Intelligent shelter hospital power load prediction and scheduling method and system

Abstract

The invention provides an intelligent shelter hospital power load prediction and scheduling method and system. The method comprises the following steps: collecting and integrating internal and external environment parameters of a shelter hospital and medical equipment operation state information in real time, collecting historical power consumption data and special power demand records, and generating a comprehensive data set containing rich background information; the power load demand trend of the shelter hospital under different environment conditions is predicted, and an accurate power load prediction report is generated; according to the power load prediction report, introducing a fuzzy logic control technology to evaluate the flexibility of a power system, and generating an optimal power distribution and scheduling scheme; and on the basis of the optimal power distribution and scheduling scheme, carrying out system performance evaluation regularly, adjusting algorithm parameters according to an evaluation result, and generating an intelligent management scheme of the shelter hospital power system. According to the technical scheme provided by the invention, the flexibility of the power distribution and scheduling method can be improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of power load forecasting and scheduling, and in particular, to a method and system for forecasting and scheduling power load in an intelligent modular hospital. Background Art

[0002] As a temporary medical facility, Fangcang shelter hospitals need to provide medical services efficiently and reliably in a short period of time. In order to ensure the normal operation of medical equipment and key systems, the stability and reliability of power supply are crucial. Fangcang shelter hospitals need to collect and integrate internal and external environmental parameters and medical equipment operating status information in real time, collect historical power consumption data and special power demand records, and generate comprehensive data sets containing rich background information. These data sets can not only help managers fully understand the changing trends of power demand, but also provide a scientific basis for the optimal management of the power system.

[0003] At present, the power management of Fangcang Hospital mainly relies on traditional power distribution and dispatching methods. These methods are usually based on fixed schedules and empirical rules, and the power distribution plan is formulated manually. In addition, some advanced methods have begun to apply machine learning and optimization algorithms, such as long short-term memory networks and mixed integer linear programming techniques, to predict power load demand and optimize power distribution. However, these methods still have certain limitations.

[0004] Traditional methods are difficult to accurately predict power load demand, especially when faced with complex and changeable environmental conditions and the operating status of medical equipment. The prediction error is large, resulting in waste or shortage of power resources. The existing power distribution and scheduling methods lack flexibility and cannot respond to sudden changes in power demand and uncertainty in the external environment in a timely manner, which easily leads to instability in the power system. Although some advanced methods have applied machine learning and optimization algorithms, they are still insufficient in multi-objective optimization and comprehensive performance evaluation. For example, the economic efficiency of power distribution, the flexibility of the power system, and resource utilization have not been fully considered, resulting in unsatisfactory optimization results. Summary of the invention

[0005] The embodiments of the present application provide a method and system for predicting and dispatching power load in an intelligent modular hospital, so as to solve the problem of lack of flexibility in power distribution and dispatching methods in the prior art.

[0006] In a first aspect, an embodiment of the present application provides a method for predicting and dispatching power load of an intelligent shelter hospital, including:

[0007] Collect and integrate the indoor and outdoor environmental parameters of Fangcang Cabin Hospital and the operating status of medical equipment in real time, collect historical power consumption data and special power demand records, and generate a comprehensive data set containing rich background information;

[0008] The long short-term memory network is combined with the seasonal autoregressive integrated moving average model to conduct an in-depth analysis of the comprehensive data set, predict the power load demand trend of the shelter hospital under different environmental conditions, and apply the multi-objective genetic algorithm to optimize the power load forecasting model to generate an accurate power load forecasting report;

[0009] According to the power load forecast report, a mixed integer linear programming technique is used to formulate a power distribution plan, fuzzy logic control technology is introduced to evaluate the flexibility of the power system, and the start and stop timing of non-critical loads is dynamically adjusted to cope with uncertain power supply fluctuations, thereby generating an optimal power distribution and scheduling plan;

[0010] Based on the optimal power distribution and scheduling scheme, system performance evaluation is carried out regularly, and algorithm parameters are adjusted according to the evaluation results to generate an intelligent management plan for the square cabin hospital power system.

[0011] Optionally, the long short-term memory network is combined with a seasonal autoregressive integrated moving average model to perform an in-depth analysis on the comprehensive data set, predict the power load demand trend of the square cabin hospital under different environmental conditions, and apply a multi-objective genetic algorithm to optimize the power load forecasting model to generate an accurate power load forecasting report, including:

[0012] Cleaning and normalizing the comprehensive data set, removing outliers and missing values, ensuring data quality, and obtaining a preprocessed data set;

[0013] Performing feature extraction and selection on the preprocessed data set to construct an input vector containing valuable features of environmental parameters and medical device working modes;

[0014] Based on the input vector, a hybrid prediction model of a long short-term memory network combined with a seasonal autoregressive integrated moving average model is constructed, and the hybrid prediction model is trained to obtain a preliminary power load prediction model;

[0015] Utilizing a multi-objective genetic algorithm, the preliminary power load forecasting model is optimized to find the best combination of model parameters and generate an optimized power load forecasting model;

[0016] Based on the optimized power load forecasting model, the power load demand of the square cabin hospital under different environmental conditions is predicted and processed to obtain an accurate power load forecasting report.

[0017] Optionally, the use of a multi-objective genetic algorithm to optimize the preliminary power load forecasting model, find the best combination of model parameters, and generate an optimized power load forecasting model includes:

[0018] In order to calculate the objective function F1(w), the generalization ability of the model is first evaluated through cross-validation, and then the key parameters affecting the model performance are determined using sensitivity analysis, and finally the initial parameters of the multi-objective genetic algorithm are set;

[0019] Objective function:

[0020]

[0021] Where w represents the model parameter vector; y i is the actual power load value of the i-th sample; is the predicted power load value of the i-th sample; N is the total number of samples; α is a weight coefficient used to balance the contribution of the root mean square error and the mean absolute error in the objective function, and its value range is between 0 and 1; γ is a weight coefficient used to balance the contribution of the mean absolute percentage error in the objective function, and its value range is between 0 and 1;

[0022] After completing the calculation of F1(w), the constraint condition F2(w) is constructed by calculating the coefficient of determination, F score and variance of the predicted value of the model to ensure that the optimization process considers both prediction accuracy and model stability;

[0023] Constraints:

[0024]

[0025] Among them, R 2 (w) is the coefficient of determination, which measures the proportion of the variability of the model explaining the variables; max(R 2 ) is R 2 The maximum possible value, usually 1; F score (w) is the F score, which measures the overall performance of the model and takes into account the precision and recall of the model; max(F score ) is the maximum possible value of the F score, usually 1; Variance(w) is the variance of the model prediction value, which measures the stability of the model prediction; max(Variance) is the maximum possible value of the variance; β is the weight coefficient used to balance R 2 The contribution of the F score to the constraints ranges from 0 to 1. δ is the weight coefficient used to balance the contribution of the variance to the constraints, and ranges from 0 to 1.

[0026] These two formulas work together in the multi-objective genetic algorithm to iteratively optimize the model parameter w to minimize the objective function F1 and the constraint condition F2, thereby generating an optimized power load forecasting model.

[0027] Optionally, the extracting and selecting features of the preprocessed data set to construct an input vector containing valuable features of environmental parameters and medical device working modes includes:

[0028] Using statistical analysis methods, performing correlation analysis on various indicators in the preprocessed data set, identifying features closely related to power load, and obtaining a list of features with high correlation;

[0029] According to the list of features with high correlation, features with high correlation are selected as components of the input vector to generate a selected feature set;

[0030] Encoding the selected feature set, converting categorical features into numerical data by using one-hot encoding or label encoding, and standardizing or normalizing continuous features according to actual needs to obtain a standardized feature set;

[0031] Based on the standardized feature set, a final input vector is constructed to obtain an accurate power load forecast report.

[0032] Optionally, based on the input vector, a hybrid prediction model of a long short-term memory network combined with a seasonal autoregressive integrated moving average model is constructed, and the hybrid prediction model is trained to obtain a preliminary power load prediction model, including:

[0033] Using the input vector, a long short-term memory network structure is designed to construct an LSTM model part capable of capturing long-term dependencies in time series data, thereby obtaining an LSTM model structure;

[0034] According to the time characteristics of the input vector, a suitable seasonal period is selected, and a seasonal autoregressive integrated moving average model part is constructed to obtain a SARIMA model structure;

[0035] Based on the LSTM model structure and the SARIMA model structure, the architecture of the hybrid prediction model is designed to ensure that the two model parts can be effectively combined to generate the hybrid prediction model architecture;

[0036] The preprocessed data set and the standardized feature set are used to train the hybrid prediction model architecture, and a preliminary power load prediction model is generated by iteratively optimizing model parameters.

[0037] Optionally, the power distribution plan is formulated based on the power load forecast report using mixed integer linear programming technology, fuzzy logic control technology is introduced to evaluate the flexibility of the power system, and the start and stop timing of non-critical loads is dynamically adjusted to cope with uncertain power supply fluctuations, thereby generating an optimal power distribution and scheduling plan, including:

[0038] According to the power load forecast report, analyze the future power demand trend, determine the basic framework of power distribution, and obtain the power demand analysis results;

[0039] Using the power demand analysis results, adopting mixed integer linear programming technology, combined with the existing power supply capacity of the Fangcang Hospital and the priorities of various loads, the power distribution is modeled and processed to generate a preliminary power distribution plan;

[0040] Introducing fuzzy logic control technology, based on the preliminary power distribution plan, evaluating the flexibility of the power system, and generating a power system flexibility evaluation report;

[0041] According to the power system flexibility assessment report, dynamically adjust the start and stop timing of non-critical loads, determine the best start and stop strategy through an optimization algorithm, and generate an optimized start and stop strategy;

[0042] Based on the preliminary power distribution plan and the optimized start-stop strategy, the accuracy of power load forecasting, the economy of power distribution and the flexibility of the power system are comprehensively considered to generate the optimal power distribution and scheduling plan.

