. The first prediction module 802 is configured to input the demand environmental parameter, a current load parameter of an air conditioner, and a historical load parameter of the air conditioner into an air conditioner temperature adjustment value prediction model to obtain an air conditioner temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model. The second prediction module 803 is configured to input the air conditioner temperature adjustment value prediction result, a work scene type, and a meteorological parameter into an air conditioner temperature adjustment range prediction model to obtain an air conditioner temperature adjustment range prediction result of the air conditioner temperature adjustment range prediction model. The adjustment module 804 is configured to perform temperature adjustment on the air conditioner based on the air conditioner temperature adjustment range prediction result to realize air conditioner load regulation. The air conditioner temperature adjustment value prediction model is constructed based on a linear model, and the air conditioner temperature adjustment range prediction model is constructed based on a nonlinear model.

[0069] Optionally, the air conditioner temperature adjustment value prediction model is constructed by using a multiple linear regression model, and the air conditioner temperature adjustment value prediction model is trained by the following steps: A plurality of data sets are acquired, wherein the data sets include a sample demand environmental parameter, a sample current load parameter of an air conditioner, and a sample historical load parameter of the air conditioner. The following steps are repeatedly executed until a set stop condition is reached: A first data set is input into the air conditioner temperature adjustment value prediction model to obtain a sample temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model; the first data set is any one of the plurality of data sets. A loss function is calculated based on the sample temperature adjustment value prediction result and a real temperature adjustment value label corresponding to the first data set. The weight coefficient of the air conditioner temperature adjustment value prediction model is adjusted based on the loss function.

[0070] Optionally, the air conditioner temperature adjustment range prediction model is constructed by using a deep learning model, and the air conditioner temperature adjustment range prediction model is trained by the following steps: A plurality of data sets are acquired, wherein the data sets include a sample air conditioner temperature adjustment value prediction result, a sample work scene type, and a sample meteorological parameter. The following steps are repeatedly executed until a set stop condition is reached: inputting a first data set into the air conditioner temperature regulation range prediction model to obtain a sample air conditioner temperature regulation range prediction result output by the air conditioner temperature regulation range prediction model; the first data set is any one of the plurality of data sets; calculating a loss function based on the sample air conditioner temperature regulation range prediction result and a true air conditioner temperature regulation range label corresponding to the first data set; adjusting network parameters of the air conditioner temperature regulation range prediction model based on the loss function.

[0071] Optionally, the device further comprises: The computing module is configured to: obtain a load regulation target value; determine an average load regulation target value of each branch node of the air conditioner based on the load regulation target value; repeatedly perform all the following steps until a first deviation and a total node power deviation of all branch nodes meet a set condition: calculate candidate load adjustment amounts of all branch nodes of the air conditioner based on an initial load adjustment amount by using a gradient descent method; perform load adjustment on all branch nodes based on the candidate load adjustment amounts; and calculate the first deviation and the total node power deviation of all branch nodes. The first deviation represents a difference between a load balance coefficient of all branch nodes after load adjustment and a historical load balance coefficient before load adjustment; the load balance coefficient represents a deviation degree of a load of each branch node from an average load of all branch nodes of the air conditioner; the total node power deviation represents a sum of node power deviations of all branch nodes after load adjustment; the node power deviation represents a deviation between the candidate load adjustment amount and a load regulation gap value; the load regulation gap value is calculated based on a deviation between a current power value of the branch node and the average load regulation target value; and the candidate load adjustment amount after iteration is taken as a load regulation strategy value of all branch nodes.

[0072] Optionally, the load balance coefficient is calculated by the following steps: Calculate a load rate of all branch nodes; the load rate is calculated based on an apparent power of the branch node and a load capacity; Calculate an average load rate based on the load rates of all branch nodes; Calculate a load balance coefficient of all branch nodes based on the load rates of all branch nodes and the average load rate.

[0073] Optionally, the device further comprises: The regulation effect determination module is configured to calculate a regulation coefficient based on the load regulation target value and the load regulation strategy value; the load regulation target value and the regulation coefficient are positively correlated; the load regulation strategy value and the regulation coefficient are negatively correlated; calculate a regulation evaluation score based on the regulation coefficient and a set evaluation coefficient; the regulation coefficient and the regulation evaluation score are positively correlated; the set evaluation coefficient and the regulation evaluation score are negatively correlated; and determine the regulation effect based on the regulation evaluation score.

[0074] The air conditioner load regulation device includes a processor and a memory, and the acquisition module 801, the first prediction module 802, the second prediction module 803, and the adjustment module 804 are stored in the memory as program units and are executed by the processor to realize corresponding functions.

