A method for simulating and constructing an air-conditioning load prediction model assisted by a computer

Abstract

The present application relates to the field of data processing technology, and specifically to a computer-aided simulation method for constructing an air conditioning load prediction model. The method comprises: real-time collection of data of each environmental parameter in a building at all times within each cycle to form a monitoring sequence; simulating the energy consumption of the air conditioner to obtain the load data of the air conditioning load at all times within each cycle and form a load sequence; obtaining the lag time; extracting the time-lag subsequence of each environmental parameter in each cycle; determining the trend synchronization between each environmental parameter and the air conditioning load in each cycle; determining the coupling coefficient of any two environmental parameters in each cycle; determining a reference evaluation value between the current cycle and the remaining cycles; obtaining each reference cycle; and constructing a load prediction model. The present application can avoid prediction bias caused by ignoring the interaction and lag effects between multiple environmental parameters, and improve the accuracy of air conditioning load prediction.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a computer-aided simulation method for constructing an air conditioning load prediction model. Background Art

[0002] Central air-conditioning systems, as core facilities in industrial plants, commercial complexes, and public buildings, are seeing a gradual increase in energy consumption. To reduce this energy consumption, building accurate prediction models allows for early understanding of future cooling and heating demand changes. This provides a basis for decision-making in optimizing refrigeration unit start-up and shutdown, scheduling ice storage systems, and utilizing renewable energy resources. This reduces energy waste and conserves energy.

[0003] Although the existing technology has adopted the multivariate linear regression method to take into account the impact of various environmental factors on the air-conditioning load when predicting the air-conditioning load, it usually analyzes the impact of various environmental factors based on the synchronous changes of the air-conditioning load and the environmental factors. It does not deeply consider the dynamic coupling effect between multiple environmental factors and the lagged effect of environmental factors on the air-conditioning load. It only relies on static weight distribution and cannot adapt to the time-varying characteristics of the interaction of multiple environmental factors, which leads to inaccurate prediction of the air-conditioning load. Summary of the Invention

[0004] In order to solve the above technical problems, a computer-aided air conditioning load prediction model simulation construction method is provided to solve the existing problems.

[0005] The solution to the technical problem of this application is to provide a computer-aided air conditioning load prediction model simulation construction method, including the following steps:

[0006] Collect data of each environmental parameter in the building at all times in each cycle in real time to form a monitoring sequence; simulate the energy consumption of air conditioners to obtain the load data of air conditioners at all times in each cycle and form a load sequence;

[0007] Analyze the hysteresis delay between the monitoring sequence of each environmental parameter and the load sequence in each cycle to obtain the hysteresis duration;

[0008] Divide all moments in each cycle into multiple time periods, and record the time periods in other cycles corresponding to the time period of the current moment in the current cycle as adjacent time periods; based on the adjacent time periods and the lag duration, slide a sliding window within the monitoring sequence to extract the time-lag subsequence of each environmental parameter in each cycle;

[0009] Determine the trend synchronization degree between each environmental parameter and the air conditioning load in each cycle by combining the trend difference and correlation between all load data in the adjacent time period in each cycle and all data in the time-lagged subsequence, and the lag time;

[0010] Analyze the causal relationship between the time-lagged subsequences of any two environmental parameters in each period, as well as the difference in trend synchronization, to determine the coupling coefficient of the any two environmental parameters in each period;

[0011] Determine a reference evaluation value between the current cycle and the remaining cycles based on the relevant changes and differences in load data between the current cycle and the remaining cycles, combined with the differences in all coupling coefficients; and obtain each reference cycle based on the reference evaluation value;

[0012] Based on the load sequence of each reference cycle, the load sequence of the current cycle and the monitoring sequence of each environmental parameter, a load forecasting model is constructed by combining the coupling coefficient and the neural network model.

[0013] Preferably, the process of obtaining the lag time is:

[0014] Using a cross-correlation function, the cross-correlation coefficients corresponding to multiple lag values between the monitoring sequence of each environmental parameter and the load sequence in each period are calculated;

[0015] The lag value corresponding to the maximum cross-correlation coefficient is used as the lag time between each environmental parameter and air-conditioning load in each cycle.

[0016] Preferably, the extraction process of the time-lag subsequence of each environmental parameter in each cycle is: constructing a sliding window of a preset size, starting from the corresponding position of the adjacent time period in the monitoring sequence of each environmental parameter in each cycle, and sliding forward according to a preset sliding step, wherein the number of slides is the lag time, and all elements in the sliding window of the last slide form a time-lag subsequence.