[0043] Optionally, dynamically adjusting the start and stop timing of non-critical loads according to the power system flexibility assessment report, determining the best start and stop strategy through an optimization algorithm, and generating an optimized start and stop strategy includes:

[0044] When calculating the startup time T of non-critical loads on (i) Previously, it was necessary to analyze the data from the power system flexibility assessment report, identify the characteristics of non-critical loads, and determine the benchmark start time and allocation weights;

[0045]

[0046] Among them, T on (i) represents the startup time of the i-th non-critical load; t0 is the benchmark startup time, which is usually the start time of the low power demand period; N is the number of non-critical loads; w j is the weight of the jth non-critical load, reflecting its importance in power distribution; P j is the power demand of the jth non-critical load; P min It is the smallest power requirement among all non-critical loads; F j is the flexibility score of the jth non-critical load; F max The highest flexibility score among all non-critical loads; D j is the startup delay time of the jth non-critical load; λ is the influence coefficient of the delay time;

[0047] After calculating Ton (i), the impact of different start and stop schedules is evaluated, and the optimal start and stop strategy S(I) is found using an optimization algorithm;

[0048] Formula 2: Optimization algorithm determines the best start-stop strategy

[0049]

[0050] Where S(i) represents the optimal start-stop strategy for the i-th non-critical load; s i is the start and stop schedule of the i-th non-critical load; C(s i ) is the start and stop schedule i The total cost under the condition includes the startup cost and the running cost; V(s i ) is the start and stop schedule i Power system volatility under i ) is the start and stop schedule i α is the cost weight coefficient, ranging from 0 to 1; β is the load balancing weight coefficient, ranging from 0 to 1; γ is the balance coefficient between volatility and load balancing; R(s i ) is the start and stop schedule i Resource utilization under

[0051] Based on S(i), a detailed start-stop schedule is generated, its effect is verified through simulation or testing, and the optimization is adjusted according to the results to form the final start-stop strategy.

[0052] Optionally, the power demand analysis result is used to adopt a mixed integer linear programming technique, combined with the existing power supply capacity of the square cabin hospital and the priority of various loads, to model the power distribution and generate a preliminary power distribution plan, including:

[0053] Using the power demand analysis results, determine the total power demand and its distribution characteristics in a specific future time period, and generate a power demand distribution map;

[0054] According to the power demand distribution map and in combination with the existing power supply capacity of the Fangcang Cabin Hospital, a detailed assessment of the power supply capacity is conducted to generate a power supply capacity assessment report;

[0055] Based on the power supply capacity assessment report, determining the priority of each type of load and generating a load priority list;

[0056] Using mixed integer linear programming technology, according to the load priority list and the power demand distribution map, a mathematical model of power distribution is established to generate a power distribution model;

[0057] The power distribution model is solved, and the optimal power distribution scheme is found by setting the objective function and constraint conditions, so as to generate a preliminary power distribution plan.

[0058] Optionally, based on the optimal power distribution and dispatching scheme, system performance evaluation is performed regularly, algorithm parameters are adjusted according to the evaluation results, and an intelligent management scheme for the square cabin hospital power system is generated, including:

[0059] Using a predetermined time period, data is collected from the power system of the shelter hospital, including power consumption, actual power consumption of each load, and the working status of power equipment, to obtain actual operating data;

[0060] Based on the actual operation data, compare the difference between the actual power distribution and dispatching situation and the optimal power distribution and dispatching plan, evaluate the performance indicators of the power system, and generate a system performance evaluation report;

[0061] Based on the system performance evaluation report, in-depth analysis of existing problems and potential areas for improvement is performed to generate a problem analysis report;

[0062] Using the problem analysis report, appropriately adjusting algorithm parameters for generating an optimal power distribution and dispatching solution to generate adjusted algorithm parameters;

[0063] Based on the adjusted algorithm parameters, the new power distribution and dispatching scheme is tested in a simulation environment to generate a test verification report;

[0064] Using the test verification report, apply the tested and verified new solution to the actual power system of the Fangcang Hospital, continuously monitor the operating status of the system, collect data on the implementation effect, and generate implementation monitoring data;

[0065] Based on the implementation monitoring data, a complete intelligent management plan for the square cabin hospital power system is formed and an intelligent management plan is generated.

[0066] In a second aspect, the embodiment of the present application provides an intelligent square cabin hospital power load forecasting and dispatching system, including:

[0067] The collection module is used to collect and integrate the environmental parameters inside and outside the Fangcang Hospital and the operating status information of medical equipment in real time, collect historical power consumption data and special power demand records, and generate a comprehensive data set containing rich background information;

[0068] An analysis module, for performing in-depth analysis on the comprehensive data set using a long short-term memory network combined with a seasonal autoregressive integrated moving average model, predicting the power load demand trend of the shelter hospital under different environmental conditions, optimizing the power load forecasting model using a multi-objective genetic algorithm, and generating an accurate power load forecasting report;

[0069] An adjustment module is used to formulate a power distribution plan based on the power load forecast report using mixed integer linear programming technology, introduce fuzzy logic control technology to evaluate the flexibility of the power system, dynamically adjust the start and stop timing of non-critical loads to cope with uncertain power supply fluctuations, and generate an optimal power distribution and scheduling plan;

[0070] The evaluation module is used to regularly evaluate system performance based on the optimal power distribution and scheduling plan, adjust algorithm parameters according to the evaluation results, and generate an intelligent management plan for the square cabin hospital power system.

[0071] In an embodiment of the present application, the environmental parameters inside and outside the Fangcang Hospital and the operating status information of the medical equipment are collected and integrated in real time, the historical power consumption data and the special power demand records are collected, and a comprehensive data set containing rich background information is generated; the long short-term memory network is combined with the seasonal autoregressive integral sliding average model to conduct an in-depth analysis of the comprehensive data set, predict the power load demand trend of the Fangcang Hospital under different environmental conditions, and apply the multi-objective genetic algorithm to optimize the power load forecasting model to generate an accurate power load forecasting report; according to the power load forecasting report, the mixed integer linear programming technology is used to formulate the power distribution plan, the fuzzy logic control technology is introduced to evaluate the flexibility of the power system, and the start and stop timing of non-critical loads is dynamically adjusted to cope with the uncertain power supply fluctuations, and the optimal power distribution and scheduling plan is generated; based on the optimal power distribution and scheduling plan, the system performance evaluation is performed regularly, the algorithm parameters are adjusted according to the evaluation results, and the intelligent management plan of the Fangcang Hospital power system is generated.

[0072] The technical solution of this application has the following beneficial effects:

[0073] By collecting and integrating the environmental parameters inside and outside the Fangcang Hospital and the operating status information of medical equipment in real time, combined with historical power consumption data and special power demand records, a comprehensive data set containing rich background information is generated. The long short-term memory network (LSTM) combined with the seasonal autoregressive integrated moving average model (SARIMA) is used for in-depth analysis, which can more accurately predict the power load demand trend of Fangcang Hospital under different environmental conditions. The multi-objective genetic algorithm is used to optimize the power load forecasting model, which further improves the prediction accuracy and generates an accurate power load forecast report; based on the power load forecast report, the mixed integer linear programming technology is used to formulate the power distribution plan to ensure the reasonable allocation of power resources. The fuzzy logic control technology is introduced to evaluate the flexibility of the power system, dynamically adjust the start and stop timing of non-critical loads, effectively respond to uncertain power supply fluctuations, generate the optimal power distribution and scheduling plan, and improve the stability and reliability of the power system; based on the optimal power distribution and scheduling plan, the system performance is evaluated regularly, and the algorithm parameters are adjusted according to the evaluation results to generate an intelligent management plan for the Fangcang Hospital power system. This process ensures the efficient operation of the power system, reduces power waste, improves energy efficiency, and also extends the service life of power equipment; by dynamically adjusting the start and stop timing of non-critical loads, the system can quickly respond to changes in power demand and the uncertainty of the external environment, enhancing the adaptability and flexibility of the power system. This not only improves the overall performance of the system, but also provides a strong guarantee for the normal operation of the Fangcang Hospital; the optimized power system can better meet the power needs of key medical equipment and life support systems, ensure the stable operation of medical equipment, and improve the satisfaction and sense of security of patients and medical staff; through intelligent management solutions, power resources are reasonably allocated, unnecessary power consumption is reduced, the operating costs of Fangcang Hospital are reduced, and economic benefits are improved.

[0074] Furthermore, the data quality was ensured by cleaning and normalizing the comprehensive data set, removing outliers and missing values. This step laid a solid foundation for subsequent feature extraction and model training, and improved the accuracy and reliability of model input; feature extraction and selection were performed on the preprocessed data set to construct an input vector containing valuable features of environmental parameters and medical equipment working modes. These features can fully reflect the key factors affecting power load, improve the explanatory power and prediction accuracy of the model; based on the input vector, a hybrid prediction model of long short-term memory network (LSTM) combined with seasonal autoregressive integrated moving average model (SARIMA) was constructed. LSTM can capture long-term dependencies in time series data, while SARIMA can handle seasonal and cyclical changes in data. The combination of the two significantly improves the prediction ability of the model; a multi-objective genetic algorithm was used to optimize the preliminary power load forecasting model to find the best combination of model parameters. Through the optimization algorithm, the model can find a balance point between multiple objectives and generate an optimized power load forecasting model, further improving the accuracy and stability of the forecast; according to the optimized power load forecasting model, the power load demand of Fangcang Hospital under different environmental conditions is forecasted and processed to generate an accurate power load forecasting report. These reports provide a scientific basis for power distribution and dispatching, help to reasonably arrange power resources, and ensure the efficient operation of the power system; accurate power load forecasting reports can help Fangcang Hospital better cope with changes in power demand and uncertainties in the external environment, dynamically adjust power distribution and dispatching strategies, and improve the flexibility and adaptability of the power system; through accurate power load forecasting and optimized power distribution and dispatching solutions, Fangcang Hospital can reasonably use power resources, reduce unnecessary power consumption, reduce operating costs, and improve economic benefits.

