Air conditioner resource allocation method and system based on cold accumulation prediction algorithm

Abstract

The invention discloses an air conditioner resource allocation method and system based on a cold accumulation prediction algorithm, and relates to the field of air conditioner cold accumulation, and the method comprises the steps: collecting historical load operation data of an air conditioner, analyzing the historical load operation data through employing a grey relational analysis algorithm, and constructing a cold load prediction model in combination with the cold accumulation prediction algorithm; training the cold load prediction model by using a back propagation neural network algorithm, inputting load operation data collected in real time into the trained cold load prediction model, and outputting a cold load prediction result; and in combination with the cooling load prediction result and the electricity price information, an air conditioner resource allocation strategy is formulated, and a model prediction control algorithm is used for optimizing the allocation strategy. Through integration of the load prediction technology and the real-time electricity price information, the cold load demand of the whole day can be accurately predicted, energy waste is reduced, the energy utilization efficiency is improved, the dynamic electricity price market can be responded, and economic maximization is achieved.

Description

Technical Field

[0001] The present invention relates to the field of air conditioning cold storage, and in particular to an air conditioning resource allocation method and system based on a cold storage prediction algorithm. Background Art

[0002] With rapid socioeconomic development and the continuous expansion of electricity consumption, power load management has become a key issue in ensuring the stable operation of the power grid. The widespread use of air conditioning systems, particularly during peak summer electricity consumption, places significant pressure on the power grid. Air conditioning energy consumption accounts for a significant proportion of total electricity consumption, and due to its operating characteristics, it often generates significant peak loads during peak hours, further increasing the pressure on the power grid. Therefore, effectively managing the energy consumption of air conditioning systems is not only a key measure for energy conservation and environmental protection, but also an effective means of ensuring the safe operation of the power grid.

[0003] In recent years, the power industry has actively sought new solutions, particularly exploring flexible control strategies to optimize air conditioning system operations. The core of these flexible control strategies lies in leveraging advanced information technology and data analysis tools, such as artificial intelligence (AI) and the Internet of Things (IoT), to achieve precise management and flexible scheduling of air conditioning loads. Specifically, by applying technologies such as air conditioning cooling plant efficiency optimization and ice storage, decentralized air conditioning resources can be effectively tapped and integrated, enabling them to participate in the grid's demand response mechanism, thereby smoothing the load curve and alleviating grid pressure.

[0004] As an innovative energy-saving technology, ice storage air conditioning systems store cooling energy at night, when electricity prices are lower, and release it during peak daytime hours. This not only significantly reduces the operating costs of air conditioning systems, but also helps to rationally distribute the power load and improve the operating efficiency of the entire power grid. However, to fully realize the potential of ice storage air conditioning systems, the key lies in accurately predicting and controlling their cooling process. Although various cooling load prediction methods have been proposed, due to the complexity and dynamic nature of the air conditioning system operating environment, existing prediction models often suffer from insufficient accuracy. In addition, how to coordinate the operating conditions of the various components within the air conditioning system to achieve optimal overall performance is also a technical challenge that needs to be solved urgently.

[0005] Currently, no effective solutions have been proposed for the problems in related technologies. Summary of the Invention

[0006] In response to the problems in the related art, the present invention proposes an air conditioning resource allocation method and system based on a cold storage prediction algorithm to overcome the above-mentioned technical problems existing in the existing related art.

[0007] To this end, the specific technical solutions adopted in the present invention are as follows:

[0008] According to one aspect of the present invention, a method for allocating air conditioning resources based on a cold storage prediction algorithm is provided, the method comprising the following steps:

[0009] S1. Collect historical load operation data of air conditioners, analyze the historical load operation data using the grey correlation analysis algorithm, and build a cooling load prediction model by combining it with the cold storage prediction algorithm;

[0010] S2. Use the back propagation neural network algorithm to train the cooling load prediction model, input the real-time collected load operation data into the trained cooling load prediction model, and output the cooling load prediction result;

[0011] S3. Combine cooling load forecast results with electricity price information to formulate an allocation strategy for air conditioning resources, and use the model predictive control algorithm to optimize the allocation strategy.

[0012] Furthermore, the historical load operation data of the air conditioner is collected, and the historical load operation data is analyzed using the grey correlation analysis algorithm. In combination with the cold storage prediction algorithm, a cooling load prediction model is constructed, which includes the following steps:

[0013] S11. Collect historical load operation data of the air conditioner, and perform feature selection on the historical load operation data to obtain feature data;

[0014] S12. Preprocessing the characteristic data using a grey relational analysis algorithm to obtain input features for a cooling load prediction model;

[0015] S13. Combining input characteristics with the cold storage prediction algorithm, a cooling load prediction model is constructed.

[0016] Furthermore, the grey relational analysis algorithm is used to pre-process the characteristic data to obtain the input features for the cooling load prediction model, which includes the following steps:

[0017] S121, performing positive processing on the feature data;

[0018] S122, performing dimensionless processing on the feature data after the forward processing, and selecting a mother sequence from the feature data after the dimensionless processing;

[0019] S123. Based on the selected parent sequence, the remaining feature data is used as a subsequence, and the correlation coefficient between each subsequence and the parent sequence is calculated using a grey correlation analysis algorithm, and the grey correlation degree is calculated based on the correlation coefficient;

[0020] S124. Determine the weight of the feature data using a hierarchical analysis algorithm, and select input features for the cooling load prediction model in combination with the grey correlation degree.

[0021] Furthermore, combining the input features with the cold storage prediction algorithm, building a cooling load prediction model includes the following steps:

[0022] S131. According to the cold storage prediction algorithm, the number of hidden layer nodes is set, and the parameters of the hidden layer nodes are randomly initialized to construct an initial cooling load prediction model;

[0023] S132, using a preset activation function, mapping the input features to the hidden layer of the initial cooling load prediction model, and calculating the output value of the hidden layer;

[0024] S133. Based on the output value of the hidden layer, the least squares algorithm is used to determine the weight of the output layer, and the parameters of the initial cooling load prediction model are optimized to obtain the cooling load prediction model.

[0025] Furthermore, the parameters of the hidden layer nodes include: input weights and biases of the hidden layer nodes.

[0026] Furthermore, the cooling load prediction model is trained using a back propagation neural network algorithm, and the real-time collected load operation data is input into the trained cooling load prediction model. Outputting the cooling load prediction result includes the following steps:

[0027] S21. Design a back propagation neural network structure based on the cooling load forecast demand, and initialize the weights and threshold parameters of the back propagation neural network;

[0028] S22. Optimizing the initial weights and threshold parameters of the back propagation neural network using a particle swarm optimization algorithm, and training the cooling load prediction model in combination with the back propagation neural network algorithm;

[0029] S23. Collect load operation data in real time and pre-process the load operation data;

[0030] S24: input the pre-processed load operation data into the trained cooling load prediction model, and output the cooling load prediction result.

[0031] Furthermore, the particle swarm optimization algorithm is used to optimize the initial weights and threshold parameters of the back propagation neural network, and the back propagation neural network algorithm is combined to train the cooling load prediction model, which includes the following steps:

[0032] S221, initialize the particle swarm and randomly generate the initial position and velocity of each particle. The position represents the weight and threshold parameters of the back propagation neural network.

[0033] S222. Using historical load operation data, train the back propagation neural network, evaluate the fitness of the particle through the prediction error, and use the error value as the fitness value of the particle;

[0034] S223, comparing the fitness value of the particle with the corresponding historical optimal fitness value, and updating the individual optimal position of the particle based on the comparison result;

[0035] S224, traverse all particles, select the particle with the highest fitness value, and set the particle with the highest fitness value as the optimal position of the group;

[0036] S225. Adjust the speed and position of each particle according to the speed update formula of the particle swarm algorithm, and train the cooling load prediction model through the back propagation algorithm based on the adjusted speed and position;

[0037] S226: Repeat steps S222 to S225 until a preset stop condition is reached, thereby obtaining a trained cooling load prediction model.

[0038] Furthermore, combining the cooling load forecast results with electricity price information, formulating an air conditioning resource allocation strategy and using the model predictive control algorithm to optimize the allocation strategy includes the following steps:

[0039] S31. Obtain electricity price information from the energy trading platform, analyze the changing trends of electricity prices in different time periods, and obtain the electricity price differences in different time periods;

[0040] S32. Based on the electricity price differences and cooling load forecast results for different time periods, a preliminary strategy for allocating air conditioning resources for each time period is formulated;

[0041] S33. Use the model predictive control algorithm to optimize the allocation strategy of air-conditioning resources in each time period.

[0042] Furthermore, the model predictive control algorithm is used to optimize the allocation strategy of air conditioning resources in each time period, including the following steps:

[0043] S331. Select a model predictive control algorithm based on the cooling load prediction results and control requirements, and establish a model predictive control framework based on the model predictive control algorithm;

[0044] S332. Taking the initially formulated air conditioning resource allocation strategy for each time period as input, combined with the model predictive control framework, and performing hourly rolling optimization calculations, the optimal allocation strategy for air conditioning resources for each time period is obtained.