[0017] Preferably, determining the trend synchronization degree between each environmental parameter and the air conditioning load in each cycle includes:

[0018] The load data at all moments in the adjacent time periods under each cycle are combined into a load subsequence;

[0019] Analyzing the difference between the fitting slope of the time-lag subsequence and the fitting slope of the load subsequence for each environmental parameter in each period, and calculating the trend difference;

[0020] Calculating the similarity between the time lag subsequence and the load subsequence; calculating the cumulative sum of the lag time and the trend difference;

[0021] The trend synchronization degree is the ratio of the similarity degree to the cumulative sum.

[0022] Preferably, the calculating trend difference includes:

[0023] Performing a linear fit on the time-delay subsequence, and recording the slope of the fitted line as the first slope;

[0024] Performing a linear fit on the data at all moments in the load subsequence, and recording the slope of the fitted line as the second slope;

[0025] The difference between the first slope and the second slope is recorded as the trend difference.

[0026] Preferably, determining the coupling coefficient of any two environmental parameters in each cycle includes:

[0027] Performing a causal test on the time-delay subsequence between the arbitrary two environmental parameters, and calculating a causal value of the arbitrary two environmental parameters;

[0028] Calculating the average value and the difference of the trend synchronization between the two arbitrary environmental parameters; calculating the ratio of the average value to the difference, which is recorded as a relative ratio;

[0029] The coupling coefficient is the product of the causal value and the relative ratio.

[0030] Preferably, the causal value is calculated as follows:

[0031] Performing a Granger causality test on the time-lagged subsequences between any two environmental parameters in each period to obtain two test statistics;

[0032] The product of the maximum value of the two test statistics and the mean value of the two test statistics is used as the causal value of the arbitrary two environmental parameters.

[0033] Preferably, determining the reference evaluation value between the current cycle and the remaining cycles includes:

[0034] Calculating the correlation coefficient of the load sequence between the current cycle and the remaining cycles, and recording the sum of the correlation coefficient and a preset value as the correlation degree;

[0035] Calculate the mean of all elements in the load subsequence in each cycle and record it as the average load; record the difference between the average load of the current cycle and the remaining cycles as the load difference;

[0036] Calculating the metric distance of the coupling coefficients of all the arbitrary two environmental parameters between the current cycle and the remaining cycles;

[0037] The sum of the load difference and the metric distance is calculated; and the reference evaluation value is a ratio of the correlation to the sum.

[0038] Preferably, a further method for obtaining each reference period is: obtaining a segmentation threshold of the reference evaluation value between the current period and all other periods; and recording the period corresponding to the reference evaluation value being greater than the segmentation threshold as each reference period.

[0039] Preferably, the load forecasting model is constructed as follows:

[0040] The average value of the coupling coefficient between each environmental parameter and all other environmental parameters in each cycle is used as the global coupling degree of each environmental parameter in each cycle;

[0041] The load sequences of all reference cycles, the load sequences of the current cycle, the monitoring sequences of each environmental parameter and the global coupling degree are input into the neural network model for training to construct a load forecasting model.

[0042] This application has at least the following beneficial effects:

[0043] This application collects air-conditioning load data through simulation, analyzes the lag between the data of each environmental parameter and the load data in each cycle, and obtains the lag time. Its beneficial effect is that it takes into account the lag of different environmental parameters and air-conditioning load; secondly, it extracts the time-lag subsequence of each environmental parameter in each cycle, and determines the trend synchronization between each environmental parameter and the air-conditioning load in each cycle. Its beneficial effect is that it eliminates the lag effect of each environmental parameter on the air-conditioning load, quantifies the influence of a single environmental parameter on the air-conditioning load, and avoids the prediction deviation caused by ignoring the lag; determines the coupling coefficient of any two environmental parameters in each cycle, and its beneficial effect is that it takes into account the coupling effect between different environmental parameters, can comprehensively capture the dynamic coupling relationship between different environmental parameters and their comprehensive influence on the air-conditioning load, and improves The prediction accuracy of air-conditioning load in complex scenarios with multiple environmental factors; determining the reference evaluation value between the current cycle and the remaining cycles; obtaining each reference cycle based on the reference evaluation value, which has the beneficial effect of considering the similar changes between the current cycle and the remaining cycles, and then selecting the remaining cycles with environmental conditions similar to the current cycle, and then referring to the prediction of subsequent load data. By constructing a load prediction model based on the load sequence of each reference cycle, the load sequence of the current cycle and the monitoring sequence of each environmental parameter, combined with the coupling coefficient and the neural network model, the beneficial effect is that it can quantify the hysteresis and coupling relationship of different environmental parameters, avoid prediction deviation caused by ignoring the interaction of multiple environmental parameters, and at the same time, use the load data of the reference cycle for prediction reference, which can improve the prediction accuracy of air-conditioning load. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The following is a detailed description of a computer-aided air conditioning load prediction model simulation construction method of the present application in conjunction with the accompanying drawings.