[0075] Furthermore, according to the power load forecast report, the future power demand trend is analyzed, the basic framework of power distribution is determined, and the power demand analysis results are generated. This step ensures the scientificity and rationality of the power distribution plan and improves the accuracy of power demand forecasting; using the power demand analysis results, the mixed integer linear programming technology is used, combined with the existing power supply capacity of the square cabin hospital and the priority of various loads, the power distribution is modeled and processed to generate a preliminary power distribution plan. Mixed integer linear programming technology can effectively solve complex optimization problems, ensure the rational allocation of power resources, and improve the efficiency and economy of power distribution; fuzzy logic control technology is introduced, based on the preliminary power distribution plan, the flexibility of the power system is evaluated, and a power system flexibility evaluation report is generated. Fuzzy logic control technology can handle uncertainty and complexity, evaluate the flexibility of the power system under different operating conditions, and provide a scientific basis for dynamic adjustment; according to the power system flexibility evaluation report, the start and stop timing of non-critical loads is dynamically adjusted, the optimal start and stop strategy is determined by the optimization algorithm, and the optimized start and stop strategy is generated. This step can effectively cope with power supply fluctuations and improve the stability and reliability of the power system. Based on the preliminary power distribution plan and the optimized start-stop strategy, the accuracy of power load forecasting, the economy of power distribution and the flexibility of the power system are comprehensively considered to generate the optimal power distribution and dispatching plan. The optimal plan can ensure that the operating efficiency and economy of the power system are maximized while meeting power demand. By dynamically adjusting the start-stop timing of non-critical loads, the system can quickly respond to changes in power demand and the uncertainty of the external environment, enhance the adaptability and responsiveness of the power system, and ensure the normal operation of the Fangcang Hospital. The optimized power distribution and dispatching plan can make rational use of power resources, reduce unnecessary power consumption, reduce operating costs, improve economic benefits, and extend the service life of power equipment.

[0076] These and other aspects of the present application will become more apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0078] Figure 1 A flowchart of a method for predicting and dispatching power load for an intelligent shelter hospital provided in an embodiment of the present application;

[0079] Figure 2A schematic diagram of the structure of an intelligent modular hospital power load forecasting and dispatching system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0080] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0081] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0082] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0083] Figure 1 A flowchart of a method for predicting and dispatching power load in an intelligent shelter hospital is provided for an embodiment of the present application, such as Figure 1 As shown, the method includes:

[0084] 101. Collect and integrate the environmental parameters inside and outside the Fangcang Hospital and the operating status information of medical equipment in real time, collect historical power consumption data and special power demand records, and generate a comprehensive data set containing rich background information;

[0085] Internal and external environmental parameters include indoor and outdoor temperature, humidity, light intensity, wind speed, etc., which affect the demand for power load. Medical equipment operating status information includes the power-on status, working mode, power consumption, etc. of medical equipment, which directly affects the change of power load. Historical power consumption data refers to the power consumption record of Fangcang Hospital over a period of time in the past, which is used to analyze the trend of power demand. Special power demand records refer to power demand records under special circumstances (such as emergencies, special medical operations), which are used to respond to changes in power demand under special circumstances.

[0086] Through sensors and monitoring equipment, internal and external environmental parameters and medical equipment operating status information are collected in real time, and these data are integrated with historical power consumption data and special power demand records to generate a comprehensive data set. Then, the comprehensive data set is cleaned and normalized to remove outliers and missing values ​​to ensure data quality and provide high-quality data support for subsequent analysis.

[0087] Assume that a variety of sensors are installed in the Fangcang hospital, including temperature sensors, humidity sensors, light sensors, and wind speed sensors, for real-time monitoring of internal and external environmental parameters. At the same time, the medical equipment is equipped with a status monitoring module that can report the operating status and power consumption of the equipment in real time. The data acquisition system collects data every 5 minutes and transmits the data to the central server. The data processing module on the central server cleans and normalizes the collected data to generate a comprehensive data set containing rich background information for subsequent analysis.

[0088] 102. Using a long short-term memory network combined with a seasonal autoregressive integrated moving average model, conduct an in-depth analysis of the comprehensive data set to predict the power load demand trend of the shelter hospital under different environmental conditions, apply a multi-objective genetic algorithm to optimize the power load forecasting model, and generate an accurate power load forecasting report;

[0089] Long short-term memory (LSTM) is a special recurrent neural network that can capture long-term dependencies in time series data; Seasonal Autoregressive Integrated Moving Average (SARIMA) is a time series analysis model that can handle seasonal and cyclical changes in data; Multi-objective genetic algorithm is an optimization algorithm that can find the optimal solution among multiple objectives.

[0090] In the embodiment of the present application, the comprehensive data set is preprocessed, features are extracted, and input vectors are constructed. Then, a hybrid prediction model of LSTM and SARIMA is constructed, and the advantages of both are combined to predict the power load demand. The hybrid prediction model is trained using the preprocessed data set, and a multi-objective genetic algorithm is applied to optimize the model parameters to improve the prediction accuracy. Finally, an accurate power load forecast report is generated based on the optimized model.

[0091] Optionally, in step 102, the long short-term memory network is combined with a seasonal autoregressive integrated moving average model to perform an in-depth analysis on the comprehensive data set, predict the power load demand trend of the square cabin hospital under different environmental conditions, and apply a multi-objective genetic algorithm to optimize the power load forecasting model to generate an accurate power load forecasting report, including:

[0092] The comprehensive data set is cleaned and normalized to remove outliers and missing values ​​to ensure data quality, and a preprocessed data set is obtained; feature extraction and selection are performed on the preprocessed data set to construct an input vector containing valuable features of environmental parameters and medical equipment working modes; based on the input vector, a hybrid prediction model of a long short-term memory network combined with a seasonal autoregressive integral moving average model is constructed, and the hybrid prediction model is trained to obtain a preliminary power load prediction model; the preliminary power load prediction model is optimized using a multi-objective genetic algorithm to find the best combination of model parameters and generate an optimized power load prediction model; based on the optimized power load prediction model, the power load demand of the shelter hospital under different environmental conditions is predicted to obtain an accurate power load prediction report.

[0093] In the embodiment of the present application, the comprehensive data set is cleaned and normalized, outliers and missing values ​​are removed, data quality is ensured, and a preprocessed data set is obtained. Feature extraction and selection are performed on the preprocessed data set to construct an input vector containing valuable features of environmental parameters and medical equipment working modes. Based on the input vector, a hybrid prediction model of a long short-term memory network (LSTM) combined with a seasonal autoregressive integrated moving average model (SARIMA) is constructed, and the hybrid prediction model is trained to obtain a preliminary power load forecasting model. The preliminary power load forecasting model is optimized using a multi-objective genetic algorithm to find the best combination of model parameters and generate an optimized power load forecasting model. According to the optimized power load forecasting model, the power load demand of the square cabin hospital under different environmental conditions is predicted and processed to obtain an accurate power load forecasting report.

[0094] It is assumed that the Fangcang hospital has collected a comprehensive data set, including internal and external environmental parameters (such as temperature, humidity, light intensity, wind speed), medical equipment operating status information (such as equipment startup status, working mode, power consumption), and historical power consumption data and special power demand records. First, the data processing module cleans and normalizes the comprehensive data set, removes outliers and missing values, ensures data quality, and generates a preprocessed data set. Next, the feature extraction module extracts valuable features from the preprocessed data set and constructs an input vector containing environmental parameters and medical equipment working modes. Then, based on the input vector, the model construction module constructs a hybrid prediction model of the long short-term memory network (LSTM) combined with the seasonal autoregressive integrated moving average model (SARIMA), and uses the preprocessed data set to train the hybrid prediction model to obtain a preliminary power load prediction model. Subsequently, the optimization module uses a multi-objective genetic algorithm to optimize the preliminary power load prediction model, find the best combination of model parameters, and generate an optimized power load prediction model. Finally, the prediction module predicts the power load demand of Fangcang hospitals under different environmental conditions based on the optimized power load prediction model, and generates an accurate power load prediction report.

[0095] The present application considers that in power load forecasting, the accuracy and stability of the model are key factors. In order to improve the prediction performance of the model, a multi-objective genetic algorithm can be used to optimize the preliminary power load forecasting model, find the best combination of model parameters, and generate an optimized power load forecasting model. By comprehensively considering the root mean square error, mean absolute error, mean absolute percentage error, determination coefficient, F score and prediction value variance, it can be ensured that the optimization process not only improves the prediction accuracy but also ensures the stability of the model.

[0096] Optionally, the use of a multi-objective genetic algorithm to optimize the preliminary power load forecasting model, find the best combination of model parameters, and generate an optimized power load forecasting model includes:

[0097] In order to calculate the objective function F1(w), the generalization ability of the model is first evaluated through cross-validation, and then the key parameters affecting the model performance are determined using sensitivity analysis, and finally the initial parameters of the multi-objective genetic algorithm are set;

[0098] Objective function:

[0099]

[0100] Where w represents the model parameter vector; y i is the actual power load value of the i-th sample; is the predicted power load value of the i-th sample; N is the total number of samples; α is a weight coefficient used to balance the contribution of the root mean square error and the mean absolute error in the objective function, and its value range is between 0 and 1; γ is a weight coefficient used to balance the contribution of the mean absolute percentage error in the objective function, and its value range is between 0 and 1;

[0101] After completing the calculation of F1(w), the constraint condition F2(w) is constructed by calculating the coefficient of determination, F score and variance of the predicted value of the model to ensure that the optimization process considers both prediction accuracy and model stability;

[0102] Constraints:

[0103]

[0104] Among them, R 2 (w) is the coefficient of determination, which measures the proportion of the variability of the model explaining the variables; max(R 2 ) is R 2 The maximum possible value, usually 1; F score (w) is the F score, which measures the overall performance of the model and takes into account the precision and recall of the model; max(F score ) is the maximum possible value of the F score, usually 1; Variance(w) is the variance of the model prediction value, which measures the stability of the model prediction; max(Variance) is the maximum possible value of the variance; β is the weight coefficient used to balance R 2 The contribution of the F score to the constraints ranges from 0 to 1. δ is the weight coefficient used to balance the contribution of the variance to the constraints, and ranges from 0 to 1.

[0105] These two formulas work together in the multi-objective genetic algorithm to iteratively optimize the model parameter w to minimize the objective function F1 and the constraint condition F2, thereby generating an optimized power load forecasting model.

[0106] The overall design purpose of this formula is to optimize the parameters of the power load forecasting model through a multi-objective genetic algorithm to ensure that the model finds the best balance between prediction accuracy and stability. By constructing objective functions and constraints, and comprehensively considering multiple performance indicators, an optimized power load forecasting model is generated.

[0107] The following is a brief introduction to the design reasons of each sub-item of the formula:

[0108]

[0109] Root Mean Square Error (RMSE):

[0110]

[0111] Definition: Root mean square error (RMSE) is the square root of the average of the squared errors between the predicted values ​​and the actual values.

[0112] Calculation: For each sample i, calculate the actual value y i With the predicted value The square error between Then find the average of all samples and finally take the square root.