[0045] According to another aspect of the present invention, an air conditioning resource allocation system based on a cold storage prediction algorithm is also provided, the system comprising a model building module, a model training and prediction module, and a strategy formulation and optimization module;

[0046] The model building module is used to collect historical load operation data of air conditioners, analyze the historical load operation data using the grey correlation analysis algorithm, and build a cooling load prediction model in combination with the cold storage prediction algorithm;

[0047] The model training and prediction module is used to train the cooling load prediction model using the back propagation neural network algorithm, input the real-time collected load operation data into the trained cooling load prediction model, and output the cooling load prediction results;

[0048] The strategy formulation and optimization module is used to combine the cooling load forecast results with electricity price information to formulate the allocation strategy of air conditioning resources and optimize the allocation strategy using the model predictive control algorithm.

[0049] Furthermore, the grey correlation analysis algorithm is used to analyze the historical load operation data, and then it is also included to determine whether to build a cooling load prediction model. The specific process is: obtaining the actual operation data of the designated equipment in the air conditioner within a preset time period; pre-allocating the designated equipment in the air conditioner based on the obtained actual operation data to obtain a cooling load distribution compliance index; judging whether the obtained cooling load distribution compliance index is within the cooling load distribution compliance threshold range, if so, completing the pre-allocation and combining the cold storage prediction algorithm to build a cooling load prediction model, otherwise adjusting the pre-allocation parameters and re-pre-allocation; the actual operation data includes cooling load amount, operation time and operation energy consumption; the cooling load distribution compliance index is used to evaluate the degree of match between the cooling load pre-allocation scheme of the designated equipment in the air conditioner and the actual operation situation; the pre-allocation parameters include the cooling load allocation ratio and the peak energy consumption of the designated equipment.

[0050] Furthermore, the specific steps for obtaining the cooling load distribution compliance index are: obtaining the actual cooling load, actual operating time and actual operating energy consumption of the specified equipment in the air conditioner within a preset time period, and combining the corresponding target value and the maximum allowable deviation value to obtain the cooling load compliance score, the operating time compliance score and the operating energy consumption compliance score; obtaining a preset ambient temperature influencing factor from a preset database, combining the obtained cooling load compliance score, the operating time compliance score and the operating energy consumption compliance score with the coupled inverse proportional processing results to obtain the cooling load distribution compliance index; the cooling load compliance score represents the ratio of the absolute value of the difference between the actual cooling load and the target cooling load to the maximum allowable cooling load deviation; the operating time compliance score represents the ratio of the absolute value of the difference between the actual operating time and the target operating time to the maximum allowable operating time deviation; the operating energy consumption compliance score represents the ratio of the absolute value of the difference between the actual operating energy consumption and the target operating energy consumption to the maximum allowable operating energy consumption deviation.

[0051] Furthermore, the method includes inputting the real-time collected load operation data into the trained cooling load prediction model, and also includes real-time monitoring of the prediction status of the cooling load prediction model to obtain a cooling load prediction evaluation value, and determining whether to output a cooling load prediction result based on the obtained cooling load prediction evaluation value; the cooling load prediction evaluation value is used to evaluate the accuracy of the prediction result of the cooling load prediction model; the cooling load prediction evaluation value is obtained by the following method: when the obtained cooling load distribution compliance index is within the cooling load distribution compliance threshold range, obtaining the parameter optimization time and determining whether the parameter optimization time is less than the preset parameter optimization time; if so, obtaining the parameter optimization time score, otherwise re-performing parameter optimization; obtaining the pre-allocation total time score and combining the obtained parameter optimization time score with the cooling load distribution compliance index to obtain the cooling load prediction evaluation value; the parameter optimization time score is the ratio of the difference between the parameter optimization time and the preset parameter optimization time to the preset parameter optimization time; the pre-allocation time score is the ratio of the absolute value of the difference between the pre-allocation total time and the pre-allocation target total time to the allowable deviation of the pre-allocation target total time.

[0052] Furthermore, the formulation of the air-conditioning resource allocation strategy also includes allocating resources to the air conditioner according to the formulated allocation strategy to obtain an allocation strategy compliance coefficient, and judging whether to execute the allocation strategy optimization based on the obtained allocation strategy compliance coefficient; the allocation strategy compliance coefficient is used to evaluate the formulation reliability of the air-conditioning resource allocation strategy; the specific steps for obtaining the allocation strategy compliance coefficient are: when the obtained cooling load prediction evaluation value is within the cooling load prediction threshold range, the actual cooling load output value and actual energy consumption output value of the specified equipment in the air conditioner at the end of the preset time period are obtained, and the cooling load output score and energy consumption output score are obtained by combining the corresponding cooling load expected output value and energy consumption expected output value respectively, and the cooling load output score and energy consumption output score are coupled with the cooling load prediction evaluation value for the results of the inverse proportional processing to obtain the allocation strategy compliance coefficient; the cooling load output score is the ratio of the difference between the expected cooling load output value and the actual cooling load output value to the expected cooling load output value; the energy consumption output score is the ratio of the difference between the actual energy consumption output value and the expected energy consumption output value to the expected energy consumption output value.

[0053] The beneficial effects of the present invention are:

[0054] 1. By integrating load forecasting technology with real-time electricity price information, the present invention can accurately predict the cooling load demand throughout the day and optimize the timing of cold storage and cooling release accordingly. Cold storage is performed during periods of low electricity prices, and the stored cold energy is used during periods of high electricity prices. This effectively reduces the operating cost of the air-conditioning system, reduces energy waste, improves energy utilization efficiency, and can respond to the dynamic electricity price market to achieve economic maximization.

[0055] 2. By utilizing historical data analysis and machine learning technology, the present invention can continuously optimize the cooling load prediction model and improve the accuracy and efficiency of the cold storage prediction algorithm. At the same time, for cooling and cold storage control, the strategy can be dynamically adjusted according to actual demand and changes in electricity prices, thereby improving the flexibility and adaptability of the system. This allows the air-conditioning system to automatically make the optimal choice under different conditions, further improving the system's operating efficiency and service quality.

[0056] 3. By comprehensively monitoring the operating status of the cold storage system and providing timely feedback, the present invention can quickly detect and respond to system anomalies, avoid potential failure risks, and ensure the stable operation of the system. At the same time, through continuous data collection and analysis, it can support long-term performance optimization and technical improvements, providing users with more reliable service guarantees.

[0057] 4. The present invention obtains the actual cooling load, actual operating hours, and actual operating energy consumption of a specified device in the air conditioner within a preset time period, and combines the corresponding target values ​​and maximum allowable deviation values ​​to obtain the cooling load compliance score, the operating hours compliance score, and the operating energy consumption compliance score. At the same time, combined with a preset ambient temperature influencing factor, the cooling load distribution compliance index is obtained, thereby achieving improved accuracy in obtaining the cooling load distribution compliance index and further achieving a more accurate assessment of the match between the cooling load pre-distribution plan and the actual operating conditions.

[0058] 5. The present invention obtains the parameter optimization duration and determines whether the parameter optimization duration is less than the preset parameter optimization duration when the obtained cooling load distribution compliance index is within the cooling load distribution compliance threshold range. If so, the parameter optimization duration score is obtained and combined with the obtained pre-allocated total duration score to obtain the cooling load prediction evaluation value. Otherwise, the parameter optimization is performed again, thereby improving the accuracy of obtaining the cooling load prediction evaluation value and further achieving a more accurate evaluation of the accuracy of the prediction results of the cooling load prediction model.

[0059] 6. The present invention obtains the actual cooling load output value and actual energy consumption output value of the specified equipment in the air conditioner at the end of the preset time period when the obtained cooling load prediction evaluation value is within the cooling load prediction threshold range, and obtains the cooling load output score and energy consumption output score in combination with the corresponding expected cooling load output value and expected energy consumption output value, respectively. At the same time, the allocation strategy compliance coefficient is obtained in combination with the obtained cooling load prediction evaluation value, thereby improving the accuracy of obtaining the allocation strategy compliance coefficient and further achieving a more accurate evaluation of the reliability of the formulation of the air conditioning resource allocation strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 is a flow chart of an air conditioning resource allocation method based on a cold storage prediction algorithm according to an embodiment of the present invention;

[0062] Figure 2 This is one of the implementation diagrams of the air conditioning resource allocation method based on the cold storage prediction algorithm according to an embodiment of the present invention;

[0063] Figure 3 This is a second implementation diagram of the air conditioning resource allocation method based on the cold storage prediction algorithm according to an embodiment of the present invention;

[0064] Figure 4 This is a third implementation diagram of the air conditioning resource allocation method based on the cold storage prediction algorithm according to an embodiment of the present invention;

[0065] Figure 5 4 is a principle block diagram of an air conditioning resource allocation system based on a cold storage prediction algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0066] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention. They are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. By referring to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and the advantages of the present invention.

[0067] According to an embodiment of the present invention, a method and system for allocating air conditioning resources based on a cold storage prediction algorithm are provided.

[0068] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to an embodiment of the present invention, an air conditioning resource allocation method based on a cold storage prediction algorithm includes the following steps:

[0069] S1. Collect historical load operation data of air conditioners, analyze the historical load operation data using the grey correlation analysis algorithm, and build a cooling load prediction model by combining it with the cold storage prediction algorithm.