[0075] The processor includes a core, and the core retrieves corresponding program units from the memory. The core can be provided with one or more than one.

[0076] The memory can include a non-permanent memory in a computer readable medium, a random access memory (RAM), and / or a non-volatile memory such as a read-only memory (ROM) or a flash memory (flash RAM), and the memory includes at least one memory chip.

[0077] Figure 9 An example of an electronic device is shown in FIG. 10, which shows a schematic diagram of a physical structure of an electronic device. Figure 9 As shown in FIG. 10, the electronic device can include a processor 910, a communications interface 920, a memory 930, and a communications bus 940, wherein the processor 910, the communications interface 920, and the memory 930 communicate with each other through the communications bus 940. The processor 910 can invoke a logical instruction in the memory 930 to execute an air conditioner load regulation method, which includes: acquiring an environment parameter regulation instruction carrying a demand environment parameter of a user; inputting the demand environment parameter, a current load parameter of an air conditioner, and a historical load parameter of the air conditioner into an air conditioner temperature adjustment value prediction model to obtain an air conditioner temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model; inputting the air conditioner temperature adjustment value prediction result, a work scene type, and a meteorological parameter into an air conditioner temperature adjustment range prediction model to obtain an air conditioner temperature adjustment range prediction result of the air conditioner temperature adjustment range prediction model; and adjusting a temperature of the air conditioner based on the air conditioner temperature adjustment range prediction result to regulate a load of the air conditioner; wherein the air conditioner temperature adjustment value prediction model is constructed based on a linear model, and the air conditioner temperature adjustment range prediction model is constructed based on a nonlinear model.

[0078] In addition, the logical instructions in the memory 930 described above can be implemented in the form of a software function unit and sold or used as an independent product, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0079] In another aspect, the present application also provides a computer program product, the computer program product comprising a computer program, the computer program being stored in a machine readable storage medium, and the computer program being executed by a processor, so that the computer can execute the air conditioner load regulation method, the method comprising: obtaining an environment parameter regulation instruction, the environment parameter regulation instruction carrying a demand environment parameter of a user; inputting the demand environment parameter, a current load parameter of the air conditioner and a historical load parameter of the air conditioner into an air conditioner temperature adjustment value prediction model to obtain an air conditioner temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model; inputting the air conditioner temperature adjustment value prediction result, a work scene type and a meteorological parameter into an air conditioner temperature adjustment range prediction model to obtain an air conditioner temperature adjustment range prediction result of the air conditioner temperature adjustment range prediction model; and adjusting the temperature of the air conditioner based on the air conditioner temperature adjustment range prediction result to realize air conditioner load regulation; wherein the air conditioner temperature adjustment value prediction model is constructed based on a linear model, and the air conditioner temperature adjustment range prediction model is constructed based on a nonlinear model.

[0080] In yet another aspect, the present application also provides a machine readable storage medium having stored thereon a computer program, which, when executed by a processor, implements a method for regulating an air conditioner load, the method comprising: obtaining an environmental parameter regulation instruction, the environmental parameter regulation instruction carrying a demand environmental parameter of a user; inputting the demand environmental parameter, a current load parameter of the air conditioner and a historical load parameter of the air conditioner into an air conditioner temperature adjustment value prediction model to obtain an air conditioner temperature adjustment value prediction result output by the air conditioner temperature adjustment value prediction model; inputting the air conditioner temperature adjustment value prediction result, a work scene type and a meteorological parameter into an air conditioner temperature adjustment range prediction model to obtain an air conditioner temperature adjustment range prediction result of the air conditioner temperature adjustment range prediction model; and adjusting a temperature of the air conditioner based on the air conditioner temperature adjustment range prediction result to regulate the air conditioner load; wherein the air conditioner temperature adjustment value prediction model is constructed based on a linear model, and the air conditioner temperature adjustment range prediction model is constructed based on a nonlinear model.

[0081] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment scheme according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0082] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software plus necessary general hardware platforms, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0083] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for regulating air conditioning load, characterized in that, include: Obtain environmental parameter control instructions, wherein the environmental parameter control instructions carry the user's required environmental parameters; The environmental demand parameters, current air conditioning load parameters, and historical air conditioning load parameters are input into the air conditioning temperature regulation value prediction model to obtain the air conditioning temperature regulation value prediction result output by the air conditioning temperature regulation value prediction model. The predicted air conditioning temperature adjustment value, the working scenario type, and the meteorological parameters are input into the air conditioning temperature adjustment range prediction model to obtain the predicted air conditioning temperature adjustment range of the air conditioning temperature adjustment range prediction model. The air conditioner temperature is adjusted based on the predicted air conditioner temperature range to achieve air conditioner load control. The air conditioning temperature adjustment value prediction model is based on a linear model, while the air conditioning temperature adjustment range prediction model is based on a nonlinear model.