[0113] Effect: RMSE is more sensitive to larger errors and is suitable for situations where large errors need to be strictly controlled. By minimizing RMSE, you can ensure that the model can provide more accurate predictions in most cases.

[0114] Mean Absolute Error (MAE):

[0115]

[0116] Definition: Mean absolute error (MAE) is a measure of the average absolute error between the predicted values ​​and the actual values.

[0117] Calculation: For each sample i, calculate the actual value y i With the predicted value The absolute error between Then find the average of all samples.

[0118] Effect: MAE treats all errors equally and is suitable for situations where all errors need to be considered evenly. By minimizing MAE, you can ensure that the model can provide relatively stable predictions in all cases.

[0119] Mean Absolute Percent Error (MAPE):

[0120]

[0121] Definition: Mean absolute percentage error (MAPE) is a measure of the average percentage error between the predicted values ​​and the actual values.

[0122] Calculation: For each sample i, calculate the actual value y i With the predicted value The percentage error between Then find the average of all samples.

[0123] Function: MAPE expresses the error in percentage form and is suitable for situations where relative error needs to be evaluated. By minimizing MAPE, it can ensure that the model can provide relatively consistent prediction accuracy on data of different magnitudes.

[0124] The following is a brief introduction to how to obtain the parameters of the formula:

[0125] w represents the model parameter vector; y i is the actual power load value of the i-th sample; is the predicted power load value of the i-th sample; N is the total number of samples; α is a weight coefficient used to balance the contribution of the root mean square error and the mean absolute error in the objective function, and its value range is between 0 and 1; γ is a weight coefficient used to balance the contribution of the mean absolute percentage error in the objective function, and its value range is between 0 and 1;

[0126] The following is a brief introduction to the design reasons of each sub-item of the formula:

[0127]

[0128] Coefficient of determination R 2 :

[0129]

[0130] Definition: Coefficient of determination R 2 Measures the proportion of the variability in the variables explained by the model, reflecting the goodness of fit of the model.

[0131] Calculation: R 2 (w) is the coefficient of determination of the model, max(R 2 ) is R 2 The maximum possible value of , usually 1.

[0132] Role: R 2 The closer it is to 1, the better the model fit is. This ensures that the model has a high degree of fit.

[0133] Weight coefficient β: used to balance R 2 The contribution to the constraint condition ranges from 0 to 1. A larger β value indicates more emphasis on R 2 .

[0134] Score F score :

[0135]

[0136] Definition: F score score Measures the overall performance of the model, taking into account the model's precision and recall.

[0137] Calculation: F score (w) is the F score of the model, max(F score ) is the maximum possible value of the F-score, usually 1.

[0138] Function: The higher the F score, the better the overall performance of the model. This can ensure that the model has high comprehensive performance.

[0139] Weight coefficient 1-β: used to balance the contribution of the F score in the constraints, and the value range is between 0 and 1. A larger 1-β value means more emphasis on the F score.

[0140] Variance of predicted value:

[0141]

[0142] Definition: Variance of predicted values ​​measures the stability of the model's predicted values ​​and reflects the robustness of the model.

[0143] Calculation: Variance(w) is the variance of the model's predictions, and max(Variance) is the maximum possible value of the variance.

[0144] Function: The smaller the variance, the more stable the model's prediction value. This ensures that the model has high stability.

[0145] Weight coefficient δ: used to balance the contribution of variance in the constraints, and its value range is between 0 and 1. A larger δ value means more emphasis on variance.

[0146] The following is a brief introduction to how to obtain the parameters of the formula:

[0147] R 2 (w) is the coefficient of determination, which measures the proportion of the variability of the model explaining the variables; max(R 2 ) is R 2 The maximum possible value, usually 1; F score (w) is the F score, which measures the overall performance of the model and takes into account the precision and recall of the model; max(F score ) is the maximum possible value of the F score, usually 1; Variance(w) is the variance of the model prediction value, which measures the stability of the model prediction; max(Variance) is the maximum possible value of the variance; β is the weight coefficient used to balance R 2 and F score The contribution of variance to the constraint condition ranges from 0 to 1. δ is the weight coefficient used to balance the contribution of variance to the constraint condition, and ranges from 0 to 1.

[0148] Assume that there is a power load forecasting model that needs to be optimized through a multi-objective genetic algorithm. The known data are as follows: N = 100 (total number of samples), α = 0.5 (weight coefficient, used to balance the root mean square error and the mean absolute error), γ = 0.2 (weight coefficient, used to balance the mean absolute percentage error), β = 0.4 (weight coefficient, used to balance R 2 and F score)， δ=0.3 (weight coefficient, used to balance variance).

[0149] First, calculate the objective function F1(w):

[0150] Assume that the actual value y i =[100,105,110,…,200](actual power load value), predicted value (Forecasted power load value).

[0151] Calculate the root mean square error (RMSE): Calculate the mean absolute error (MAE):

[0152] Calculate the mean absolute percentage error (MAPE):

[0153] Substituting into the objective function F1(w): F1(w)=0.5·2+0.5·2+0.2·0.0133=1+1+0.00266=2.00266.

[0154] Next, calculate the constraint F2(w): the coefficient of determination R of the assumed model 2 (w) = 0.95, F score F score (w)=0.90, predicted value variance Variance(w)=10.

[0155] F2(w)=0.4.

[0156] The power load forecasting model optimized by the multi-objective genetic algorithm has an objective function F1(w) value of 2.00266, which is lower than the set threshold of 2.5, indicating that the model performs well in terms of prediction accuracy and has a small error. The value of the constraint condition F2(w) is 0.35, which is lower than the set threshold of 0.5, indicating that the model performs well in terms of stability and can maintain a high degree of fit and comprehensive performance. At the same time, the variance of the predicted value is small, and the model has high robustness. Finally, the optimized power load forecasting model can better serve the power system management of the Fangcang Hospital and ensure the efficient operation of the power system.

[0157] Optionally, the extracting and selecting features of the preprocessed data set to construct an input vector containing valuable features of environmental parameters and medical device working modes includes:

[0158] Utilize statistical analysis methods to perform correlation analysis on various indicators in the preprocessed data set, identify features closely related to the power load, and obtain a list of features with high correlation; select features with high correlation as components of the input vector based on the list of features with high correlation, and generate a selected feature set; encode the selected feature set, convert categorical features into numerical data using unique hot encoding or label encoding, and standardize or normalize continuous features according to actual needs to obtain a standardized feature set; construct a final input vector based on the standardized feature set to obtain an accurate power load forecast report.

[0159] In an embodiment of the present application, a statistical analysis method is used to perform a correlation analysis on various indicators in the preprocessed data set, identify features that are closely related to the power load, and obtain a list of features with high correlation. Based on the list of features with high correlation, features with high correlation are selected as components of the input vector to generate a selected feature set. The selected feature set is encoded, and for categorical features, unique hot encoding or label encoding is used to convert them into numerical data. For continuous features, standardization or normalization is performed according to actual needs to obtain a standardized feature set. Based on the standardized feature set, the final input vector is constructed to generate an accurate power load forecast report.

[0160] Assume that the Fangcang Hospital has cleaned and normalized the comprehensive data set to obtain a preprocessed data set, which includes internal and external environmental parameters (such as temperature, humidity, light intensity, wind speed) and medical equipment operating status information (such as equipment startup status, working mode, power consumption). First, the statistical analysis method is used to perform correlation analysis on the various indicators in the preprocessed data set, identify the features closely related to the power load, and obtain a list of features with high correlation, such as temperature, humidity, and power consumption of certain medical equipment. Then, based on the list of features with high correlation, features with high correlation are selected as components of the input vector to generate a selected feature set. Next, the selected feature set is encoded, and the categorical features (such as equipment working mode) are converted into numerical data using unique hot encoding or label encoding. For continuous features (such as temperature and humidity), they are standardized or normalized according to actual needs to obtain a standardized feature set. Finally, based on the standardized feature set, the final input vector is constructed to train a hybrid prediction model of a long short-term memory network (LSTM) combined with a seasonal autoregressive integrated moving average model (SARIMA) to generate an accurate power load forecast report.

[0161] Optionally, based on the input vector, a hybrid prediction model of a long short-term memory network combined with a seasonal autoregressive integrated moving average model is constructed, and the hybrid prediction model is trained to obtain a preliminary power load prediction model, including:

[0162] The input vector is used to design a long short-term memory network structure, and an LSTM model part capable of capturing the medium- and long-term dependencies in time series data is constructed to obtain an LSTM model structure; according to the time characteristics of the input vector, a suitable seasonal cycle is selected to construct a seasonal autoregressive integrated moving average model part to obtain a SARIMA model structure; based on the LSTM model structure and the SARIMA model structure, an architecture of a hybrid prediction model is designed to ensure that the two model parts can be effectively combined to generate a hybrid prediction model architecture; the preprocessed data set and the standardized feature set are used to train the hybrid prediction model architecture, and a preliminary power load prediction model is generated by iteratively optimizing model parameters.

[0163] In the embodiment of the present application, it is assumed that the square cabin hospital has generated a standardized feature set, including internal and external environmental parameters (such as temperature, humidity, light intensity, wind speed) and medical equipment operating status information (such as equipment startup status, working mode, power consumption). First, the long short-term memory network (LSTM) structure is designed using these feature sets to construct an LSTM model part that can capture long-term dependencies in time series data, such as setting multiple LSTM layers and hidden units to ensure that the model can effectively learn long-term trends in historical data. Then, according to the time characteristics of the input vector, a suitable seasonal cycle (such as daily cycle, weekly cycle) is selected to construct a seasonal autoregressive integrated moving average model (SARIMA) part, such as setting seasonal parameters (p, d, q, P, D, Q) to ensure that the model can handle seasonal and periodic changes in the data. Next, based on the LSTM model structure and the SARIMA model structure, the architecture of the hybrid prediction model is designed to ensure that the two model parts can be effectively combined to complement each other's advantages, such as merging the output of the LSTM and the output of the SARIMA through a fusion layer to generate a hybrid prediction model architecture. Finally, the hybrid prediction model architecture is trained using the preprocessed data set and the standardized feature set. By iteratively optimizing the model parameters, a preliminary power load forecasting model is generated to ensure that the model has high prediction accuracy and generalization ability on the training data.