[0070] It should be noted that the grey correlation analysis algorithm is used to analyze the historical load operation data, and then it is also included to determine whether to build a cooling load prediction model. The specific process is: obtain the actual operation data of the specified equipment in the air conditioner (usually including refrigerators, cold storage devices and cooling towers) within the preset time period; pre-allocate the specified equipment in the air conditioner based on the actual operation data obtained to obtain the cooling load distribution compliance index; determine whether the obtained cooling load distribution compliance index is within the cooling load distribution compliance threshold range (including the maximum and minimum values ​​of the historical cooling load distribution compliance index). If so, complete the pre-allocation and combine the cold storage prediction algorithm to build a cooling load prediction model. If the cooling load prediction model is not met, the pre-allocation parameters are adjusted and pre-allocation is performed again until the obtained cooling load distribution compliance index is within the cooling load distribution compliance threshold range. Actual operation data includes cooling load (obtained by flow sensors), operation time (obtained by time counters), and operation energy consumption (monitored in real time by energy consumption monitoring systems). Pre-allocation means initial resource allocation to designated equipment in the air conditioner based on actual operation data within a preset time period. The cooling load distribution compliance index is used to evaluate the degree of match between the cooling load pre-allocation plan and the actual operation status of designated equipment in the air conditioner. Pre-allocation parameters include cooling load allocation ratio and peak energy consumption of designated equipment.

[0071] Among them, the cooling load distribution compliance threshold range represents the range corresponding to the maximum and minimum values ​​of the historical cooling load distribution compliance index of the specified air-conditioning equipment in the preset database within the historical time period; when the obtained cooling load distribution compliance index is not within the cooling load distribution compliance threshold range, it is necessary to prompt the preset personnel to increase the cooling load ratio to obtain the limit for reducing the peak energy consumption of the specified equipment until the pre-allocation is completed when the obtained cooling load distribution compliance index is within the cooling load distribution compliance threshold range.

[0072] It should be noted that the specific steps for obtaining the cooling load distribution compliance index are as follows: obtaining the actual cooling load, actual operating time, and actual operating energy consumption of the specified equipment in the air conditioner within the preset time period, and combining the corresponding target value (acquired based on the historical load operation data of the air conditioner, including the target cooling load, target operating time, and target operating energy consumption) and the maximum allowable deviation value (including the maximum allowable cooling load deviation, the maximum allowable operating time deviation, and the maximum allowable operating energy consumption deviation) to obtain the cooling load compliance score (i.e., the H in the restriction expression of the cooling load distribution compliance index). t ), the running time meets the score (that is, the Y in the restriction expression of the cooling load distribution meets the index t ) and the operating energy consumption compliance score (i.e., the N in the cooling load distribution compliance index constraint expression t); obtain a preset ambient temperature influence factor (i.e., δ in the restriction expression of the cooling load distribution compliance index) from a preset database, and perform coupled inverse proportional processing on the obtained cooling load compliance score, operating time compliance score, and operating energy consumption compliance score to obtain the cooling load distribution compliance index; the cooling load compliance score represents the ratio of the absolute value of the difference between the actual cooling load and the target cooling load to the maximum allowable cooling load deviation; the operating time compliance score represents the ratio of the absolute value of the difference between the actual operating time and the target operating time to the maximum allowable operating time deviation; the operating energy consumption compliance score represents the ratio of the absolute value of the difference between the actual operating energy consumption and the target operating energy consumption to the maximum allowable operating energy consumption deviation.

[0073] The specific restriction expression of the cooling load distribution compliance index is:

[0074]

[0075] Where t is the number of the preset time period, t = 1, 2, ..., T, T is the total number of preset time periods, e is a natural constant, FU represents the cooling load distribution compliance index of the specified equipment in the air conditioner within the preset time period, δ represents the ambient temperature influence factor, H t Indicates the cooling load compliance score of the specified equipment in the air conditioner during the t-th preset time period, H1 t represents the actual cooling load of the specified equipment in the air conditioner during the t-th preset time period, H10 represents the target cooling load, ΔH1 max Indicates the maximum allowable cooling load deviation, Y t Indicates the compliance score of the specified equipment in the air conditioner during the t-th preset time period, Y1 t represents the actual operating time of the specified equipment in the air conditioner during the tth preset time period (recorded in real time by the time counter), Y10 represents the target operating time, ΔY1 max Indicates the maximum allowable running time deviation, N t Indicates the energy consumption compliance score of the specified equipment in the air conditioner during the tth preset time period, N1 t represents the actual operating energy consumption of the specified equipment in the air conditioner during the t-th preset time period (recorded in real time by the power sensor), N10 represents the target operating energy consumption, ΔN1 max Indicates the maximum allowable operating energy consumption deviation.

[0076] Among them, the target cooling load, target operating time and target operating energy consumption are respectively represented by the sum and average of the historical cooling load, historical operating time and historical operating energy consumption of the specified equipment in the preset database in each historical time period; the maximum allowable cooling load deviation, maximum allowable operating time deviation and maximum allowable operating energy consumption deviation are respectively represented by the sum and average of the maximum values ​​corresponding to the historical cooling load deviation, historical operating time deviation and historical operating energy consumption deviation of the specified equipment in the preset database in each historical time period.

[0077] The ambient temperature impact factor is the impact factor used in the cooling load distribution compliance index acquisition process from a preset database. It represents the numerical value corresponding to the degree of influence of the ambient temperature on the cooling load distribution compliance index. The ambient temperature impact factor for the operating area of ​​a specified device, corresponding to the cooling load distribution compliance index, can be directly obtained from the preset database. The corresponding relationship can be a pre-set mapping relationship. For example, a mapping set is formed by combining the cooling load distribution compliance index corresponding to the ambient temperature during the cooling load distribution compliance index acquisition process with the ambient temperature impact factor corresponding to the cooling load distribution compliance index in the preset database. The corresponding ambient temperature impact factor is then input into the mapping set. The mapping relationship can be a one-to-one or many-to-one relationship. In this example, the ambient temperature impact factor ranges from [0, 1].

[0078] It should be understood that the cooling load distribution compliance index decreases with the increase of the cooling load compliance score, the operating time compliance score and the operating energy consumption compliance score. The cooling load compliance score increases with the cooling load deviation (i.e., |H1 t -H10|) increases, and the running time compliance score increases with the running time deviation (i.e., |Y1 t -Y10|) increases, and the running energy consumption compliance score increases with the running energy consumption deviation (i.e. |N1 t -N10|) increases.

[0079] It's important to note that the runtime compliance score also indirectly affects the cooling load compliance score and the operating energy consumption compliance score. For example, if an air conditioning system's ideal runtime is 4 hours during a certain time period, but it actually runs for 6 hours, this will result in a runtime deviation of 2 hours, increasing the runtime compliance score. At this point, the air conditioning system may need to provide more cooling capacity to meet the additional runtime, thereby increasing the cooling load deviation (i.e., an increase in the cooling load compliance score). Similarly, due to the increase in actual runtime, the air conditioning system needs to consume more energy to maintain operation, which increases the operating energy consumption deviation (i.e., an increase in the operating energy consumption compliance score).

[0080] Through the above analysis, it can be seen that the increase in the operating hours compliance score not only directly reflects the deviation of the operating hours, but also indirectly increases the cooling load compliance score and the operating energy consumption compliance score by affecting the deviation of the cooling load and operating energy consumption. This helps to more accurately evaluate the operating performance of the air-conditioning system, improves the accuracy and reliability of the cooling load distribution compliance index, and thus realizes a more accurate formulation of the air-conditioning resource allocation strategy.

[0081] It should be noted that the data collection system of the energy station energy management system can collect, store, process, and analyze large amounts of energy data. The cold storage prediction algorithm can identify historical energy operation conditions. Through learning and simulation of the cold storage prediction algorithm and continuous logarithmic fitting, it can provide guidance for energy consumption prediction and allocation control, which serves as a basis for allocation. In the absence of large amounts of data, the cold storage prediction algorithm can simulate and learn the actual functional area based on the performance curve of the equipment. The performance curve of the equipment decreases with the length of use, and corrections are made simultaneously, achieving dual simulation training of historical and equipment performance curves. Equipment data is collected and stored through smart meters, and then verified using fitting functions and training algorithm data.

[0082] In addition, the integration of real-time weather data is crucial for energy management. The cold storage prediction algorithm uses real-time weather data to calibrate existing data using a fitting function. This is supported by historical data from similar conditions, improving the accuracy of the cold storage prediction algorithm in resource allocation. The cold storage prediction algorithm first collects real-time weather data, including weather conditions and other relevant meteorological information, and then uses this data to calibrate the model to ensure the accuracy of the prediction results. Next, the cold storage prediction algorithm analyzes the existing data using a fitting function to identify relationships and patterns between the data. To verify the reliability of the model, the cold storage prediction algorithm references historical data from previous years under similar weather conditions. By comparing historical data with model predictions, the algorithm further adjusts and optimizes model parameters. This process helps improve the algorithm's accuracy in resource allocation, ensuring that reasonable energy allocation decisions can be made under different weather conditions. By monitoring indoor and outdoor temperatures in real time and combining historical data with weather forecasts, the cold storage prediction algorithm can predict future temperature changes and adjust energy supply in advance to maintain optimal indoor comfort and energy efficiency.

[0083] Specifically, the historical load operation data of air conditioners is collected, the grey correlation analysis algorithm is used to analyze the historical load operation data, and the cooling load prediction model is constructed in combination with the cooling storage prediction algorithm. The following steps are included:

[0084] S11. Collect historical load operation data of the air conditioner, and perform feature selection on the historical load operation data to obtain feature data.

[0085] It should be noted that the actual situation of shopping malls in a certain area is analyzed and the key factors affecting the cooling load are selected as characteristic data.