[0045] Figure 1 A flowchart of a computer-aided air conditioning load prediction model simulation construction method provided in an embodiment of the present application;

[0046] Figure 2 A flowchart of the steps of a method for obtaining the coupling coefficient of any two environmental parameters in each cycle provided in an embodiment of the present application;

[0047] Figure 3 A flowchart of the steps of a method for obtaining reference evaluation values between the current cycle and the remaining cycles provided in an embodiment of the present application. DETAILED DESCRIPTION

[0048] To make the objectives, technical solutions, and advantages of this application more clearly understood, the following, in conjunction with the accompanying drawings and implementation examples, further details are provided on a computer-aided simulation method for constructing an air conditioning load forecasting model. It should be understood that the specific embodiments described herein are intended only to explain this application and are not intended to limit this application.

[0049] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0050] See also Figure 1 , which shows a flowchart of a computer-aided air conditioning load prediction model simulation construction method provided by an embodiment of the present application, the method comprising the following steps:

[0051] Step 1: Collect data of each environmental parameter in the building at all times in each cycle in real time to form a monitoring sequence; simulate the energy consumption of the air conditioner to obtain the load data of the air conditioner load at all times in each cycle and form a load sequence.

[0052] The temperature and humidity sensors and wind speed sensors are used to collect data of each environmental parameter in the real building environment at different times in real time. The environmental parameters include indoor temperature, outdoor temperature, indoor humidity, outdoor humidity, and outdoor wind speed;

[0053] In this embodiment, the time interval for data collection is 1 minute, and the collection cycle is one day. As other implementation methods, the implementer can set it according to actual conditions.

[0054] Therefore, the data of each environmental parameter in the building at all times in each cycle are combined to form a monitoring sequence for each environmental parameter in each cycle.

[0055] Secondly, the RC (Resistor-Capacitor Network) model can better reflect the thermophysical properties of buildings. It analogizes the thermal resistance R and stored heat capacity C of building envelopes, such as walls, to the resistor R and capacitor C in the electrical field. By decomposing the building structure into multiple nodes, each representing a building part such as a wall, window, roof, or interior space, the model simulates heat conduction and heat capacity effects using thermal resistance and thermal capacitance. Thermal resistance represents the resistance to heat transfer, while thermal capacitance represents the ability to store heat. Therefore, by analyzing the building envelope and material properties, and collecting real-world environmental parameters in real time, an RC thermal network model is constructed. A building simulation scenario is created using energy consumption modeling software. Based on the output of the RC thermal network model, the temperature and heat flow at different building nodes are set, and the building's air conditioning energy consumption is simulated. Real-time air conditioning load data is obtained during the simulation, resulting in air conditioning load data at different times within each cycle, forming a load sequence for each cycle.

[0056] In this embodiment, the collection time interval of the air conditioner load data is 1 minute, and the collection cycle is one day. Secondly, the energy consumption modeling software uses Design Builder for simulation.

[0057] It should be noted that the construction of the RC thermal network model and the process of performing simulation using the energy consumption modeling software are well-known technologies and will not be described in detail here.

[0058] Secondly, all collected data are normalized. In this embodiment, the maximum and minimum normalization method is used for normalization. The maximum and minimum normalization method is a well-known technology and will not be described here. As other implementation methods, the implementer can adopt other methods of the existing technology, such as the Z-score normalization method, etc. This embodiment does not impose any special restrictions on this.

[0059] At this point, the load sequence of each cycle in the building and the monitoring sequence of each environmental parameter are obtained.