[0164] 103. Based on the power load forecast report, a mixed integer linear programming technique is used to formulate a power distribution plan, fuzzy logic control technology is introduced to evaluate the flexibility of the power system, and the start and stop timing of non-critical loads is dynamically adjusted to cope with uncertain power supply fluctuations and generate an optimal power distribution and scheduling plan;

[0165] Mixed integer linear programming (MILP) is an optimization technique used to solve optimization problems involving discrete variables and continuous variables; fuzzy logic control technology is a control method for dealing with uncertainty and complexity, and can evaluate the flexibility of the system; non-critical loads refer to loads that have little impact on the normal operation of the shelter hospital, and the start and stop timing can be adjusted within a certain range.

[0166] In the embodiment of the present application, based on the power load forecast report, the future power demand trend is analyzed to determine the basic framework of power distribution. Using MILP technology, combined with the power supply capacity of the square cabin hospital and the priority of various loads, the power distribution is modeled to generate a preliminary power distribution plan. Fuzzy logic control technology is introduced to evaluate the flexibility of the power system and generate a power system flexibility assessment report. According to the flexibility assessment report, the start and stop timing of non-critical loads is dynamically adjusted to generate an optimized start and stop strategy. Finally, based on the preliminary power distribution plan and the optimized start and stop strategy, the optimal power distribution and scheduling plan is generated.

[0167] Optionally, in step 103, according to the power load forecast report, a power distribution plan is formulated using a mixed integer linear programming technique, fuzzy logic control technology is introduced to evaluate the flexibility of the power system, and the start and stop timing of non-critical loads is dynamically adjusted to cope with uncertain power supply fluctuations, thereby generating an optimal power distribution and scheduling plan, including:

[0168] According to the power load forecast report, the future power demand trend is analyzed, the basic framework of power distribution is determined, and the power demand analysis results are obtained; using the power demand analysis results, the mixed integer linear programming technology is adopted, combined with the existing power supply capacity of the cabin hospital and the priority of various loads, the power distribution is modeled and processed to generate a preliminary power distribution plan; fuzzy logic control technology is introduced, and based on the preliminary power distribution plan, the flexibility of the power system is evaluated to generate a power system flexibility evaluation report; according to the power system flexibility evaluation report, the start and stop timing of non-critical loads is dynamically adjusted, the best start and stop strategy is determined by the optimization algorithm, and the optimized start and stop strategy is generated; based on the preliminary power distribution plan and the optimized start and stop strategy, the accuracy of power load forecasting, the economy of power distribution and the flexibility of the power system are comprehensively considered to generate the optimal power distribution and scheduling plan.

[0169] In the embodiment of the present application, based on the power load forecast report, the future power demand trend is analyzed, the basic framework of power distribution is determined, and the power demand analysis results are obtained. Using the power demand analysis results, the mixed integer linear programming technology is adopted, combined with the existing power supply capacity of the square cabin hospital and the priority of various loads, the power distribution is modeled and processed to generate a preliminary power distribution plan. Fuzzy logic control technology is introduced, and based on the preliminary power distribution plan, the flexibility of the power system is evaluated to generate a power system flexibility evaluation report. According to the power system flexibility evaluation report, the start and stop timing of non-critical loads is dynamically adjusted, and the optimal start and stop strategy is determined by the optimization algorithm to generate an optimized start and stop strategy. Based on the preliminary power distribution plan and the optimized start and stop strategy, the accuracy of the power load forecast, the economy of power distribution and the flexibility of the power system are comprehensively considered to generate the optimal power distribution and scheduling plan.

[0170] Assume that the Fangcang Hospital has generated a power load forecast report, predicting the power demand trend under different environmental conditions in the future. First, according to the power load forecast report, the future power demand trend is analyzed to determine the basic framework of power distribution, such as increasing the power supply of key medical equipment during peak hours and reducing the power consumption of non-critical loads during off-peak hours, and obtaining the power demand analysis results. Then, using the power demand analysis results, the mixed integer linear programming (MILP) technology is used to model the power distribution, combined with the existing power supply capacity of the Fangcang Hospital and the priority of various loads (such as the highest priority for key medical equipment and the lower priority for non-critical loads), and generate a preliminary power distribution plan to ensure that the power demand of key equipment is met. Next, the fuzzy logic control technology is introduced to evaluate the flexibility of the power system based on the preliminary power distribution plan, and generate a power system flexibility evaluation report, which includes the system's responsiveness and stability under different operating conditions. According to the power system flexibility assessment report, the start and stop timing of non-critical loads is dynamically adjusted, and the best start and stop strategy is determined through optimization algorithms (such as genetic algorithms or particle swarm optimization algorithms). The optimized start and stop strategy is generated to ensure that the start and stop timing of non-critical loads can be flexibly adjusted when the power supply fluctuates to maintain the stable operation of the system. Finally, based on the preliminary power distribution plan and the optimized start and stop strategy, the accuracy of power load forecasting, the economy of power distribution, and the flexibility of the power system are comprehensively considered to generate the optimal power distribution and scheduling plan to ensure that the Fangcang Hospital can operate efficiently and stably under various environmental conditions.

[0171] Optionally, the power demand analysis result is used to adopt a mixed integer linear programming technique, combined with the existing power supply capacity of the square cabin hospital and the priority of various loads, to model the power distribution and generate a preliminary power distribution plan, including:

[0172] Using the power demand analysis results, determine the total power demand and its distribution characteristics in a specific time period in the future, and generate a power demand distribution map; based on the power demand distribution map and in combination with the existing power supply capacity of the Fangcang Cabin Hospital, conduct a detailed assessment of the power supply capacity and generate a power supply capacity assessment report; based on the power supply capacity assessment report, determine the priority of each type of load and generate a load priority list; using mixed integer linear programming technology, establish a mathematical model of power distribution based on the load priority list and the power demand distribution map, and generate a power distribution model; solve the power distribution model, find the optimal power distribution plan by setting the objective function and constraints, and generate a preliminary power distribution plan.

[0173] In the embodiment of the present application, based on the power load forecast report, the future power demand trend is analyzed, the basic framework of power distribution is determined, and the power demand analysis results are obtained. Using the power demand analysis results, the mixed integer linear programming technology is adopted, combined with the existing power supply capacity of the square cabin hospital and the priority of various loads, the power distribution is modeled and processed to generate a preliminary power distribution plan. Fuzzy logic control technology is introduced, and based on the preliminary power distribution plan, the flexibility of the power system is evaluated to generate a power system flexibility evaluation report. According to the power system flexibility evaluation report, the start and stop timing of non-critical loads is dynamically adjusted, and the optimal start and stop strategy is determined by the optimization algorithm to generate an optimized start and stop strategy. Based on the preliminary power distribution plan and the optimized start and stop strategy, the accuracy of the power load forecast, the economy of power distribution and the flexibility of the power system are comprehensively considered to generate the optimal power distribution and scheduling plan.

[0174] Assume that the Fangcang Hospital has generated a power load forecast report, predicting the power demand in different time periods in the next week. First, based on the power load forecast report, the future power demand trend is analyzed to determine the basic framework of power distribution, such as increasing the power supply of key medical equipment during peak hours (such as 9 am to 5 pm) and reducing the power consumption of non-critical loads during trough hours (such as 1 am to 5 am), and obtaining the power demand analysis results. Then, using the power demand analysis results, the mixed integer linear programming technology is used, combined with the existing power supply capacity of the Fangcang Hospital and the priority of various loads (such as key medical equipment has the highest priority, and non-critical loads have a lower priority), the power distribution is modeled and processed, and a preliminary power distribution plan is generated to ensure that the power demand of key equipment is met. Next, the fuzzy logic control technology is introduced to evaluate the flexibility of the power system based on the preliminary power distribution plan, and a power system flexibility evaluation report is generated, which includes the responsiveness and stability of the system under different operating conditions. According to the power system flexibility assessment report, the start and stop timing of non-critical loads is dynamically adjusted, and the best start and stop strategy is determined through optimization algorithms (such as genetic algorithms or particle swarm optimization algorithms). The optimized start and stop strategy is generated to ensure that the start and stop timing of non-critical loads can be flexibly adjusted when the power supply fluctuates to maintain the stable operation of the system. Finally, based on the preliminary power distribution plan and the optimized start and stop strategy, the accuracy of power load forecasting, the economy of power distribution, and the flexibility of the power system are comprehensively considered to generate the optimal power distribution and scheduling plan to ensure that the Fangcang Hospital can operate efficiently and stably under various environmental conditions.

[0175] This application takes into account that in the power system, the start and stop timing of non-critical loads has an important impact on the stability and economy of the power system. In order to ensure the efficient operation of the power system, it is necessary to dynamically adjust the start and stop timing of non-critical loads according to the power system flexibility assessment report. By determining the optimal start and stop strategy through optimization algorithms, the flexibility and resource utilization of the power system can be improved, and the operating costs can be reduced.

[0176] Optionally, dynamically adjusting the start and stop timing of non-critical loads according to the power system flexibility assessment report, determining the best start and stop strategy through an optimization algorithm, and generating an optimized start and stop strategy includes:

[0177] When calculating the startup time T of non-critical loads on (i) Previously, it was necessary to analyze the data from the power system flexibility assessment report, identify the characteristics of non-critical loads, and determine the benchmark start time and allocation weights;

[0178]

[0179] Among them, T on (i) represents the startup time of the i-th non-critical load; t0 is the benchmark startup time, which is usually the start time of the low power demand period; N is the number of non-critical loads; w j is the weight of the jth non-critical load, reflecting its importance in power distribution; P j is the power demand of the jth non-critical load; P min It is the smallest power requirement among all non-critical loads; F j is the flexibility score of the jth non-critical load; F max The highest flexibility score among all non-critical loads; D j is the startup delay time of the jth non-critical load; λ is the influence coefficient of the delay time;

[0180] After calculating T on (i), the impact of different start-stop schedules is evaluated, and the optimal start-stop strategy S(i) is found using an optimization algorithm;

[0181]

[0182] Where S(i) represents the optimal start-stop strategy for the i-th non-critical load; s i is the start and stop schedule of the i-th non-critical load; C(s i ) is the start and stop schedule i The total cost under the condition includes the startup cost and the running cost; V(s i ) is the start and stop schedule i Power system volatility under i) is the start and stop schedule i α is the cost weight coefficient, ranging from 0 to 1; β is the load balancing weight coefficient, ranging from 0 to 1; γ is the balance coefficient between volatility and load balancing; R(s i ) is the start and stop schedule i Resource utilization under

[0183] Based on S(i), a detailed start-stop schedule is generated, its effect is verified through simulation or testing, and the optimization is adjusted according to the results to form the final start-stop strategy.