[0086] S12. Preprocess the feature data using a grey relational analysis algorithm to obtain input features for a cooling load prediction model.

[0087] Specifically, preprocessing the feature data using the grey relational analysis algorithm to obtain input features for the cooling load prediction model includes the following steps:

[0088] S121, performing positive processing on the feature data;

[0089] S122, performing dimensionless processing on the feature data after the forward processing, and selecting a mother sequence from the feature data after the dimensionless processing;

[0090] S123. Based on the selected parent sequence, the remaining feature data is used as a subsequence, and the correlation coefficient between each subsequence and the parent sequence is calculated using a grey correlation analysis algorithm, and the grey correlation degree is calculated based on the correlation coefficient;

[0091] S124. Determine the weight of the feature data using a hierarchical analysis algorithm, and select input features for the cooling load prediction model in combination with the grey correlation degree.

[0092] It should be noted that the grey correlation analysis method is used to pre-process the model input feature data to predict the dynamic cooling load of the ice storage air conditioning system of large public buildings, including:

[0093] All indicators are converted into extremely large indices so that they change in the same direction for easy comparison. This step ensures that all indicators change in the same direction, that is, increases or decreases are positive. A reference sequence (parent sequence) is selected, usually the data sequence that best reflects the system characteristics, as a comparison benchmark. The data is dimensionless processed to eliminate the influence of dimension and order of magnitude, making the data comparable on the same scale. The correlation coefficient between each indicator and the parent sequence is calculated, which reflects the degree of similarity between the indicator and the parent sequence. The gray correlation degree, that is, the degree of correlation between the indicator and the parent sequence, is calculated based on the correlation coefficient. The weight of each indicator can be determined using methods such as expert scoring and the hierarchical analysis method. The weighted correlation degree is calculated by combining the weight and correlation degree to determine the contribution of each indicator to the cooling load prediction. Based on the gray correlation degree results, factors with low correlation with the output are eliminated, and factors with high correlation are retained as model input features. The filtered input features are used to establish a cooling load prediction model to predict the dynamic cooling load of ice storage air-conditioning systems in large public buildings. This helps improve the prediction accuracy and convergence speed of the model and meet the actual application needs of the project.

[0094] S13. Combining input characteristics with the cold storage prediction algorithm, a cooling load prediction model is constructed.

[0095] It's important to note that the configuration and deployment of the cold storage prediction algorithm in a microservices architecture involves developing applications within a distributed framework. Leveraging its fast startup and simplified configuration, the cold storage prediction algorithm is encapsulated as a standalone microservice. Through Spring Boot's automatic configuration (its built-in server and production-grade features enable developers to quickly build and deliver high-quality applications), and embedded Tomcat (a core project in the Apache Software Foundation's Jakarta project, a free, open-source, and lightweight web application server), the cold storage service can be easily deployed and run. In a microservices architecture, an API gateway such as Spring CloudGateway (a gateway framework designed to provide a simple and effective routing management solution for microservices) serves as the system's single entry point, responsible for request routing, load balancing, and security control. It integrates with Spring Boot applications, processes client requests, and routes them to the correct microservices. Real-time collection and processing of environmental data dynamically adjusts the cold storage strategy. For configuration management, Spring Cloud Config (a configuration center providing centralized external configuration support for microservice applications in distributed systems) is used to centrally manage configurations for different environments. Configuration files are stored in a Git repository or the local file system, enabling centralized management and dynamic refresh of configurations. Service discovery and registration are implemented through Eureka (a registration center primarily used for service registration and discovery in a microservices architecture, playing a crucial role in the Spring Cloud microservices system). This allows the cold storage service to register with the Eureka server upon startup and be dynamically discovered by other services. Furthermore, Spring Cloud Sleuth (a routing listener primarily used for distributed link tracing in distributed systems, helping developers better understand and monitor the interactions and call flows between microservices) and Zipkin (which helps developers collect, store, and query call chain information between services in a distributed system for better performance analysis, troubleshooting, and system monitoring) are used to implement link tracing and monitor the call chain between microservices, facilitating problem diagnosis and performance optimization. This configuration and deployment method enables the cold storage prediction algorithm service to be efficiently integrated into the microservice architecture, realizing intelligent and automated cold storage management.

[0096] Specifically, combining input features with the cold storage prediction algorithm, building a cooling load prediction model includes the following steps:

[0097] S131. According to the cold storage prediction algorithm, the number of hidden layer nodes is set, and the parameters of the hidden layer nodes are randomly initialized to construct an initial cooling load prediction model.

[0098] Specifically, the parameters of the hidden layer nodes include: input weights and biases of the hidden layer nodes.

[0099] S132: Using a preset activation function, the input features are mapped to the hidden layer of the initial cooling load prediction model, and the output value of the hidden layer is calculated.

[0100] S133. Based on the output value of the hidden layer, the least squares algorithm is used to determine the weight of the output layer, and the parameters of the initial cooling load prediction model are optimized to obtain the cooling load prediction model.

[0101] It's important to note that the cold storage prediction algorithm platform relies on multi-dimensional data integration, including historical data, unit performance and sample data support, as well as algorithm training and learning. This data integration is crucial to the algorithm's accuracy and efficiency. First, historical data provides a record of past cold storage system operations, which can be used to train machine learning models to predict future cooling load demand. Second, unit performance data, such as the impact of compressor configuration on air conditioning performance, including the evaporation pressure of the cooling system, the degree of condensation on indoor units, the cooling air volume, and the air outlet temperature, are all key parameters that the algorithm needs to consider. These performance parameters directly affect the cold storage efficiency and the overall performance of the system.

[0102] S2. Use the back propagation neural network algorithm to train the cooling load prediction model, input the real-time collected load operation data into the trained cooling load prediction model, and output the cooling load prediction result.

[0103] Specifically, the cooling load prediction model is trained using a back propagation neural network algorithm, and the real-time collected load operation data is input into the trained cooling load prediction model. Outputting the cooling load prediction result includes the following steps:

[0104] S21. Design a back propagation neural network structure based on the cooling load forecast demand, and initialize the weights and threshold parameters of the back propagation neural network.

[0105] S22. Use the particle swarm optimization algorithm to optimize the initial weights and threshold parameters of the back propagation neural network, and combine the back propagation neural network algorithm to train the cooling load prediction model.

[0106] Specifically, the particle swarm optimization algorithm is used to optimize the initial weights and threshold parameters of the back propagation neural network, and the back propagation neural network algorithm is combined to train the cooling load prediction model, which includes the following steps:

[0107] S221, initialize the particle swarm and randomly generate the initial position and velocity of each particle. The position represents the weight and threshold parameters of the back propagation neural network.

[0108] S222. Using historical load operation data, train the back propagation neural network, evaluate the fitness of the particle through the prediction error, and use the error value as the fitness value of the particle;

[0109] S223, comparing the fitness value of the particle with the corresponding historical optimal fitness value, and updating the individual optimal position of the particle based on the comparison result;

[0110] S224, traverse all particles, select the particle with the highest fitness value, and set the particle with the highest fitness value as the optimal position of the group;

[0111] S225. Adjust the speed and position of each particle according to the speed update formula of the particle swarm algorithm, and train the cooling load prediction model through the back propagation algorithm based on the adjusted speed and position;

[0112] S226: Repeat steps S222 to S225 until a preset stop condition is reached, thereby obtaining a trained cooling load prediction model.

[0113] S23. Collect load operation data in real time and pre-process the load operation data.

[0114] S24: input the pre-processed load operation data into the trained cooling load prediction model, and output the cooling load prediction result.

[0115] It should be noted that the load operation data collected in real time is input into the cooling load prediction model after training. It then includes real-time monitoring of the prediction situation of the cooling load prediction model to obtain a cooling load prediction evaluation value and, based on the obtained cooling load prediction evaluation value, determines whether to output the cooling load prediction result; the cooling load prediction evaluation value is used to evaluate the accuracy of the prediction result of the cooling load prediction model.

[0116] Specifically, the cooling load prediction evaluation value is obtained by the following method: when the obtained cooling load distribution compliance index is within the cooling load distribution compliance threshold range, the parameter optimization time is obtained and it is determined whether the parameter optimization time is less than the preset parameter optimization time (represented by the result of summing and averaging the historical parameter optimization time of the initial cooling load prediction model in the preset database). If so, the parameter optimization time score is obtained (i.e., C in the restriction expression of the cooling load prediction evaluation value), otherwise, the parameter optimization is performed again; the pre-allocated total time score (i.e., F in the restriction expression of the cooling load prediction evaluation value) is obtained and combined with the obtained parameter optimization time score, the parameter optimization time score is obtained. The cooling load prediction evaluation value is obtained by coupling the inverse proportional processing result of the optimization duration score and the cooling load distribution compliance index (i.e., FU in the restriction expression of the cooling load prediction evaluation value); the parameter optimization duration is the total duration of parameter optimization of the initial cooling load prediction model; the parameter optimization duration score is the ratio of the difference between the parameter optimization duration and the preset parameter optimization duration to the preset parameter optimization duration; the pre-allocation duration score is the ratio of the absolute value of the difference between the pre-allocation total duration and the pre-allocation target total duration to the allowable deviation of the pre-allocation target total duration (represented by the sum and average of the historical air-conditioning resource pre-allocation durations in the preset database).