[0060] Step 2: Analyze the correlation between the lag delay between the monitoring sequence of each environmental parameter in each cycle and the load sequence to obtain the lag duration; divide all moments in each cycle into multiple time periods, and record the time periods in the remaining cycles corresponding to the time period of the current moment in the current cycle as adjacent time periods; based on the adjacent time periods and the lag duration, slide the sliding window in the monitoring sequence to extract the time-lag subsequence of each environmental parameter in each cycle; determine the trend synchronization between each environmental parameter and the air-conditioning load in each cycle through the trend difference and correlation between all load data in the adjacent time periods in each cycle and all data in the time-lag subsequence, combined with the lag duration.

[0061] Because buildings exhibit thermal inertia, changes in environmental factors like outdoor temperature, humidity, and wind speed require a period of heat transfer and accumulation before gradually affecting indoor temperatures. Air conditioners then adjust their cooling strategies based on these changes. Consequently, changes in air conditioning load do not respond instantly to changes in these influencing factors, but rather experience a time delay. This means that changes in air conditioning load lag behind those in various environmental parameters.

[0062] Secondly, because the speed of heat transfer and accumulation varies under different weather conditions, the hysteresis effect of external environmental factors on air conditioning load under different weather conditions is also inconsistent. Therefore, in order to improve the prediction accuracy of load data, it is necessary to analyze the hysteresis effect of different environmental parameters on air conditioning load data, specifically:

[0063] Using a cross-correlation function, the cross-correlation coefficients corresponding to multiple lag values between the monitoring sequence of each environmental parameter and the load sequence in each period are calculated;

[0064] It should be noted that the cross-correlation function is a well-known technology and will not be described in detail here. The lag values are calculated as follows: The cross-correlation coefficient when , as another implementation method, can be set by the implementer according to the actual situation.

[0065] The lag value corresponding to the maximum cross-correlation coefficient is used as the lag time between each environmental parameter and air conditioning load in each cycle;

[0066] It should be noted that the lag period can reflect the time lag between the change in the air-conditioning load and each environmental parameter; secondly, the lag period can also reflect the influence of different environmental parameters on the air-conditioning load. The smaller the lag period, the shorter the time it takes for the air-conditioning load to change after the environmental parameter changes, which reflects that the impact of the environmental parameter on the air-conditioning load is more direct and the greater the impact.

[0067] Furthermore, the data between the air conditioning load and each environmental parameter is matched by the hysteresis time, specifically:

[0068] Divide all moments in each cycle into multiple time periods;

[0069] In this embodiment, every hour is regarded as a time period, and each cycle is divided into 24 time periods.

[0070] The time period in each cycle that corresponds to the time period of the current moment in the current cycle is recorded as the adjacent time period;

[0071] It should be noted that, for ease of understanding, it is assumed that the time period of the current moment in the current cycle is 09:00-10:00, each cycle represents a day, and the time period of 09:00-10:00 in each cycle is regarded as the adjacent time period of each cycle;

[0072] Constructing a sliding window of a preset size, starting from the position corresponding to the adjacent time period in the monitoring sequence of each environmental parameter in each cycle, sliding forward according to a preset sliding step size, wherein the number of slides is the lag time, and all elements in the sliding window of the last slide are combined into a time-lag subsequence of each environmental parameter in each cycle;

[0073] In this embodiment, the time length of the sliding window is 1 hour, that is, one period; the preset sliding step is 1. As other implementation methods, the implementer can set it according to actual conditions.

[0074] It should be noted that the time-lagged subsequence has temporal matching and consistency with the load data in the adjacent time period. The time-lagged subsequence can provide more accurate environmental factor data for the subsequent load forecasting model and correct the forecast deviation caused by delayed conduction.

[0075] Furthermore, the relative changes between all the data in the time-lagged subsequence and all the load data of the air-conditioning load in the adjacent period are analyzed, and the trend synchronization degree is calculated, specifically:

[0076] Performing a linear fit on the time-delay subsequence, and recording the slope of the fitted line as the first slope;

[0077] The load data of the air-conditioning load at all times in the adjacent time periods under each cycle are used to form a load subsequence;

[0078] Performing a linear fit on the data at all moments in the load subsequence, and recording the slope of the fitted line as the second slope;

[0079] Recording the difference between the first slope and the second slope as the trend difference;

[0080] In this embodiment, the least square method is used for linear fitting, wherein the least square method is a well-known technology and is not described here in detail. Secondly, the absolute value of the difference between the first slope and the second slope is calculated and recorded as the trend difference.