[0184] The overall design goal of the formula is to calculate the optimal start time for each non-critical load by comprehensively considering the characteristics of non-critical loads, power demand trough period, flexibility score and start delay time. Then, by evaluating the impact of different start and stop schedules, an optimization algorithm is used to find the best start and stop strategy to ensure the stability and economy of the power system.

[0185] The following is a brief introduction to the design reasons of each sub-item of the formula:

[0186]

[0187] Benchmark start time t0: Select the start time of the low power demand period as the benchmark start time to reduce the power load at startup and avoid starting non-critical loads during the peak power demand period, thereby reducing the pressure on the power system and improving the overall efficiency of the system. j :Reflects the importance of each non-critical load in power distribution. By giving different weights to different loads, it ensures that important loads are started first, improving the flexibility and reliability of the system. The startup time of important loads should be given priority to ensure the normal operation of key services. Power demand P j and the minimum power requirement P min :Power demand P j represents the power demand of the jth non-critical load, and P min is the smallest power requirement among all non-critical loads. By calculating P j -P min , can highlight the importance of high power demand loads, ensure that these loads are started at the right time, and avoid overloading the power system. Flexibility score F j and maximum flexibility score F max : Flexibility score F j Reflects the flexibility of the jth non-critical load, that is, its ability to start at different times. max is the highest flexibility score among all non-critical loads. The startup time can be adjusted to ensure that highly flexible loads can start quickly when needed, improving the responsiveness of the system. j And the delay time influence coefficient λ: start delay time D j represents the startup delay time of the jth non-critical load, that is, the time required from the issuance of the command to the actual startup. The delay time influence coefficient λ is used to adjust the impact of the delay time on the startup time. By introducing exp(-λ·D j ), can ensure that loads with longer delay times are started at the appropriate time, avoiding electricity charges caused by delays. The weight, power demand, flexibility score and startup delay time are taken into account to calculate the startup time contribution of each non-critical load. The larger the product of these factors, the greater the startup time contribution of the load, indicating that the load has a higher priority in startup time. The value used to normalize the numerator ensures that the startup time calculation result is within a reasonable range. The denominator also takes into account the weight, power demand and startup delay time, but does not take into account the flexibility score to keep the calculation relatively simple and stable.

[0188] The following is a brief introduction to how to obtain the parameters of the formula:

[0189]

[0190] T on (i) represents the startup time of the i-th non-critical load; t0 is the benchmark startup time, which is usually the start time of the low power demand period; N is the number of non-critical loads; w j is the weight of the jth non-critical load, reflecting its importance in power distribution; P j is the power demand of the jth non-critical load; P min It is the smallest power requirement among all non-critical loads; F j is the flexibility score of the jth non-critical load; F max The highest flexibility score among all non-critical loads; D j is the startup delay time of the jth non-critical load; λ is the influence coefficient of the delay time;

[0191] The following is a brief introduction to the design reasons of each sub-item of the formula:

[0192]

[0193] The optimal start-stop strategy S(i) represents the optimal start-stop strategy for the i-th non-critical load. The optimal start-stop strategy is found through the optimization algorithm to ensure the stability and economy of the power system. irepresents the start and stop schedule of the i-th non-critical load. Different start and stop schedules will affect the cost, volatility, load balance and resource utilization of the power system, so it is necessary to find the optimal start and stop schedule. Cost C(s i ) indicates the start and stop time schedule s i The total cost under the power system includes the startup cost and the operating cost. The cost is an important indicator of the power system operation. By minimizing the cost, the operating cost of the power system can be reduced and the economy can be improved. i ) indicates the start and stop time schedule s i The power system volatility under load reflects the imbalance between power supply and demand. The power system volatility is an important indicator for evaluating system stability. By minimizing the volatility, the stability and reliability of the power system can be improved. Load balancing L(s i ) indicates the start and stop time schedule s i The load balance under the load reflects the uniformity of power distribution among the loads. Load balance is an important indicator for evaluating whether the resource allocation of the power system is reasonable. By optimizing the load balance, local overload can be avoided and the overall efficiency of the system can be improved. Resource utilization R(s i ) indicates the start and stop time schedule s i The resource utilization rate under the condition reflects the effective utilization of power resources. Resource utilization is an important indicator for evaluating the efficiency of resource utilization in the power system. By maximizing resource utilization, the overall efficiency and economy of the system can be improved. The cost weight coefficient α represents the weight of cost in the optimization target, and the value range is between 0 and 1. By adjusting the value of α, a balance can be found between cost and other factors to ensure the comprehensiveness of the optimization target. The balance coefficient γ of volatility and load balance represents the balance coefficient of volatility and load balance in the optimization target. By adjusting the value of γ, a balance can be found between volatility and load balance to ensure the comprehensiveness of the optimization target. The load balance weight coefficient β represents the weight of resource utilization in the optimization target, and the value range is between 0 and 1. By adjusting the value of β, a balance can be found between resource utilization and other factors to ensure the comprehensiveness of the optimization target. Through the design of the above sub-items, formula S(i) can comprehensively consider cost, power system volatility, load balance and resource utilization, find the best start-stop strategy, help improve the stability and economy of the power system, and ensure the efficient operation of the power system of the square cabin hospital.

[0194] The following is a brief introduction to how to obtain the parameters of the formula:

[0195] S(i) represents the optimal start-stop strategy for the i-th non-critical load; s i is the start and stop schedule of the i-th non-critical load; C(s i) is the start and stop schedule i The total cost under the condition includes the startup cost and the running cost; V(s i ) is the start and stop schedule i Power system volatility under i ) is the start and stop schedule i α is the cost weight coefficient, ranging from 0 to 1; β is the load balancing weight coefficient, ranging from 0 to 1; γ is the balance coefficient between volatility and load balancing; R(s i ) is the start and stop schedule i Resource utilization under

[0196] The following is a brief introduction to the design reasons of each sub-item of the formula:

[0197]

[0198] Benchmark start time t0: Select the start time of the low power demand period as the benchmark start time to reduce the power load at startup and avoid starting non-critical loads during the peak power demand period, thereby reducing the pressure on the power system and improving the overall efficiency of the system. j :Reflects the importance of each non-critical load in power distribution. By giving different weights to different loads, it ensures that important loads are started first, improving the flexibility and reliability of the system. The startup time of important loads should be given priority to ensure the normal operation of key services. Power demand P j and the minimum power requirement P min :Power demand P j represents the power demand of the jth non-critical load, and P min is the smallest power requirement among all non-critical loads. By calculating P j -P min , can highlight the importance of high power demand loads, ensure that these loads are started at the right time, and avoid overloading the power system. Flexibility score F j and maximum flexibility score F max : Flexibility score F j Reflects the flexibility of the jth non-critical load, that is, its ability to start at different times. max is the highest flexibility score among all non-critical loads. The startup time can be adjusted to ensure that highly flexible loads can start quickly when needed, improving the responsiveness of the system. j And the delay time influence coefficient λ: start delay time D jrepresents the startup delay time of the jth non-critical load, that is, the time required from the issuance of the command to the actual startup. The delay time influence coefficient λ is used to adjust the impact of the delay time on the startup time. By introducing exp(-λ·D j ), can ensure that loads with longer delay times are started at the appropriate time, avoiding electricity charges caused by delays. The weight, power demand, flexibility score and startup delay time are taken into account to calculate the startup time contribution of each non-critical load. The larger the product of these factors, the greater the startup time contribution of the load, indicating that the load has a higher priority in startup time. The value used to normalize the numerator ensures that the startup time calculation result is within a reasonable range. The denominator also takes into account the weight, power demand and startup delay time, but does not take into account the flexibility score to keep the calculation relatively simple and stable.

[0199] The following is a brief introduction to how to obtain the parameters of the formula:

[0200]

[0201] T on (i) represents the startup time of the i-th non-critical load; t0 is the benchmark startup time, which is usually the start time of the low power demand period; N is the number of non-critical loads; w j is the weight of the jth non-critical load, reflecting its importance in power distribution; P j is the power demand of the jth non-critical load; P min It is the smallest power requirement among all non-critical loads; F j is the flexibility score of the jth non-critical load; F max The highest flexibility score among all non-critical loads; D j is the startup delay time of the jth non-critical load; λ is the influence coefficient of the delay time;

[0202] The following is a brief introduction to the design reasons of each sub-item of the formula:

[0203]

[0204] The optimal start-stop strategy S(i) represents the optimal start-stop strategy for the i-th non-critical load. The optimal start-stop strategy is found through the optimization algorithm to ensure the stability and economy of the power system. i represents the start and stop schedule of the i-th non-critical load. Different start and stop schedules will affect the cost, volatility, load balance and resource utilization of the power system, so it is necessary to find the optimal start and stop schedule. Cost C(s i) indicates the start and stop time schedule s i The total cost under the power system includes the startup cost and the operating cost. The cost is an important indicator of the power system operation. By minimizing the cost, the operating cost of the power system can be reduced and the economy can be improved. i ) indicates the start and stop time schedule s i The power system volatility under load reflects the imbalance between power supply and demand. The power system volatility is an important indicator for evaluating system stability. By minimizing the volatility, the stability and reliability of the power system can be improved. Load balancing L(s i ) indicates the start and stop time schedule s i The load balance under the load reflects the uniformity of power distribution among the loads. Load balance is an important indicator for evaluating whether the resource allocation of the power system is reasonable. By optimizing the load balance, local overload can be avoided and the overall efficiency of the system can be improved. Resource utilization R(s i ) indicates the start and stop time schedule s i The resource utilization rate under the condition reflects the effective utilization of power resources. Resource utilization is an important indicator for evaluating the efficiency of resource utilization in the power system. By maximizing resource utilization, the overall efficiency and economy of the system can be improved. The cost weight coefficient α represents the weight of cost in the optimization target, and the value range is between 0 and 1. By adjusting the value of α, a balance can be found between cost and other factors to ensure the comprehensiveness of the optimization target. The balance coefficient γ of volatility and load balance represents the balance coefficient of volatility and load balance in the optimization target. By adjusting the value of γ, a balance can be found between volatility and load balance to ensure the comprehensiveness of the optimization target. The load balance weight coefficient β represents the weight of resource utilization in the optimization target, and the value range is between 0 and 1. By adjusting the value of β, a balance can be found between resource utilization and other factors to ensure the comprehensiveness of the optimization target. Through the design of the above sub-items, formula S(i) can comprehensively consider cost, power system volatility, load balance and resource utilization, find the best start-stop strategy, help improve the stability and economy of the power system, and ensure the efficient operation of the power system of the square cabin hospital.