[0117] Specifically, assuming that the cooling load distribution meets the threshold range between 0 and 1, and the preset parameter optimization time is 60 seconds, the specific constraint expression of the cooling load prediction evaluation value is:

[0118]

[0119] Where e is a natural constant, YU represents the cooling load prediction evaluation value of the cooling load prediction model within the preset time period, FU represents the cooling load distribution compliance index of the specified equipment in the air conditioner within the preset time period, ΔFU represents the cooling load distribution compliance threshold range, C represents the parameter optimization duration score of the cooling load prediction model within the preset time period, C1 represents the parameter optimization duration of the cooling load prediction model within the preset time period (recorded in real time by the time counter), C10 represents the preset parameter optimization duration, F represents the pre-allocation duration score of the specified equipment in the air conditioner within the preset time period, F1 represents the total pre-allocation duration of the specified equipment in the air conditioner within the preset time period (recorded in real time by the time counter), F10 represents the pre-allocation target total duration, and ΔF1 represents the allowable deviation of the pre-allocation target total duration.

[0120] It should be understood that the cooling load prediction evaluation value increases with the increase of the cooling load distribution compliance index, and decreases with the increase of the parameter optimization duration score and the pre-allocation duration score. Among them, the parameter optimization duration score decreases with the increase of the parameter optimization duration, and the pre-allocation duration score increases with the increase of the pre-allocation duration deviation (i.e., |F1-F10|).

[0121] It should be noted that (1) the pre-allocation duration score also indirectly affects the value of the cooling load distribution compliance index. An increase in the pre-allocation duration score usually means an increase in the deviation of the pre-allocation duration, which may lead to low rationality of the cooling load in the initial distribution stage. Specifically, if the pre-allocation duration is too long, the system may start too early and consume a large amount of cooling capacity, while insufficient cooling capacity may occur in subsequent periods; conversely, if the pre-allocation duration is too short, the system may start too late and fail to meet the initial cooling load demand.

[0122] (2) The cooling load distribution compliance index also indirectly affects the value of the parameter optimization duration score. Assuming that before the initial cooling load prediction model is optimized, the cooling load distribution compliance index increases, indicating that the distribution is unreasonable. At this time, through parameter optimization adjustment, the cooling load distribution of the air-conditioning system becomes more reasonable, that is, the cooling load distribution compliance index decreases. At this time, the operating time that the air-conditioning system may need will be closer to the ideal time (because the air-conditioning system can meet the cooling load demand more efficiently), which helps to reduce the parameter optimization duration score.

[0123] Through the above-mentioned indirect influence mechanism, the cooling load prediction model can further achieve a more accurate prediction of the cooling load output, because reasonable cooling load distribution and more efficient parameter optimization can reduce the system's operating deviation and energy waste, thereby improving the overall performance and prediction accuracy of the air-conditioning system.

[0124] It should be noted that the specific process of determining whether to output the cooling load prediction result based on the obtained cooling load prediction evaluation value is as follows: determining whether the obtained cooling load prediction evaluation value is within the cooling load prediction threshold range (including the maximum and minimum values ​​equal to the historical cooling load prediction evaluation values). If so, it indicates that the accuracy of the prediction result of the cooling load prediction model meets the expected demand and the cooling load prediction result is output; otherwise, it indicates that the accuracy of the prediction result of the cooling load prediction model does not meet the expected demand. In this case, the cooling load prediction result is not output and the parameters of the cooling load prediction model are adjusted.

[0125] Among them, the cooling load prediction threshold range represents the range corresponding to the maximum and minimum values ​​of the historical cooling load prediction evaluation values ​​of the cooling load prediction model in the preset database at the end of the historical time period; parameter adjustment generally includes learning rate adjustment, number of iterations adjustment and regularization coefficient adjustment. When the obtained cooling load prediction evaluation value is not within the cooling load prediction threshold range, it is necessary to increase the learning rate, number of iterations and regularization coefficient of the cooling load model. While improving the prediction accuracy of the cooling load prediction model, the stability and reliability of the cooling load prediction model are also improved, thereby improving the prediction accuracy and reliability of the cooling load prediction model.

[0126] It should be noted that the back propagation neural network algorithm is a PSO-BP (particle swarm optimization-back propagation) neural network algorithm, which introduces the particle swarm optimization algorithm into the back propagation neural network, including:

[0127] Each particle represents a set of weights and threshold parameters of the BP (back propagation) neural network, and each particle has its own position and speed; a BP neural network is constructed based on the weights and threshold parameters of each particle and trained using historical data. After training, a prediction error indicator (such as mean square error) is used to evaluate the fitness of the particle; based on the evaluated fitness value, the individual optimal position of each particle is updated, and the solution with the best fitness is retained as the individual optimal position of the particle; based on the individual optimal positions of all particles, the global optimal position, that is, the group optimal position, is determined, which is the solution with the best fitness in the entire particle swarm; according to the speed update formula of the particle swarm algorithm, the speed and position of the particle are updated, and the speed update will take into account the difference between the individual optimal position and the group optimal position; based on the updated weights and threshold parameters, the BP algorithm is used for training to further optimize the particle solution so that the BP neural network can better capture the patterns and trends of time series data; the steps of evaluating fitness, updating individual optimal positions, updating group optimal positions, and updating particle speed and position are repeated until the stopping condition is met, such as reaching the maximum number of iterations or meeting a certain prediction accuracy requirement.

[0128] By utilizing the global search characteristics of the PSO algorithm and the fast local search capabilities of the BP algorithm, good prediction results can be achieved. This algorithm, which combines global and local search capabilities, can effectively train the model and improve the accuracy of predictions.

[0129] S3. Combine cooling load forecast results with electricity price information to formulate an allocation strategy for air conditioning resources, and use the model predictive control algorithm to optimize the allocation strategy.

[0130] It should be noted that the formulation of the air conditioning resource allocation strategy also includes allocating air conditioning resources according to the formulated allocation strategy to obtain the allocation strategy compliance coefficient, and at the same time, judging whether to perform allocation strategy optimization based on the obtained allocation strategy compliance coefficient (judging whether the obtained allocation strategy compliance coefficient is greater than the preset allocation strategy compliance coefficient; if so, completing resource allocation; otherwise, performing allocation strategy optimization and re-allocating resources according to the optimized allocation strategy); the allocation strategy compliance coefficient is used to evaluate the reliability of the formulation of the air conditioning resource allocation strategy;

[0131] Specifically, the specific steps for obtaining the allocation strategy compliance coefficient are as follows: when the obtained cooling load prediction evaluation value is within the cooling load prediction threshold range, the actual cooling load output value and the actual energy consumption output value of the specified equipment in the air conditioner at the end of the preset time period are obtained, and the corresponding expected cooling load output value and expected energy consumption output value are respectively combined to obtain the cooling load output score (i.e., P in the restriction expression of the allocation strategy compliance coefficient) and the energy consumption output score (i.e., Q in the restriction expression of the allocation strategy compliance coefficient). At the same time, the results of the inverse proportional coupling processing of the cooling load output score and the energy consumption output score are coupled with the cooling load prediction evaluation value (i.e., YU in the restriction expression of the allocation strategy compliance coefficient) to obtain the allocation strategy compliance coefficient.

[0132] The cooling load output score is the ratio of the difference between the expected cooling load output value and the actual cooling load output value to the expected cooling load output value (since the expected cooling load output value is the sum and average of the historical maximum actual cooling load output values ​​of the specified air-conditioning equipment at the end of each historical time period in the preset database, the expected cooling load output value is greater than or equal to the actual cooling load output value); the energy consumption output score is the ratio of the difference between the actual energy consumption output value and the expected energy consumption output value to the expected energy consumption output value (since the expected energy consumption output value is the sum and average of the historical minimum actual energy consumption output values ​​of the specified air-conditioning equipment at the end of each historical time period in the preset database, the expected energy consumption output value is less than the actual energy consumption output value).

[0133] Specifically, the allocation strategy complies with the specific restriction expression of the coefficient:

[0134]

[0135] Where e is a natural constant, PEI represents the compliance coefficient of the allocation strategy of the specified equipment in the air conditioner at the end of the preset time period, YU represents the cooling load prediction evaluation value of the cooling load prediction model within the preset time period, ΔYU represents the cooling load prediction threshold range, P represents the cooling load output score of the specified equipment in the air conditioner at the end of the preset time period, P1 represents the actual cooling load output value of the specified equipment in the air conditioner at the end of the preset time period (recorded in real time by the flow meter), P10 represents the expected cooling load output value, Q represents the energy consumption output score of the specified equipment in the air conditioner at the end of the preset time period, Q10 represents the expected energy consumption output value, and Q1 represents the actual energy consumption output value of the specified equipment in the air conditioner at the end of the preset time period (recorded in real time by the power sensor).

[0136] It should be understood that the allocation strategy compliance coefficient increases with the increase of the cooling load forecast evaluation value, and decreases with the increase of the cooling load output score and the energy consumption output score. Among them, the cooling load output score decreases with the increase of the actual cooling load output value, and the energy consumption output score increases with the increase of the actual energy consumption output value.

[0137] It should be noted that (1) the cooling load output score also indirectly affects the value of the cooling load prediction evaluation value. When the cooling load output score increases, it indicates that the deviation between the actual cooling load output value and the expected output value increases. At this time, the prediction accuracy of the corresponding cooling load prediction model decreases, that is, the cooling load prediction evaluation value decreases.