[0081] calculating the similarity between the time-delay subsequence and the load subsequence;

[0082] In this embodiment, the similarity is measured by calculating the mutual information coefficient between the time-delay subsequence and the load subsequence. The calculation of the mutual information coefficient is a well-known technique and will not be described in detail here.

[0083] Calculating the cumulative sum of the lag time and the trend difference, and using the ratio of the similarity to the cumulative sum as the trend synchronization degree between each environmental parameter and the air-conditioning load in each cycle;

[0084] It should be noted that the similarity reflects the strength of the nonlinear dependency between the lag subsequence and the load subsequence of the environmental parameter and the consistency of the synchronous change trend. The greater the similarity, the stronger the nonlinear dependency between the lag subsequence and the load subsequence of the environmental parameter and the more synchronous change trend they have. The trend difference reflects whether the change rates of the lag subsequence and the load subsequence of the environmental parameter are relatively consistent. The smaller the trend difference, the more consistent the change rates of the lag subsequence and the load subsequence of the environmental parameter. By introducing the lag time, the trend synchronization degree reflects the hysteresis difference effect of the change of the environmental parameter on the air-conditioning load. The greater the obtained trend synchronization degree, the more significant the effect of the environmental parameter on the air-conditioning load, the more synchronous change trend there is between the environmental parameter and the air-conditioning load, and the smaller the hysteresis effect caused by it.

[0085] At this point, the trend synchronization between each environmental parameter and air-conditioning load in each cycle is obtained.

[0086] Step 3: Analyze the causal relationship between the time-lagged subsequences of any two environmental parameters in each period, as well as the difference in trend synchronization, to determine the coupling coefficient of the any two environmental parameters in each period.

[0087] Furthermore, considering that environmental parameters not only affect air conditioning load, but also interact with each other, forming complex interactions, it is important to consider that different environmental parameters not only directly affect air conditioning load but also alter other environmental parameters such as humidity and wind speed. The interactions between these environmental parameters further affect air conditioning load. Traditional analysis methods often focus solely on the impact of a single environmental parameter on air conditioning load, ignoring the coupling and interactions between multiple environmental parameters. This makes it difficult to fully reflect the actual situation, resulting in the prediction model being unable to accurately capture the dynamic relationships between different environmental parameters, thus affecting prediction accuracy. Therefore, to improve the prediction accuracy of air conditioning load, further analysis of the coupling effects between multiple environmental parameters is necessary.

[0088] Based on the above analysis, the coupling coefficient is calculated by the coupling change of the data between different environmental parameters. The step flow chart of the method for obtaining the coupling coefficient of any two environmental parameters in each cycle provided in the embodiment of the present application is as follows: Figure 2 As shown, specifically including:

[0089] A causal test is performed on the time-lagged subsequences between any two environmental parameters in each period, and two test statistics are obtained;

[0090] In this embodiment, a Granger causality test algorithm is used to perform a causal test. Assuming that the two environmental parameters are outdoor temperature and outdoor humidity, the time-lagged subsequence corresponding to the outdoor temperature is used as the dependent variable, and the time-lagged subsequence corresponding to the outdoor humidity is used as the independent variable. A Granger causality test is performed to calculate the test statistic. Then, the time-lagged subsequence corresponding to the outdoor temperature is used as the independent variable, and the time-lagged subsequence corresponding to the outdoor humidity is used as the dependent variable. A Granger causality test is performed to calculate the test statistic, thereby obtaining two test statistics. It should be noted that the Granger causality test algorithm is a well-known technology and will not be described in detail here.

[0091] The product of the maximum value of the two test statistics and the mean value of the two test statistics is used as the causal value of the arbitrary two environmental parameters in each period;

[0092] It should be noted that the larger the causal value is, the stronger the coupling relationship between the time-delay subsequences of any two environmental parameters is. For the two environmental parameters of outdoor temperature and outdoor humidity, the greater the coupling effect between outdoor temperature and outdoor humidity is.

[0093] Calculating the average value of the trend synchronization between any two environmental parameters in each period;

[0094] Calculating the difference in trend synchronization between any two environmental parameters in each period;

[0095] In this embodiment, the absolute value of the difference between the trend synchronization degrees of the arbitrary two environmental parameters and the air-conditioning load parameter in each cycle is calculated.