[0205] The following is a brief introduction to how to obtain the parameters of the formula:

[0206] S(i) represents the optimal start-stop strategy for the i-th non-critical load; s i is the start and stop schedule of the i-th non-critical load; C(s i ) is the start and stop schedule i The total cost under the condition includes the startup cost and the running cost; V(s i ) is the start and stop schedule i Power system volatility underi ) is the start and stop schedule i α is the cost weight coefficient, ranging from 0 to 1; β is the load balancing weight coefficient, ranging from 0 to 1; γ is the balance coefficient between volatility and load balancing; R(s i ) is the start and stop schedule i Resource utilization under

[0207] Assume that the Fangcang Hospital has three non-critical loads, namely load A, load B and load C. The known data are as follows: t0 = 2 hours (the start time of the low power demand period), N = 3 (the number of non-critical loads), w1 = 0.5, w2 = 0.3, w3 = 0.2 (weight), P1 = 100kW, P2 = 80kW, P3 = 60kW (power demand), P min =60kW (minimum power requirement), F1=0.8, F2=0.7, F3=0.6 (flexibility score), F max =0.8 (maximum flexibility score), D1 = 0.5 hours, D2 = 0.3 hours, D3 = 0.4 hours (startup delay time), λ = 0.1 (delay time influence coefficient), α = 0.5 (cost weight coefficient), β = 0.3 (load balancing weight coefficient), γ = 0.2 (balance coefficient between volatility and load balancing). First, calculate the startup time T of each non-critical load on (i):

[0208]

[0209] Calculation results:

[0210] T on (1)≈2.029

[0211] Similarly, calculate T on (2) and T on (3):T on (2)≈2.023T on (3)≈2.018 Next, we evaluate the impact of different start-stop schedules and use an optimization algorithm to find the best start-stop strategy S(i): Assuming a start-stop schedule s i The parameters are as follows: C(s1)=1000 yuan, V(s1)=0.05, L(s1)=0.9, R(s1)=0.85.

[0212] Calculate S(1):

[0213]

[0214] Similarly, calculate S(2) and S(3):

[0215]

[0216] The smaller S(1), S(2), and S(3) are, the better the overall performance of the start-stop strategy is after considering all factors. Minimizing this objective function value means finding the best balance point between these factors, thereby ensuring the efficient operation of the power system. Therefore, 500.16, 500.18, and 500.20 are the smallest among all possible start-stop schedules, indicating the best start-stop strategy. Finally, based on the calculation results, a detailed start-stop schedule is generated, and its effect is verified by simulation or testing. Assume that the simulation results show that this start-stop strategy performs well in terms of cost, volatility, load balancing, and resource utilization, and is finally confirmed as the best start-stop strategy. The results show that the optimal start-stop strategy determined by the optimization algorithm can effectively reduce the operating cost of the power system, improve the stability of the system and resource utilization, and ensure the efficient operation of the power system of the square cabin hospital.

[0217] 104. Based on the optimal power distribution and dispatching scheme, regular system performance evaluation is performed, and algorithm parameters are adjusted according to the evaluation results to generate an intelligent management plan for the square cabin hospital power system.

[0218] System performance evaluation refers to the regular evaluation of the operating efficiency, stability and economy of the power system; algorithm parameter adjustment refers to adjusting the parameters of the optimization algorithm based on the evaluation results to further improve system performance. Regularly collect actual operating data of the power system, including power consumption, load operating status, etc. Based on the collected data, evaluate the operating efficiency, stability and economy of the power system, and generate a system performance evaluation report. According to the evaluation results, adjust the parameters of the optimization algorithm, such as weight coefficients, constraints, etc. Finally, based on the adjusted parameters, generate an intelligent management plan for the square cabin hospital power system to ensure efficient and stable operation of the system.

[0219] In the embodiment of the present application, the actual operation data of the power system is collected regularly, including power consumption, load operation status, etc. Based on the collected data, the operation efficiency, stability and economy of the power system are evaluated, and a system performance evaluation report is generated. According to the evaluation results, the parameters of the optimization algorithm, such as weight coefficients, constraints, etc., are adjusted. Finally, based on the adjusted parameters, an intelligent management plan for the square cabin hospital power system is generated to ensure efficient and stable operation of the system.

[0220] Optionally, based on the optimal power distribution and dispatching scheme described in step 104, system performance evaluation is performed regularly, algorithm parameters are adjusted according to the evaluation results, and an intelligent management scheme for the square cabin hospital power system is generated, including:

[0221] Using a predetermined time period, data is collected from the electric power system of the square cabin hospital, including power consumption, actual power consumption of each load, and working status of power equipment, to obtain actual operation data; based on the actual operation data, the difference between the actual power distribution and scheduling situation and the optimal power distribution and scheduling plan is compared, the performance indicators of the power system are evaluated, and a system performance evaluation report is generated; based on the system performance evaluation report, the existing problems and potential improvement space are deeply analyzed to generate a problem analysis report; using the problem analysis report, the algorithm parameters used to generate the optimal power distribution and scheduling plan are appropriately adjusted to generate adjusted algorithm parameters; based on the adjusted algorithm parameters, the new power distribution and scheduling plan is tested in a simulation environment to generate a test verification report; using the test verification report, the new plan that has been tested and verified is applied to the actual power system of the square cabin hospital, the operating status of the system is continuously monitored, data on the implementation effect is collected, and implementation monitoring data is generated; based on the implementation monitoring data, a complete set of intelligent management plans for the square cabin hospital power system is formed to generate an intelligent management plan.

[0222] In the embodiment of the present application, a predetermined time period is used to collect data from the power system of the square cabin hospital, including power consumption, actual power consumption of each load, and working status of power equipment, to obtain actual operation data. Based on the actual operation data, the difference between the actual power distribution and scheduling situation and the optimal power distribution and scheduling scheme is compared, the performance indicators of the power system are evaluated, and a system performance evaluation report is generated. According to the system performance evaluation report, the existing problems and potential improvement space are deeply analyzed to generate a problem analysis report. Using the problem analysis report, the algorithm parameters used to generate the optimal power distribution and scheduling scheme are appropriately adjusted to generate the adjusted algorithm parameters. Based on the adjusted algorithm parameters, the new power distribution and scheduling scheme is tested in a simulation environment to generate a test verification report. Using the test verification report, the new scheme that has been tested and verified is applied to the actual power system of the square cabin hospital, the operating status of the system is continuously monitored, the data of the implementation effect is collected, and the implementation monitoring data is generated. Based on the implementation monitoring data, a complete set of intelligent management solutions for the square cabin hospital power system is formed, and an intelligent management solution is generated.

[0223] Figure 2 A schematic diagram of the structure of an intelligent square cabin hospital power load forecasting and dispatching system is provided for the embodiment of the present application, such as Figure 2 As shown, the device comprises:

[0224] The collection module 21 is used to collect and integrate the environmental parameters inside and outside the Fangcang Hospital and the operating status information of medical equipment in real time, collect historical power consumption data and special power demand records, and generate a comprehensive data set containing rich background information;

[0225] An analysis module 22 is used to use a long short-term memory network combined with a seasonal autoregressive integrated moving average model to perform in-depth analysis on the comprehensive data set, predict the power load demand trend of the square cabin hospital under different environmental conditions, and use a multi-objective genetic algorithm to optimize the power load forecasting model to generate an accurate power load forecasting report;

[0226] The adjustment module 23 is used to formulate a power distribution plan based on the power load forecast report by using mixed integer linear programming technology, introduce fuzzy logic control technology to evaluate the flexibility of the power system, dynamically adjust the start and stop timing of non-critical loads to cope with uncertain power supply fluctuations, and generate an optimal power distribution and scheduling plan;

[0227] The evaluation module 24 is used to regularly evaluate system performance based on the optimal power distribution and scheduling plan, adjust algorithm parameters according to the evaluation results, and generate an intelligent management plan for the square cabin hospital power system.

[0228] Figure 2 The intelligent square cabin hospital power load forecasting and dispatching system can perform Figure 1 The implementation principle and technical effects of the intelligent square cabin hospital power load forecasting and scheduling method described in the embodiment are not repeated here. The specific way in which each module and unit performs operations in the intelligent square cabin hospital power load forecasting and scheduling system in the above embodiment has been described in detail in the embodiment of the method, and will not be elaborated here.

[0229] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for predicting and dispatching power load of an intelligent shelter hospital, characterized in that: include: Collect and integrate the indoor and outdoor environmental parameters of Fangcang Cabin Hospital and the operating status of medical equipment in real time, collect historical power consumption data and special power demand records, and generate a comprehensive data set containing rich background information; The long short-term memory network is combined with the seasonal autoregressive integrated moving average model to conduct an in-depth analysis of the comprehensive data set, predict the power load demand trend of the shelter hospital under different environmental conditions, and apply the multi-objective genetic algorithm to optimize the power load forecasting model to generate an accurate power load forecasting report; According to the power load forecast report, a mixed integer linear programming technique is used to formulate a power distribution plan, fuzzy logic control technology is introduced to evaluate the flexibility of the power system, and the start and stop timing of non-critical loads is dynamically adjusted to cope with uncertain power supply fluctuations, thereby generating an optimal power distribution and scheduling plan; Based on the optimal power distribution and scheduling scheme, system performance evaluation is carried out regularly, and algorithm parameters are adjusted according to the evaluation results to generate an intelligent management plan for the square cabin hospital power system.