[0138] (2) The cooling load prediction evaluation value also indirectly affects the value of the energy consumption output score. When the cooling load prediction evaluation value decreases, it indicates that the cooling load prediction model's prediction accuracy of the cooling load demand decreases, which may cause the air-conditioning system to over-supply cooling during actual operation, thereby increasing the output energy consumption (i.e., increasing the energy consumption output score).

[0139] The above-mentioned indirect impact mechanism helps operation and maintenance personnel to improve the cooling load prediction algorithm more accurately, thereby improving the prediction accuracy of the cooling load prediction model to better adapt to changes in the predicted values, thereby achieving the overall operating energy efficiency of the air-conditioning system, and thus achieving improved resource allocation management and optimization accuracy and reliability.

[0140] It should be noted that the allocation strategy includes both cold storage and cold release technologies. Cold storage technologies include water storage, ice storage, and compound storage. Cold storage and cold release technologies are key technologies for balancing cooling load supply and demand in air conditioning systems. They are linked in that they both utilize the phase change properties of materials to store and release cold energy. These technologies include:

[0141] Cold storage stage: During periods of low electricity demand, low-cost electricity is used to drive refrigeration equipment (such as chillers) to produce cold water or ice, which is then stored in cold storage equipment. This process involves the selection and application of different cold storage methods, including water storage technology, ice storage technology, and compound storage technology.

[0142] Cooling stage: During peak electricity demand periods, the stored cold energy is released to meet the cooling load requirements of the air-conditioning system. The cooling operation can be achieved by controlling valves and adjusting flow rates, including shutting down all refrigeration equipment and opening the cooling valve of the cold storage tank to release the stored cold energy.

[0143] Technical connection: The connection between cold storage and cooling technology is that they together constitute a complete energy storage and release cycle; cold storage technology is responsible for storing energy when electricity prices are low, while cooling technology releases energy when electricity prices are high. The two work together to achieve optimal energy allocation and cost savings.

[0144] Achieve allocation: Dynamically adjust the cooling storage and cooling release strategies based on real-time electricity prices and load demand. For example, when electricity prices are low, the system increases the amount of cooling stored. During peak electricity price periods, the stored cooling capacity is prioritized to meet demand, thereby achieving optimal allocation of air conditioning resources.

[0145] Optimal control: Optimal control of cold storage and cooling technology can be achieved through advanced control strategies such as model predictive control (MPC) algorithms. These strategies can be used to calculate the optimal control strategy for the cold storage and cooling of the cold storage air conditioning system on an hourly basis based on the building's hourly cooling load forecast model for the next 24 hours, thereby achieving a reasonable distribution of cooling capacity throughout the day.

[0146] Among them, water storage technology belongs to sensible heat storage technology, while ice storage technology belongs to phase change latent heat storage technology. The development and application of these technologies provide practical physical basis and technical support for the cold storage prediction algorithm.

[0147] Specifically, combining cooling load forecast results with electricity price information, formulating an air conditioning resource allocation strategy, and using the model predictive control algorithm to optimize the allocation strategy includes the following steps:

[0148] S31. Obtain electricity price information from the energy trading platform, analyze the changing trends of electricity prices in different time periods, and obtain the electricity price differences in different time periods;

[0149] S32. Based on the electricity price differences and cooling load forecast results for different time periods, a preliminary strategy for allocating air conditioning resources for each time period is formulated;

[0150] S33. Use the model predictive control algorithm to optimize the allocation strategy of air-conditioning resources in each time period.

[0151] It should be noted that, through electricity price information, the algorithm can be trained with data on price optimization and energy optimization, ultimately obtaining the optimal control solution. The optimal control solution is obtained by using the PSO-BP neural network algorithm based on electricity price information, as follows:

[0152] Collect historical electricity price data and data on related factors affecting electricity prices, perform data cleaning, standardization or normalization processing to prepare for model input; initialize the weights and bias parameters of the BP neural network, and set the network structure, including the input layer, hidden layer and output layer; apply the PSO algorithm to the parameter optimization of the BP neural network, where each particle represents a set of network parameters, and find the optimal solution by iteratively updating the position and velocity of the particles; use the training set data to train the BP neural network, and use the PSO algorithm to optimize the network parameters to minimize the prediction error; divide the data into training set, validation set and test set, and evaluate the generalization ability and prediction accuracy of the model through cross-validation and other techniques; during the model training process, consider the impact of electricity price factors on load demand, analyze the data sets with the best price and the best energy through algorithmic analysis, and adjust the model parameters to adapt to changes in electricity prices; based on the trained PSO-BP neural network model, combined with electricity price factors, perform real-time prediction and optimization, and ultimately derive the optimal control solution, and use evaluation indicators such as MAE, MSE, RMSE and MAPE to measure the accuracy and reliability of the model prediction.

[0153] Based on the results of algorithm optimization, energy distribution can be dynamically adjusted. For example, when it is predicted that energy demand will increase, the system can increase energy supply in advance to avoid energy shortages; conversely, when it is predicted that energy demand will decrease, the system can reduce supply to avoid waste.

[0154] Furthermore, data is typically divided into training, validation, and test sets to ensure the model's generalization capabilities. Model training uses the training set data; for example, in exponential smoothing, historical data is used to fit model parameters. Evaluation and tuning utilize the validation set to assess model performance. Common evaluation metrics include MAE, MSE, RMSE, and MAPE. When training a model, the training set data is first used to fit the model parameters. For example, in exponential smoothing, historical data is used to determine the model's smoothing coefficient. Model training includes data preprocessing, feature selection, model initialization, parameter optimization, and model validation. During the model evaluation phase, the validation set is used to assess model performance. Common evaluation metrics include mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), and mean absolute percentage error (MAPE).

[0155] Furthermore, model training involves steps such as data cleaning and processing, selecting relevant features, setting the initial model parameters, iteratively optimizing the model parameters using the training data, evaluating model performance on a validation set, and adjusting the model parameters to prevent overfitting or underfitting. When evaluating the model, the aforementioned metrics are calculated to measure the difference between the model's predicted values ​​and the actual values. Based on the results of these metrics, the model parameters or structure are further adjusted to improve the model's predictive accuracy and generalization ability.

[0156] After determining the model parameters, data visualization based on the energy management system allows for intuitive analysis of changes in algorithm parameters and predicted results, improving the accuracy and practicality of the predictions. Comparing the predicted results with actual data through charts helps understand the accuracy of the model's predictions.

[0157] Specifically, the optimization process of initially formulating the allocation strategy of air conditioning resources for each time period using the model predictive control algorithm includes the following steps:

[0158] S331. Select a model predictive control algorithm based on the cooling load prediction results and control requirements, and establish a model predictive control framework based on the model predictive control algorithm;

[0159] S332. Taking the initially formulated air conditioning resource allocation strategy for each time period as input, combined with the model predictive control framework, and performing hourly rolling optimization calculations, the optimal allocation strategy for air conditioning resources for each time period is obtained.

[0160] It should be noted that the model predictive control algorithm is based on machine learning algorithms, such as the improved LSTM (Long Short-Term Memory) algorithm, which can be used to predict energy demand and optimize energy distribution. The use of the improved LSTM algorithm to predict energy demand and optimize energy distribution includes:

[0161] Collect historical energy consumption data, weather data, economic indicators and other relevant data, which will serve as input for model training; clean, normalize or standardize the collected data to eliminate the influence of dimension and order of magnitude, and prepare for model training; use correlation analysis, such as the Pearson coefficient, to describe the correlation between influencing factors and load, and screen out features that have a significant impact on the prediction results; select a suitable algorithm, such as the CNN-LSTM (Convolutional Neural Network - Long Short-Term Memory Network) model, and use historical data to train the model. During the training process, the genetic algorithm particle swarm optimization (GA-PSO) can be used to improve the LSTM model. , improve the prediction performance of the model; evaluate the accuracy and reliability of the model through methods such as cross-validation and mean square error (MSE) to ensure the generalization ability of the model; apply the trained model to actual energy demand forecasting, formulate corresponding energy management strategies based on the forecast results, and optimize energy distribution; dynamically adjust energy production and consumption according to real-time energy demand and grid operation status to achieve economic, efficient and stable operation of the energy system; in the integrated energy system, consider the characteristic relationship between different energy sources, realize multi-energy complementarity and cross-scheduling, and improve energy utilization efficiency. These models can process complex data sets and learn from them to improve the accuracy of predictions.

[0162] In a specific embodiment, the cold storage prediction algorithm achieves dynamic prediction and optimal control of the cooling load of the ice storage air conditioning system through a comprehensive analysis of historical data, unit performance, price and other factors; the algorithm first uses historical load operation data, combined with the actual situation of a shopping mall in a certain area, to analyze the working principles and common control strategies of the four operating modes of the ice storage air conditioning system.

[0163] Collect historical load operation data, which will serve as input for model training; analyze the actual situation of shopping malls in a certain area, and select key factors affecting cooling load as input features of the model; the cold storage prediction algorithm uses the basic principles of extreme learning machines to randomly initialize the parameters of the hidden layer nodes, including input weights and biases, and then map them to the hidden layer through an activation function. The output layer weights are determined using the least squares method, and the model parameters are optimized to improve prediction accuracy; the trained model is applied to real-time data to perform daily hourly cooling load predictions. By comparing with actual cooling load data, the model's prediction accuracy is verified, and necessary adjustments are made.