2. The air conditioning load control method according to claim 1, characterized in that, The air conditioning temperature regulation value prediction model is constructed using a multiple linear regression model; the air conditioning temperature regulation value prediction model is trained through the following steps: Acquire multiple datasets, which include sample demand environment parameters, sample air conditioning current load parameters, and sample air conditioning historical load parameters; Repeat the following steps until the set stopping condition is met: The first dataset is input into the air conditioner temperature regulation value prediction model to obtain the sample temperature regulation value prediction result output by the air conditioner temperature regulation value prediction model; the first dataset is any one of the multiple datasets; The loss function is calculated based on the predicted temperature regulation value of the sample and the actual temperature regulation value label corresponding to the first dataset. The weight coefficients of the air conditioning temperature regulation value prediction model are adjusted based on the loss function.

3. The air conditioning load control method according to claim 1, characterized in that, The air conditioner temperature adjustment range prediction model is constructed using a deep learning model; the air conditioner temperature adjustment range prediction model is trained through the following steps: Multiple datasets are acquired, including sample air conditioning temperature regulation value prediction results, sample work scenario types, and sample meteorological parameters; Repeat the following steps until the set stopping condition is met: The first dataset is input into the air conditioner temperature adjustment range prediction model to obtain the sample air conditioner temperature adjustment range prediction result output by the air conditioner temperature adjustment range prediction model; the first dataset is any one of the multiple datasets; The loss function is calculated based on the predicted air conditioning temperature adjustment range of the sample and the actual air conditioning temperature adjustment range label corresponding to the first dataset. The network parameters of the air conditioning temperature regulation range prediction model are adjusted based on the loss function.

4. The air conditioning load control method according to claim 1, characterized in that, After adjusting the air conditioner temperature based on the predicted air conditioner temperature range to achieve air conditioner load control, the method further includes: Obtain the target value for load control; Based on the load control target value, determine the average load control target value for each branch node of the air conditioner; Repeat all of the following steps until the first deviation and the total node power deviation meet the set conditions: The gradient descent method is used to calculate the candidate load adjustment values ​​for all branch nodes of the air conditioning system based on the initial load adjustment value. Load adjustment is performed on all branch nodes based on the candidate load adjustment amount; Calculate the first deviation and the total node power deviation of all branch nodes; Wherein, the first deviation represents the difference between the load balance coefficient of all branch nodes after load adjustment and the historical load balance coefficient before load adjustment; the load balance coefficient represents the degree of deviation between the load of each branch node and the average load of all branch nodes of the air conditioning system; the total node power deviation represents the sum of the node power deviations of all branch nodes after load adjustment; the node power deviation represents the deviation between the candidate load adjustment amount and the load control gap value; the load control gap value is calculated based on the deviation between the current power value of the branch node and the average load control target value; the candidate load adjustment amount after the iteration is completed is used as the load control strategy value of all branch nodes.

5. The air conditioning load control method according to claim 4, characterized in that, The load balancing coefficient is calculated through the following steps: Calculate the load factor for all branch nodes; the load factor is calculated based on the apparent power and load capacity of the branch nodes. Calculate the average load rate based on the load rates of all the branch nodes; Based on the load rate of all branch nodes and the average load rate, calculate the load balancing coefficient of all branch nodes.

6. The air conditioning load control method according to claim 4, characterized in that, The method further includes: The control coefficient is calculated based on the load control target value and the load control strategy value; there is a positive correlation between the load control target value and the control coefficient; there is a negative correlation between the load control strategy value and the control coefficient. The regulation evaluation score is calculated based on the regulation coefficient and the set evaluation coefficient; there is a positive correlation between the regulation coefficient and the regulation evaluation score; there is a negative correlation between the set evaluation coefficient and the regulation evaluation score. The effectiveness of regulation is determined based on the regulation evaluation score.