2. The method according to claim 1, characterized in that The long short-term memory network is combined with the seasonal autoregressive integrated moving average model to conduct an in-depth analysis of the comprehensive data set, predict the power load demand trend of the square cabin hospital under different environmental conditions, and apply the multi-objective genetic algorithm to optimize the power load forecasting model to generate an accurate power load forecasting report, including: Cleaning and normalizing the comprehensive data set, removing outliers and missing values, ensuring data quality, and obtaining a preprocessed data set; Performing feature extraction and selection on the preprocessed data set to construct an input vector containing valuable features of environmental parameters and medical device working modes; Based on the input vector, a hybrid prediction model of a long short-term memory network combined with a seasonal autoregressive integrated moving average model is constructed, and the hybrid prediction model is trained to obtain a preliminary power load prediction model; Utilizing a multi-objective genetic algorithm, the preliminary power load forecasting model is optimized to find the best combination of model parameters and generate an optimized power load forecasting model; Based on the optimized power load forecasting model, the power load demand of the square cabin hospital under different environmental conditions is predicted and processed to obtain an accurate power load forecasting report.

3. The method according to claim 2, characterized in that The method of optimizing the preliminary power load forecasting model by using a multi-objective genetic algorithm, finding the best combination of model parameters, and generating an optimized power load forecasting model includes: In order to calculate the objective function F1(w), the generalization ability of the model is first evaluated through cross-validation, and then the key parameters affecting the model performance are determined using sensitivity analysis, and finally the initial parameters of the multi-objective genetic algorithm are set; Objective function: Where w represents the model parameter vector; y i is the actual power load value of the i-th sample; is the predicted power load value of the i-th sample; N is the total number of samples; α is a weight coefficient used to balance the contribution of the root mean square error and the mean absolute error in the objective function, and its value range is between 0 and 1; γ is a weight coefficient used to balance the contribution of the mean absolute percentage error in the objective function, and its value range is between 0 and 1; After completing the calculation of F1(w), the constraint condition F2(w) is constructed by calculating the coefficient of determination, F score and variance of the predicted value of the model to ensure that the optimization process considers both prediction accuracy and model stability; Constraints: Among them, R 2 (w) is the coefficient of determination, which measures the proportion of the variability of the model explaining the variables; max(R 2 ) is R 2 The maximum possible value, usually 1; F score (w) is the F score, which measures the overall performance of the model and takes into account the precision and recall of the model; max(F score ) is the maximum possible value of the F score, usually 1; Variance(w) is the variance of the model prediction value, which measures the stability of the model prediction; max(Variance) is the maximum possible value of the variance; β is the weight coefficient used to balance R 2 The contribution of the F score to the constraints ranges from 0 to 1. δ is the weight coefficient used to balance the contribution of the variance to the constraints, and ranges from 0 to 1. These two formulas work together in the multi-objective genetic algorithm to iteratively optimize the model parameter w to minimize the objective function F1 and the constraint condition F2, thereby generating an optimized power load forecasting model.

4. The method according to claim 2, characterized in that: The extracting and selecting features of the preprocessed data set to construct an input vector containing valuable features of environmental parameters and medical device working modes includes: Using statistical analysis methods, performing correlation analysis on various indicators in the preprocessed data set, identifying features closely related to power load, and obtaining a list of features with high correlation; According to the list of features with high correlation, features with high correlation are selected as components of the input vector to generate a selected feature set; Encoding the selected feature set, converting categorical features into numerical data by using one-hot encoding or label encoding, and standardizing or normalizing continuous features according to actual needs to obtain a standardized feature set; Based on the standardized feature set, a final input vector is constructed to obtain an accurate power load forecast report.

5. The method according to claim 2, characterized in that: Based on the input vector, a hybrid prediction model of a long short-term memory network combined with a seasonal autoregressive integrated moving average model is constructed, and the hybrid prediction model is trained to obtain a preliminary power load prediction model, including: Using the input vector, a long short-term memory network structure is designed to construct an LSTM model part capable of capturing long-term dependencies in time series data, thereby obtaining an LSTM model structure; According to the time characteristics of the input vector, a suitable seasonal period is selected, and a seasonal autoregressive integrated moving average model part is constructed to obtain a SARIMA model structure; Based on the LSTM model structure and the SARIMA model structure, the architecture of the hybrid prediction model is designed to ensure that the two model parts can be effectively combined to generate the hybrid prediction model architecture; The preprocessed data set and the standardized feature set are used to train the hybrid prediction model architecture, and a preliminary power load prediction model is generated by iteratively optimizing model parameters.

6. The method according to claim 1, characterized in that The power distribution plan is formulated based on the power load forecast report using mixed integer linear programming technology, fuzzy logic control technology is introduced to evaluate the flexibility of the power system, and the start and stop timing of non-critical loads is dynamically adjusted to cope with uncertain power supply fluctuations, generating the optimal power distribution and scheduling plan, including: According to the power load forecast report, analyze the future power demand trend, determine the basic framework of power distribution, and obtain the power demand analysis results; Using the power demand analysis results, adopting mixed integer linear programming technology, combined with the existing power supply capacity of the Fangcang Hospital and the priorities of various loads, the power distribution is modeled and processed to generate a preliminary power distribution plan; Introducing fuzzy logic control technology, based on the preliminary power distribution plan, evaluating the flexibility of the power system, and generating a power system flexibility evaluation report; According to the power system flexibility assessment report, dynamically adjust the start and stop timing of non-critical loads, determine the best start and stop strategy through an optimization algorithm, and generate an optimized start and stop strategy; Based on the preliminary power distribution plan and the optimized start-stop strategy, the accuracy of power load forecasting, the economy of power distribution and the flexibility of the power system are comprehensively considered to generate the optimal power distribution and scheduling plan.

7. The method according to claim 6, characterized in that According to the power system flexibility assessment report, dynamically adjusting the start and stop timing of non-critical loads, determining the best start and stop strategy through an optimization algorithm, and generating an optimized start and stop strategy, including: When calculating the startup time T of non-critical loads on (i) Previously, it was necessary to analyze the data from the power system flexibility assessment report, identify the characteristics of non-critical loads, and determine the benchmark start time and allocation weights; Among them, T on (i) represents the startup time of the i-th non-critical load; t0 is the benchmark startup time, which is usually the starting time of the low power demand period; N is the number of non-critical loads; w j is the weight of the jth non-critical load, reflecting its importance in power distribution; P j is the power demand of the jth non-critical load; P min It is the smallest power requirement among all non-critical loads; F j is the flexibility score of the jth non-critical load; F max The highest flexibility score among all non-critical loads; D j is the startup delay time of the jth non-critical load; λ is the influence coefficient of the delay time; After calculating T on (i), the impact of different start-stop schedules is evaluated, and the optimal start-stop strategy S(i) is found using an optimization algorithm; Formula 2: Optimization algorithm determines the best start-stop strategy Where S(i) represents the optimal start-stop strategy for the i-th non-critical load; s i is the start and stop schedule of the i-th non-critical load; C(s i ) is the start and stop schedule i The total cost under the condition includes the startup cost and the running cost; V(s i ) is the start and stop schedule i Power system volatility under i ) is the start and stop schedule i α is the cost weight coefficient, ranging from 0 to 1; β is the load balancing weight coefficient, ranging from 0 to 1; γ is the balance coefficient between volatility and load balancing; R(s i ) is the start and stop schedule i Resource utilization under Based on S(i), a detailed start-stop schedule is generated, its effect is verified through simulation or testing, and the optimization is adjusted according to the results to form the final start-stop strategy.

8. The method according to claim 6, characterized in that The power demand analysis results are used to model the power distribution by adopting mixed integer linear programming technology, combined with the existing power supply capacity of the Fangcang Hospital and the priorities of various loads, to generate a preliminary power distribution plan, including: Using the power demand analysis results, determine the total power demand and its distribution characteristics in a specific future time period, and generate a power demand distribution map; According to the power demand distribution map and in combination with the existing power supply capacity of the Fangcang Cabin Hospital, a detailed assessment of the power supply capacity is conducted to generate a power supply capacity assessment report; Based on the power supply capacity assessment report, determining the priority of each type of load and generating a load priority list; Using mixed integer linear programming technology, a mathematical model of power distribution is established according to the load priority list and the power demand distribution map to generate a power distribution model; The power distribution model is solved, and the optimal power distribution scheme is found by setting the objective function and constraint conditions, so as to generate a preliminary power distribution plan.

9. The method according to claim 1, characterized in that: Based on the optimal power distribution and dispatching scheme, system performance evaluation is performed regularly, algorithm parameters are adjusted according to the evaluation results, and an intelligent management scheme for the square cabin hospital power system is generated, including: Using a predetermined time period, data is collected from the power system of the shelter hospital, including power consumption, actual power consumption of each load, and the working status of power equipment, to obtain actual operating data; Based on the actual operation data, compare the difference between the actual power distribution and dispatching situation and the optimal power distribution and dispatching plan, evaluate the performance indicators of the power system, and generate a system performance evaluation report; Based on the system performance evaluation report, in-depth analysis of existing problems and potential areas for improvement is performed to generate a problem analysis report; Using the problem analysis report, appropriately adjusting algorithm parameters for generating an optimal power distribution and dispatching solution to generate adjusted algorithm parameters; Based on the adjusted algorithm parameters, the new power distribution and dispatching scheme is tested in a simulation environment to generate a test verification report; Using the test verification report, apply the tested and verified new solution to the actual power system of the Fangcang Cabin Hospital, continuously monitor the operating status of the system, collect data on the implementation effect, and generate implementation monitoring data; Based on the implementation monitoring data, a complete intelligent management plan for the square cabin hospital power system is formed and an intelligent management plan is generated.

10. An intelligent square cabin hospital power load forecasting and dispatching system, characterized in that: include: The collection module is used to collect and integrate the environmental parameters inside and outside the Fangcang shelter hospital and the operating status information of medical equipment in real time, collect historical power consumption data and special power demand records, and generate a comprehensive data set containing rich background information; An analysis module, for performing in-depth analysis on the comprehensive data set using a long short-term memory network combined with a seasonal autoregressive integrated moving average model, predicting the power load demand trend of the shelter hospital under different environmental conditions, optimizing the power load forecasting model using a multi-objective genetic algorithm, and generating an accurate power load forecasting report; An adjustment module is used to formulate a power distribution plan based on the power load forecast report using mixed integer linear programming technology, introduce fuzzy logic control technology to evaluate the flexibility of the power system, dynamically adjust the start and stop timing of non-critical loads to cope with uncertain power supply fluctuations, and generate an optimal power distribution and scheduling plan; The evaluation module is used to regularly evaluate system performance based on the optimal power distribution and scheduling plan, adjust algorithm parameters according to the evaluation results, and generate an intelligent management plan for the square cabin hospital power system.

Images