[0164] In addition, the working principles and common control strategies of the four operating modes of ice storage air conditioning systems are as follows:

[0165] Ice storage process selection: During the operation of the ice storage air conditioning system, the refrigerator can have two operating conditions, namely ice storage condition and cooling condition. In the ice storage condition, the low-temperature ethylene glycol solution cooled by the refrigerator enters the ice storage heat exchanger of the ice storage tank, cooling the still water in the ice storage tank and freezing it into ice.

[0166] Parallel process and series process: There are two processes in the ethylene glycol solution system, parallel process and series process; in the parallel process, the refrigerator and the ice storage tank are in parallel in the system and can jointly provide cooling.

[0167] Refrigerator ice storage and ice storage tank cooling: During the period when the air conditioning system is not operating (such as at night), the refrigerator automatically switches to ice storage mode, closes the V2 and V4 valves, and opens the V1 and V3 valves, allowing the ethylene glycol solution to circulate between the refrigerator and the ice storage tank; when the ice storage tank is needed to provide cooling by melting ice, the refrigerator stops running, but still serves as a passage for the system.

[0168] Dynamic ice making and optimized operation: Based on load forecasts and daily ice melting conditions, the ice storage tank's melting rate is controlled to avoid problems with insufficient ice to melt and insufficient cooling temperatures. During peak and peak electricity price periods, more ice is melted to reduce the number of cooling units in operation. When electricity prices are approaching low and the load begins to decrease, the number of cooling units in operation is minimized or eliminated to melt all remaining ice in the ice storage tanks.

[0169] By collecting and analyzing summer shopping mall air conditioning load operating data, training and testing samples were provided for establishing a cooling load prediction model. Secondly, targeting the characteristics of ice-storage air conditioning in a specific shopping mall, the real-time steps for establishing a cooling load prediction model using an extreme learning machine (ELM) include: collecting and preprocessing historical data and related features that affect cooling load, such as indoor and outdoor temperature, humidity, and building characteristics; using this data to train the ELM model, randomly initializing input weights and hidden layer biases, mapping them to the hidden layer using an activation function, and then determining the output layer weights using the least squares method; applying the trained model to real-time data to perform daily hourly cooling load forecasts; and adjusting the cooling source equipment operating strategy based on the forecast results to achieve energy efficiency optimization. The entire process involves data collection, model training, forecast execution, and strategy adjustment.

[0170] The study also compared the prediction results of an incremental extreme learning machine (I-ELM) with those of a traditional extreme learning machine, finding that the I-ELM reduced its root mean square error by 13.6% and its relative error by 11.8%, demonstrating excellent generalization and prediction accuracy. Furthermore, the cold storage prediction algorithm considers the impact of unit performance, such as the effect of compressor configuration on air conditioning performance, including the evaporation pressure of the refrigeration system, the degree of condensation in the indoor unit, the cooling air volume, and the outlet air temperature. Energy efficiency ratios and price factors were also considered, indicating that high-efficiency units do not offer an advantage in winter heating conditions, are more expensive, and require a longer period of use to recover the investment.

[0171] In summary, the cold storage prediction algorithm provides a new approach for energy-saving optimization of ice storage air-conditioning systems by comprehensively considering factors such as historical data, unit performance, and price. It helps to achieve peak load shifting and valley filling of power load in the power grid and reduce the operating costs of users' air-conditioning systems.

[0172] like Figure 2 Figure 1 shows one implementation diagram of an air conditioning resource allocation method based on a cold storage prediction algorithm according to an embodiment of the present invention. First, equipment performance curves and external factors are obtained from historical data. After data preparation, preprocessing is performed to remove outliers using mean absolute error. The data is then characterized, with load-specific data such as temperature and power labeled. The data is then partitioned into training and test sets. The model is trained and weighted using the training set. The model is then evaluated using the test set. Model parameters are adjusted based on the error to optimize the model, achieving accurate predictions of equipment load throughout the day.

[0173] like Figure 3The figure shows the second implementation diagram of the air conditioning resource allocation method based on the cold storage prediction algorithm according to an embodiment of the present invention. First, information such as equipment parameters, all-day load, all-day temperature and humidity, all-day electricity price, and total cold storage capacity is obtained to determine the stage, status, and decision. The cold storage process is then divided into a cold storage period and a cold release period by time period division. Dynamic planning tables are constructed for the cold storage period and the cold release period, respectively, to store the cold storage capacity, cold release capacity, and costs under different conditions for each time period. Finally, with each time period as the stage, the remaining cold storage capacity as the status, and the cold release capacity for each time period as the decision output, a scientific strategy is provided for the operation of the cold storage system to achieve efficient and economical cold storage management.

[0174] like Figure 4 The figure shows the third implementation diagram of the air conditioning resource allocation method based on the cold storage prediction algorithm according to an embodiment of the present invention. First, information such as equipment parameters and all-day load is obtained. After data preprocessing and characterization, the data is divided into a training set and a test set. The model is trained and weighted using the training set, and the model is evaluated using the test set, with optimization adjustments based on the error. Simultaneously, the cold storage and cooling release periods are determined based on the time period divisions. Dynamic planning tables are constructed to store the relevant data, and the optimal strategy is ultimately output. The IoT devices execute the strategy, and the platform monitors it in real time, forming a complete cold storage strategy formulation and execution monitoring system.

[0175] It is important to understand that Figure 2 The data in the are processed and input into Figure 3 In the model training module, the trained model and its prediction results are input into Figure 4 In the strategy formulation and optimization module, the allocation strategy of air conditioning resources is formulated and optimized. Specifically, the A side represents Figure 2 The output end of data collection, preprocessing or model building is connected to the B end; the B end represents the input end of the model training module, which is used to receive the preprocessed historical load operation data or the preliminary constructed cooling load prediction model and is connected to the C end; the C end represents Figure 3 The output end of the model training and prediction process, that is, the cooling load prediction model after training and its output prediction results, is connected to the D end; the D end represents the input end of the strategy formulation and optimization module, which is used to receive the cooling load prediction results and electricity price information in order to formulate and optimize the allocation strategy.

[0176] like Figure 5 FIG. 1 is a block diagram of a system for allocating air-conditioning resources based on a cold storage prediction algorithm according to an embodiment of the present invention. The system includes a model building module 1, a model training and prediction module 2, and a strategy formulation and optimization module 3.

[0177] Model building module 1 is used to collect historical load operation data of air conditioners, analyze the historical load operation data using the grey correlation analysis algorithm, and build a cooling load prediction model in combination with the cold storage prediction algorithm;

[0178] Model training and prediction module 2 is used to train the cooling load prediction model using the back propagation neural network algorithm, input the real-time collected load operation data into the trained cooling load prediction model, and output the cooling load prediction results;

[0179] Strategy formulation and optimization module 3 is used to combine the cooling load forecast results with electricity price information to formulate the allocation strategy of air conditioning resources and optimize the allocation strategy using the model predictive control algorithm.

[0180] In summary, the technical solutions of the present invention, through the integration of load forecasting technology and real-time electricity price information, can accurately predict cooling load demand throughout the day. This allows for optimized timing of cold storage and cooling release, enabling cold storage during periods of lower electricity prices and utilization of stored cold energy during periods of higher prices. This effectively reduces the operating costs of the air conditioning system, reducing energy waste and improving energy efficiency while also responsive to dynamic electricity pricing markets, achieving economic maximization. By leveraging historical data analysis and machine learning techniques, the cooling load forecasting model can be continuously optimized, enhancing the accuracy and efficiency of the cold storage prediction algorithm. Furthermore, the cooling release and storage control strategies can be dynamically adjusted based on actual demand and electricity price fluctuations, increasing system flexibility and adaptability. This allows the air conditioning system to automatically make the optimal choice under varying conditions, further improving system efficiency and service quality. By comprehensively monitoring the operating status of the cold storage system and providing timely feedback, system anomalies can be quickly detected and responded to, avoiding potential failure risks and ensuring stable system operation. Furthermore, continuous data collection and analysis support long-term performance optimization and technical improvements, providing users with more reliable service assurance.

[0181] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. The air conditioning resource allocation method based on the cold storage prediction algorithm is characterized by: The method comprises the following steps: S1. Collect historical load operation data of air conditioners, analyze the historical load operation data using the grey correlation analysis algorithm, and build a cooling load prediction model by combining it with the cold storage prediction algorithm; S2. Use the back propagation neural network algorithm to train the cooling load prediction model, input the real-time collected load operation data into the trained cooling load prediction model, and output the cooling load prediction result; S3. Combine cooling load forecast results with electricity price information to formulate an allocation strategy for air conditioning resources, and use the model predictive control algorithm to optimize the allocation strategy.

2. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 1 is characterized in that: The process of collecting historical load operation data of the air conditioner, analyzing the historical load operation data using a grey correlation analysis algorithm, and combining the cold storage prediction algorithm to construct a cold load prediction model includes the following steps: S11. Collect historical load operation data of the air conditioner, and perform feature selection on the historical load operation data to obtain feature data; S12. Preprocessing the characteristic data using a grey relational analysis algorithm to obtain input features for a cooling load prediction model; S13. Combining input characteristics with the cold storage prediction algorithm, a cooling load prediction model is constructed.

3. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 2 is characterized in that: The method of preprocessing the characteristic data using the grey relational analysis algorithm to obtain the input characteristics for the cooling load prediction model includes the following steps: S121, performing positive processing on the feature data; S122, performing dimensionless processing on the feature data after the forward processing, and selecting a mother sequence from the feature data after the dimensionless processing; S123. Based on the selected parent sequence, the remaining feature data is used as a subsequence, and the correlation coefficient between each subsequence and the parent sequence is calculated using a grey correlation analysis algorithm, and the grey correlation degree is calculated based on the correlation coefficient; S124. Determine the weight of the feature data using a hierarchical analysis algorithm, and select input features for the cooling load prediction model in combination with the grey correlation degree.

4. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 3 is characterized in that: The construction of a cooling load prediction model by combining input features with a cold storage prediction algorithm includes the following steps: S131. According to the cold storage prediction algorithm, the number of hidden layer nodes is set, and the parameters of the hidden layer nodes are randomly initialized to construct an initial cooling load prediction model; S132, using a preset activation function, mapping the input features to the hidden layer of the initial cooling load prediction model, and calculating the output value of the hidden layer; S133. Based on the output value of the hidden layer, the least squares algorithm is used to determine the weight of the output layer, and the parameters of the initial cooling load prediction model are optimized to obtain the cooling load prediction model.

5. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 4 is characterized in that: The parameters of the hidden layer nodes include: input weights and biases of the hidden layer nodes.

6. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 1 is characterized in that: The method of training the cooling load prediction model using the back propagation neural network algorithm, inputting the real-time collected load operation data into the trained cooling load prediction model, and outputting the cooling load prediction result includes the following steps: S21. Design a back propagation neural network structure based on the cooling load forecast demand, and initialize the weights and threshold parameters of the back propagation neural network; S22. Optimizing the initial weights and threshold parameters of the back propagation neural network using a particle swarm optimization algorithm, and training the cooling load prediction model in combination with the back propagation neural network algorithm; S23. Collect load operation data in real time and pre-process the load operation data; S24: input the pre-processed load operation data into the trained cooling load prediction model, and output the cooling load prediction result.

7. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 6 is characterized in that: The method of optimizing the initial weights and threshold parameters of the back propagation neural network using the particle swarm optimization algorithm and training the cooling load prediction model in combination with the back propagation neural network algorithm includes the following steps: S221, initializing the particle swarm, randomly generating the initial position and velocity of each particle, wherein the position represents the weight and threshold parameters of the back propagation neural network; S222. Using historical load operation data, train the back propagation neural network, evaluate the fitness of the particle through the prediction error, and use the error value as the fitness value of the particle; S223, comparing the fitness value of the particle with the corresponding historical optimal fitness value, and updating the individual optimal position of the particle based on the comparison result; S224, traverse all particles, select the particle with the highest fitness value, and set the particle with the highest fitness value as the optimal position of the group; S225. Adjust the speed and position of each particle according to the speed update formula of the particle swarm algorithm, and train the cooling load prediction model through the back propagation algorithm based on the adjusted speed and position; S226: Repeat steps S222 to S225 until a preset stop condition is reached, thereby obtaining a trained cooling load prediction model.

8. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 1 is characterized in that: The method of combining cooling load prediction results with electricity price information to formulate an air conditioning resource allocation strategy and optimizing the allocation strategy using a model predictive control algorithm includes the following steps: S31. Obtain electricity price information from the energy trading platform, analyze the changing trends of electricity prices in different time periods, and obtain the electricity price differences in different time periods; S32. Based on the electricity price differences and cooling load forecast results for different time periods, a preliminary strategy for allocating air conditioning resources for each time period is formulated; S33. Use the model predictive control algorithm to optimize the allocation strategy of air-conditioning resources in each time period.

9. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 8 is characterized in that: The method of optimizing the allocation strategy of air conditioning resources for each time period by using the model predictive control algorithm includes the following steps: S331. Select a model predictive control algorithm based on the cooling load prediction results and control requirements, and establish a model predictive control framework based on the model predictive control algorithm; S332. Taking the initially formulated air conditioning resource allocation strategy for each time period as input, combined with the model predictive control framework, and performing hourly rolling optimization calculations, the optimal allocation strategy for air conditioning resources for each time period is obtained.

10. An air conditioning resource allocation system based on a cold storage prediction algorithm, used to implement the air conditioning resource allocation method based on a cold storage prediction algorithm according to any one of claims 1 to 9, characterized in that: The system includes a model building module, a model training and prediction module, and a strategy formulation and optimization module; The model building module is used to collect historical load operation data of air conditioners, analyze the historical load operation data using a grey correlation analysis algorithm, and build a cooling load prediction model in combination with a cold storage prediction algorithm; The model training and prediction module is used to train the cooling load prediction model using the back propagation neural network algorithm, input the real-time collected load operation data into the trained cooling load prediction model, and output the cooling load prediction result; The strategy formulation and optimization module is used to combine the cooling load prediction results and electricity price information to formulate an allocation strategy for air conditioning resources and optimize the allocation strategy using a model predictive control algorithm.

11. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 1, characterized in that: The grey correlation analysis algorithm is used to analyze the historical load operation data, and then it is also included to determine whether to build a cooling load prediction model. The specific process is as follows: Obtain actual operating data of a specified device in the air conditioner within a preset time period; Pre-allocate designated equipment in the air conditioner based on the actual operating data obtained to obtain a cooling load distribution compliance index; Determine whether the obtained cooling load distribution compliance index is within the cooling load distribution compliance threshold range. If so, complete pre-distribution and build a cooling load prediction model in combination with the cold storage prediction algorithm. Otherwise, adjust the pre-distribution parameters and re-perform pre-distribution. The actual operation data includes cooling load, operation time and operation energy consumption; The cooling load distribution compliance index is used to evaluate the matching degree between the cooling load pre-distribution plan of a designated device in the air conditioner and the actual operating conditions; The pre-allocation parameters include a cooling load distribution ratio and a peak energy consumption of a designated device.

12. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 11, characterized in that: The specific steps for obtaining the cooling load distribution compliance index are as follows: Obtain the actual cooling load, actual operating hours, and actual operating energy consumption of a specified device in the air conditioner within a preset time period, and combine the corresponding target values ​​and maximum allowable deviation values ​​to obtain the cooling load compliance score, operating hours compliance score, and operating energy consumption compliance score; The preset ambient temperature impact factor is obtained from the preset database, and the cooling load distribution compliance index is obtained by combining the obtained cooling load compliance score, operating time compliance score and operating energy consumption compliance score with the coupled inverse proportional processing results; The cooling load compliance score represents the ratio of the absolute value of the difference between the actual cooling load and the target cooling load to the maximum allowable cooling load deviation; The runtime compliance score represents the ratio of the absolute value of the difference between the actual runtime and the target runtime to the maximum allowable runtime deviation; The operating energy consumption compliance score represents the ratio of the absolute value of the difference between the actual operating energy consumption and the target operating energy consumption to the maximum allowable operating energy consumption deviation.

13. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 1, characterized in that: The method includes inputting the real-time collected load operation data into the trained cooling load prediction model, and then monitoring the prediction of the cooling load prediction model in real time to obtain a cooling load prediction evaluation value and determining whether to output a cooling load prediction result based on the obtained cooling load prediction evaluation value; The cooling load prediction evaluation value is used to evaluate the accuracy of the prediction results of the cooling load prediction model; The cooling load prediction evaluation value is obtained by the following method: When the obtained cooling load distribution compliance index is within the cooling load distribution compliance threshold range, the parameter optimization time is obtained and it is determined whether the parameter optimization time is less than the preset parameter optimization time. If so, the parameter optimization time score is obtained, otherwise the parameter optimization is performed again. Obtain the total pre-allocation duration score and combine it with the obtained parameter optimization duration score and the cooling load distribution compliance index to obtain the cooling load forecast evaluation value; The parameter optimization duration score is the ratio of the difference between the parameter optimization duration and the preset parameter optimization duration to the preset parameter optimization duration; The pre-allocated duration score is the ratio of the absolute value of the difference between the pre-allocated total duration and the pre-allocated target total duration to the allowable deviation of the pre-allocated target total duration.

14. The air conditioning resource allocation method based on the cold storage prediction algorithm according to claim 1, characterized in that: The method further comprises formulating an allocation strategy for air conditioning resources, and then allocating resources to the air conditioners according to the formulated allocation strategy to obtain an allocation strategy compliance coefficient, and determining whether to perform allocation strategy optimization based on the obtained allocation strategy compliance coefficient; The allocation strategy compliance coefficient is used to evaluate the reliability of the formulation of the air conditioning resource allocation strategy; The specific steps for obtaining the allocation strategy compliance coefficient are as follows: When the obtained cooling load forecast evaluation value is within the cooling load forecast threshold range, the actual cooling load output value and actual energy consumption output value of the specified air-conditioning equipment at the end of the preset time period are obtained, and the cooling load output score and energy consumption output score are obtained by combining the corresponding expected cooling load output value and expected energy consumption output value, respectively. At the same time, the cooling load output score and energy consumption output score are coupled with the cooling load forecast evaluation value and the result of the inverse proportional processing, and then coupled to obtain the allocation strategy compliance coefficient; The cooling load output fraction is the ratio of the difference between the expected cooling load output value and the actual cooling load output value to the expected cooling load output value; The energy consumption output score is the ratio of the difference between the actual energy consumption output value and the expected energy consumption output value to the expected energy consumption output value.

Images