[0096] Calculating the ratio of the average value to the difference value, which is recorded as a relative ratio; multiplying the causal value by the relative ratio as the coupling coefficient of the arbitrary two environmental parameters in each cycle;

[0097] In this embodiment, the Taking the two environmental parameters of outdoor temperature and outdoor humidity under a cycle as an example, the calculation formula of the coupling coefficient is:

[0098]

[0099] in, is the coupling coefficient between outdoor temperature and outdoor humidity, is the causal value between outdoor temperature and outdoor humidity, is the average value of the trend synchronization between outdoor temperature and outdoor humidity, is the trend synchronization degree between outdoor temperature and air conditioning load parameters, is the trend synchronization degree between outdoor humidity and air conditioning load parameters, To preset a value greater than 0, avoid the denominator being 0, The value range is , in this embodiment, The value is 0.01. As other implementation methods, the implementer can set it according to actual conditions.

[0100] It should be noted that, the larger the average value and the smaller the difference, the greater the relative contrast obtained, indicating that the joint influence of the arbitrary two environmental parameters on the air-conditioning load is more significant, that is, the arbitrary two environmental parameters are synchronized with the change of the air-conditioning load, and the coupling coefficient reflects the comprehensive coupling strength of the arbitrary two environmental parameters on the air-conditioning load under the condition of time lag. The larger the coupling coefficient, the more significant the interaction relationship between the arbitrary two environmental parameters, and the more significant the impact on the air-conditioning load under the action of coupling.

[0101] At this point, the coupling coefficient of any two environmental parameters in each cycle is obtained.

[0102] Step 4: Determine the reference evaluation value between the current cycle and the remaining cycles based on the relevant changes and differences in load data between the current cycle and the remaining cycles, combined with the differences in all coupling coefficients; and obtain each reference cycle based on the reference evaluation value.

[0103] Furthermore, by analyzing the change of the air conditioning load parameter data in the adjacent time periods between the current cycle and the remaining cycles, combined with the coupling coefficient, the reference evaluation value is calculated. The step flow chart of the method for obtaining the reference evaluation value between the current cycle and the remaining cycles provided in the embodiment of the present application is as follows: Figure 3 As shown, specifically including:

[0104] Calculating the correlation coefficient of the load sequence between the current cycle and the remaining cycles, and recording the sum of the correlation coefficient and a preset value as the correlation degree;

[0105] In this embodiment, the correlation coefficient is measured by calculating the Pearson correlation coefficient of the load sequence between the current cycle and the remaining cycles. The calculation of the Pearson correlation coefficient is a well-known technique and will not be described in detail here. As other implementation methods, the implementer may adopt other methods of the prior art, such as the Spearman correlation coefficient, the cosine similarity, etc. This embodiment does not impose any special restrictions on this. Secondly, the preset value is 1. Since the value range of the correlation coefficient is By adding the preset value 1, the range of the correlation degree is .

[0106] It should be noted that the correlation reflects whether the overall change trend of the air-conditioning load between the current cycle and the remaining cycles is consistent. The greater the correlation, the more consistent the overall change trend of the air-conditioning load between the current cycle and the remaining cycles.

[0107] Calculate the mean of all elements in the load subsequence in each period and record it as the average load;

[0108] The difference between the average load of the current cycle and the average load of the remaining cycles is recorded as the load difference;

[0109] Calculating the metric distance of the coupling coefficients of all the arbitrary two environmental parameters between the current cycle and the remaining cycles;

[0110] In this embodiment, the distance is measured by calculating the Euclidean distance of the coupling coefficients of all the arbitrary two environmental parameters between the current cycle and the remaining cycles. As other implementation methods, the implementer may adopt other methods of the prior art, such as Mahalanobis distance, Manhattan distance, etc., and this embodiment does not impose any special restrictions on this.

[0111] Calculating a sum of the load difference and the metric distance, and using a ratio of the correlation to the sum as a reference evaluation value between the current cycle and the remaining cycles;

[0112] It should be noted that the smaller the load difference, the smaller the difference in air-conditioning load between the current cycle and the remaining cycles, the smaller the measurement distance, the more consistent the coupling influence of environmental conditions between the current cycle and the remaining cycles, and the closer the changing trends of environmental conditions in the two cycles are. The larger the obtained reference evaluation value, the more similar the changes in environmental parameters on air-conditioning load between the current cycle and the remaining cycles are, and the stronger the reference of data in the remaining cycles to the load forecast of the current cycle.