7. An air conditioning load control device, characterized in that, include: The acquisition module is used to acquire environmental parameter control instructions, which carry the user's required environmental parameters. The first prediction module is used to input the demand environment parameters, the current air conditioning load parameters and the historical air conditioning load parameters into the air conditioning temperature regulation value prediction model, and obtain the air conditioning temperature regulation value prediction result output by the air conditioning temperature regulation value prediction model. The second prediction module is used to input the predicted result of the air conditioner temperature adjustment value, the working scenario type and meteorological parameters into the air conditioner temperature adjustment range prediction model to obtain the predicted result of the air conditioner temperature adjustment range of the air conditioner temperature adjustment range prediction model. The adjustment module is used to adjust the temperature of the air conditioner based on the predicted temperature adjustment range of the air conditioner, so as to realize the control of the air conditioner load; The air conditioning temperature adjustment value prediction model is based on a linear model, while the air conditioning temperature adjustment range prediction model is based on a nonlinear model.

8. The air conditioning load control device according to claim 7, characterized in that, The air conditioning temperature regulation value prediction model is constructed using a multiple linear regression model; the air conditioning temperature regulation value prediction model is trained through the following steps: Acquire multiple datasets, which include sample demand environment parameters, sample air conditioning current load parameters, and sample air conditioning historical load parameters; Repeat the following steps until the set stopping condition is met: The first dataset is input into the air conditioner temperature regulation value prediction model to obtain the sample temperature regulation value prediction result output by the air conditioner temperature regulation value prediction model; the first dataset is any one of the multiple datasets; The loss function is calculated based on the predicted temperature regulation value of the sample and the actual temperature regulation value label corresponding to the first dataset. The weight coefficients of the air conditioning temperature regulation value prediction model are adjusted based on the loss function.

9. The air conditioning load control device according to claim 7, characterized in that, The air conditioner temperature adjustment range prediction model is constructed using a deep learning model; the air conditioner temperature adjustment range prediction model is trained through the following steps: Multiple datasets are acquired, including sample air conditioning temperature regulation value prediction results, sample work scenario types, and sample meteorological parameters; Repeat the following steps until the set stopping condition is met: The first dataset is input into the air conditioner temperature adjustment range prediction model to obtain the sample air conditioner temperature adjustment range prediction result output by the air conditioner temperature adjustment range prediction model; the first dataset is any one of the multiple datasets; The loss function is calculated based on the predicted air conditioning temperature adjustment range of the sample and the actual air conditioning temperature adjustment range label corresponding to the first dataset. The network parameters of the air conditioning temperature regulation range prediction model are adjusted based on the loss function.

10. The air conditioning load control device according to claim 7, characterized in that, The device further includes: A calculation module is used to obtain the load control target value; determine the average load control target value for each branch node of the air conditioner based on the load control target value; repeat all the following steps until the first deviation and the total node power deviation meet the set conditions: calculate the candidate load adjustment amount for all branch nodes of the air conditioner based on the initial load adjustment amount using the gradient descent method; adjust the load of all branch nodes based on the candidate load adjustment amount; calculate the first deviation and the total node power deviation of all branch nodes; Wherein, the first deviation represents the difference between the load balance coefficient of all branch nodes after load adjustment and the historical load balance coefficient before load adjustment; the load balance coefficient represents the degree of deviation between the load of each branch node and the average load of all branch nodes of the air conditioning system; the total node power deviation represents the sum of the node power deviations of all branch nodes after load adjustment; the node power deviation represents the deviation between the candidate load adjustment amount and the load control gap value; the load control gap value is calculated based on the deviation between the current power value of the branch node and the average load control target value; the candidate load adjustment amount after the iteration is completed is used as the load control strategy value of all branch nodes.

11. The air conditioning load control device according to claim 10, characterized in that, The load balancing coefficient is calculated through the following steps: Calculate the load factor for all branch nodes; the load factor is calculated based on the apparent power and load capacity of the branch nodes. Calculate the average load rate based on the load rates of all the branch nodes; Based on the load rate of all branch nodes and the average load rate, calculate the load balancing coefficient of all branch nodes.

12. The air conditioning load control device according to claim 10, characterized in that, The device further includes: The control effect determination module is used to calculate the control coefficient based on the load control target value and the load control strategy value; the load control target value and the control coefficient are positively correlated; the load control strategy value and the control coefficient are negatively correlated; calculate the control evaluation score based on the control coefficient and the set evaluation coefficient; the control coefficient and the control evaluation score are positively correlated; the set evaluation coefficient and the control evaluation score are negatively correlated; and determine the control effect based on the control evaluation score.

13. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the air conditioning load control method according to any one of claims 1 to 6.

14. A machine-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the air conditioning load control method according to any one of claims 1 to 6.

15. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the air conditioning load control method according to any one of claims 1 to 6.

Images