[0113] Furthermore, based on the reference evaluation value, the data of all cycles are screened, specifically:

[0114] A threshold segmentation algorithm is used to obtain a segmentation threshold of the reference evaluation value between the current cycle and all other cycles, and the cycles corresponding to the reference evaluation value being greater than the segmentation threshold are recorded as reference cycles;

[0115] In this embodiment, the Otsu threshold segmentation algorithm is used to obtain the segmentation threshold, wherein the Otsu threshold segmentation algorithm is a well-known technology and will not be described in detail here. As other implementation methods, the implementer can adopt other methods of the existing technology, such as cross-validation method, etc. This embodiment does not impose any special restrictions on this.

[0116] At this point, each reference cycle is obtained.

[0117] Step 5: Based on the load sequence of each reference cycle, the load sequence of the current cycle, and the monitoring sequence of each environmental parameter, a load forecasting model is constructed by combining the coupling coefficient and the neural network model.

[0118] Furthermore, based on the coupling coefficient, the global coupling degree is calculated, specifically:

[0119] The average value of the coupling coefficient between each environmental parameter and all other environmental parameters in each cycle is used as the global coupling degree of each environmental parameter in each cycle;

[0120] It should be noted that the global coupling degree reflects the comprehensive coupling effect of each environmental parameter with the other environmental parameters, and can dynamically adapt to the changes in the coupling relationship under different scenarios and different time lags, thereby comprehensively reflecting the impact intensity of each environmental parameter on the air-conditioning load in the interaction of multiple environmental parameters.

[0121] The data of air conditioning load parameters at all times in all reference periods, the data of each environmental parameter at all times in the current period, the global coupling degree of each environmental parameter, and the data of air conditioning load parameters at all times in the current period are input into the neural network model for training to construct a load forecasting model;

[0122] In this embodiment, a BP neural network model is used for training, wherein the training of the BP neural network model is a well-known technology and will not be described in detail here. As other implementation methods, the implementer can adopt other methods of the existing technology, such as the long short-term memory network model, etc. This embodiment does not impose special restrictions on this; secondly, the input layer in the BP neural network model has a total of 12 input nodes, each node corresponds to an environmental parameter and the global coupling degree, the hidden layer has 5 layers, and the output layer has 60 nodes, corresponding to the load data within a time period of the output, that is, the load data within the next 1 hour is output; secondly, the optimization algorithm of the BP neural network model is the SGD algorithm, the loss function is MSE, and the number of iterations is 60 times.

[0123] The load data of the air conditioner is predicted through the load forecasting model, and the load data of the air conditioner energy consumption is simulated by the energy consumption modeling software and compared with the predicted load data. The prediction accuracy of the load forecasting model is evaluated by the mean absolute error, thereby verifying the effectiveness of the load forecasting model and realizing accurate prediction and simulation verification of the air conditioner load.

[0124] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 1At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0125] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0126] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the present application. It should be noted that a person skilled in the art can make various modifications and improvements without departing from the spirit of the present application. Therefore, any simple modifications, equivalent variations, and modifications to the above embodiments made in accordance with the technical essence of the present application without departing from the content of the present application's technical solution fall within the scope of protection of the present application's technical solution.

Claims

1. A computer-aided air conditioning load prediction model simulation construction method, characterized in that: The method comprises the following steps: Collect data of each environmental parameter in the building at all times in each cycle in real time to form a monitoring sequence; simulate the energy consumption of air conditioners to obtain the load data of air conditioners at all times in each cycle and form a load sequence; Analyze the hysteresis delay between the monitoring sequence of each environmental parameter and the load sequence in each cycle to obtain the hysteresis duration; Divide all moments in each cycle into multiple time periods, and record the time periods in other cycles that correspond to the time period of the current moment in the current cycle as adjacent time periods; based on the adjacent time periods and the lag duration, slide a sliding window within the monitoring sequence to extract the time-lag subsequence of each environmental parameter in each cycle; Determine the trend synchronization degree between each environmental parameter and the air conditioning load in each cycle by combining the trend difference and correlation between all load data in the adjacent time period in each cycle and all data in the time-lagged subsequence, and the lag time; Analyze the causal relationship between the time-lagged subsequences of any two environmental parameters in each period, as well as the difference in trend synchronization, to determine the coupling coefficient of the any two environmental parameters in each period; Determine a reference evaluation value between the current cycle and the remaining cycles based on the relevant changes and differences in load data between the current cycle and the remaining cycles, combined with the differences in all coupling coefficients; and obtain each reference cycle based on the reference evaluation value; Based on the load sequence of each reference cycle, the load sequence of the current cycle and the monitoring sequence of each environmental parameter, a load forecasting model is constructed by combining the coupling coefficient and the neural network model.

2. The computer-aided air conditioning load prediction model simulation construction method according to claim 1, characterized in that: The process of obtaining the lag time is as follows: Using a cross-correlation function, the cross-correlation coefficients corresponding to multiple lag values between the monitoring sequence of each environmental parameter and the load sequence in each period are calculated; The lag value corresponding to the maximum cross-correlation coefficient is used as the lag time between each environmental parameter and air-conditioning load in each cycle.

3. The computer-aided air conditioning load prediction model simulation construction method according to claim 1, characterized in that: The extraction process of the time-lag subsequence of each environmental parameter in each cycle is as follows: constructing a sliding window of a preset size, starting from the corresponding position of the adjacent time period in the monitoring sequence of each environmental parameter in each cycle, sliding forward according to a preset sliding step, wherein the number of sliding times is the lag time, and all elements in the sliding window of the last sliding are combined into a time-lag subsequence.

4. The computer-aided air conditioning load prediction model simulation construction method according to claim 1, characterized in that: Determining the trend synchronization between each environmental parameter and the air conditioning load in each cycle includes: The load data at all moments in the adjacent time periods under each cycle are combined into a load subsequence; Analyzing the difference between the fitting slope of the time-lag subsequence and the fitting slope of the load subsequence for each environmental parameter in each period, and calculating the trend difference; Calculating the similarity between the time lag subsequence and the load subsequence; calculating the cumulative sum of the lag time and the trend difference; The trend synchronization degree is the ratio of the similarity degree to the cumulative sum.

5. The computer-aided air conditioning load prediction model simulation construction method according to claim 4, characterized in that: The calculated trend difference includes: Performing a linear fit on the time-delay subsequence, and recording the slope of the fitted line as the first slope; Performing a linear fit on the data at all moments in the load subsequence, and recording the slope of the fitted line as the second slope; The difference between the first slope and the second slope is recorded as the trend difference.

6. The computer-aided air conditioning load prediction model simulation construction method according to claim 1, characterized in that: Determining the coupling coefficient of any two environmental parameters in each cycle includes: Performing a causal test on the time-delay subsequence between the arbitrary two environmental parameters, and calculating a causal value of the arbitrary two environmental parameters; Calculating the average value and the difference of the trend synchronization between the two arbitrary environmental parameters; calculating the ratio of the average value to the difference, which is recorded as a relative ratio; The coupling coefficient is the product of the causal value and the relative ratio.

7. The computer-aided air conditioning load prediction model simulation construction method according to claim 6, characterized in that: The calculation method of the causal value is: Performing a Granger causality test on the time-lagged subsequences between any two environmental parameters in each period to obtain two test statistics; The product of the maximum value of the two test statistics and the mean value of the two test statistics is used as the causal value of the arbitrary two environmental parameters.

8. The computer-aided air conditioning load prediction model simulation construction method according to claim 4, characterized in that: Determining the reference evaluation value between the current cycle and the remaining cycles includes: Calculating the correlation coefficient of the load sequence between the current cycle and the remaining cycles, and recording the sum of the correlation coefficient and a preset value as the correlation degree; Calculate the mean of all elements in the load subsequence in each cycle and record it as the average load; record the difference between the average load of the current cycle and the remaining cycles as the load difference; Calculating the metric distance of the coupling coefficients of all the arbitrary two environmental parameters between the current cycle and the remaining cycles; The sum of the load difference and the metric distance is calculated; and the reference evaluation value is a ratio of the correlation to the sum.

9. The computer-aided air conditioning load forecasting model simulation construction method according to claim 1, characterized in that: A further method for obtaining each reference period is: obtaining a segmentation threshold of the reference evaluation value between the current period and all other periods; and recording the period corresponding to the reference evaluation value being greater than the segmentation threshold as each reference period.

10. The computer-aided air conditioning load forecasting model simulation construction method according to claim 1, characterized in that: The construction process of the load forecasting model: The average value of the coupling coefficient between each environmental parameter and all other environmental parameters in each cycle is used as the global coupling degree of each environmental parameter in each cycle; The load sequences of all reference cycles, the load sequences of the current cycle, the monitoring sequences of each environmental parameter and the global coupling degree are input into the neural network model for training to construct a load forecasting model.

Images