Ventilation strategy optimization analog simulation training system
Through ventilation strategy optimization, simulation training system, and using technologies such as working condition matching and trend prediction, ventilation strategies are automatically optimized, which solves the problems of low efficiency and safety risks of ventilation system caused by artificial experience dependence, and achieves more efficient and safe ventilation management .
Patent Information
- Application Number
- CN202411411912.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-10
- Publication Date
- 2025-01-10
AI Technical Summary
The existing ventilation strategy management model relies on manual experience, resulting in low efficiency of ventilation systems, inability to achieve optimal results, safety risks, and difficulty in establishing a long-term, stable and sustainable management system.
A ventilation strategy optimization simulation training system was designed, including ventilation strategy optimization unit, ventilation effect knowledge base and business system interface service module. Through operating condition matching, trend prediction, single ventilation effect evaluation, fit model training and heuristics Modules such as evolution automatically identify environmental changes and match optimized ventilation strategies to reduce manual intervention.
It improves the accuracy and energy consumption efficiency of ventilation strategies, reduces safety risks, ensures the timeliness and forward-looking of ventilation systems, reduces personnel input costs, and avoids fluctuations in the production environment.
Smart Images
Figure CN119294095A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of ventilation strategy optimization, and specifically relates to a ventilation strategy optimization simulation training system. Background Art
[0002] The purpose of ventilation strategy optimization simulation training is to find ways to improve ventilation efficiency, reduce energy consumption, optimize air volume distribution and improve air quality by simulating and analyzing the behavior of the ventilation system. This kind of training is crucial to ensure the safe production of underground spaces such as mines, because a reasonable ventilation system can provide enough fresh air, maintain appropriate temperature and humidity, prevent the accumulation of harmful gases, and effectively control wind direction and air volume in the event of a disaster to limit the spread of the accident.
[0003] At present, the ventilation strategy management model usually relies on manual experience to operate and maintain the ventilation system. Due to the lack of systematic and scientific management methods, extensive management often leads to low efficiency of the ventilation system and failure to achieve the optimal ventilation effect. There are also certain safety risks, which in turn leads to the lack of rationality in the installation of ventilation devices, making it difficult to play their due role. In addition, long-term reliance on manual experience management is not conducive to the establishment of a long-term, stable and sustainable ventilation management system. Summary of the invention
[0004] In order to overcome the above-mentioned defects, the present invention provides a ventilation strategy optimization simulation training system, which solves the problem that the ventilation strategy management mode usually relies on manual experience to operate and maintain the ventilation system. Due to the lack of systematic and scientific management methods, extensive management often leads to low efficiency of the ventilation system and failure to achieve the optimal ventilation effect. There are also certain safety risks, which leads to the lack of rationality in the installation of ventilation devices and long-term reliance on manual experience management.
[0005] To achieve the above-mentioned object, the present invention provides the following technical solutions: a ventilation strategy optimization simulation training system, comprising a ventilation strategy optimization unit, a ventilation effect knowledge base and a business system interface service module, wherein the ventilation strategy optimization unit is bidirectionally connected to the ventilation effect knowledge base and the business system interface service module by signals, and the business system interface service module is bidirectionally connected to the ventilation effect knowledge base;
[0006] The ventilation strategy optimization unit includes a working condition matching module, an evaluation and recommendation module, a trend prediction model training module, a trend prediction calculation module, a single ventilation effect evaluation module, a ventilation effect fitting model training module, a ventilation effect fitting prediction module, a priori evolution rule model, and a heuristic evolution module;
[0007] Working condition matching module: used to identify the internal and external environment of the current ventilation system operation, and call the algorithm to match multiple ventilation strategies from the historical single ventilation effect evaluation results;
[0008] Evaluation and recommendation module: Based on the operating effects of multiple ventilation strategies provided by the working condition matching module and the evolution strategies and effects output by the ventilation strategy evolution module;
[0009] Trend prediction model training module: build trend prediction models based on real-time environmental monitoring data;
[0010] Trend prediction calculation module: uses the trained trend prediction model to make real-time trend predictions on indoor temperature and humidity indicators based on real-time environmental monitoring data, and is used to predict temperature and humidity changes in the future.
[0011] Single ventilation effect evaluation module: used to calculate the index reflecting the ventilation effect after each ventilation, and store the evaluation results in the ventilation effect knowledge base;
[0012] Ventilation effect fitting model training module: Based on the historical single ventilation effect evaluation result data, historical internal and external temperature and humidity start and stop data, and historical ventilation strategy data, the RBF neural network algorithm is used to train the ventilation effect fitting model;
[0013] Ventilation effect fitting prediction module: by inputting new independent variables, the ventilation effect fitting model gives the ventilation effect fitting results, and the ventilation effect fitting model has the function of regular update;
[0014] Prior evolution rule model: by inputting the system number, the temperature and humidity indicators inside and outside the system and their change gradients, the ventilation strategy and the operation results of the ventilation strategy, the prior evolution rule is called, the number of equipment turned on and the opening gear are adjusted according to the operation effect of the current ventilation strategy, and the adjusted ventilation strategy is output;
[0015] Heuristic evolution module: According to the current strategy, the heuristic evolution algorithm is used to evolve the current strategy through multiple rounds of mutation, crossover and evaluation. The evolution algorithm is used to randomly, mutate and cross-expand the initial strategy to form multiple new strategies, which are then substituted into the RBF ventilation effect fitting model to fit the ventilation effect results.
[0016] As a further solution of the present invention: the internal and external environment includes internal temperature, internal humidity, external temperature, and external humidity. The working condition matching module matches multiple ventilation strategies and corresponding historical operating effects of similar working conditions from the ventilation effect knowledge base according to the similarity between the current working condition and the historical working condition, and provides them to the evaluation and recommendation module;
[0017] The specific implementation process of the working condition matching module is as follows:
[0018] When a ventilation strategy recommendation request is received, the working condition matching module is triggered to obtain the current real-time working condition data and environmental monitoring data, followed by access to the single ventilation effect evaluation table data;
[0019] Clean the data, check whether there are null values or abnormal values in the acquired data, process the abnormalities before calling the algorithm, replace the missing values with the mean of ventilation data, replace the time series data with the mean before and after the missing values, and process the abnormal values with outliers, and replace the outliers with the n-sigma rule;
[0020] Filter out the key fields of real-time working condition data and ventilation effect knowledge base data, and standardize the field data to eliminate the impact of different dimensions on ventilation strategy matching;
[0021] Based on the standardized data, the similarity between the current working condition and the historical ventilation strategy data is calculated through a similarity matching algorithm, and the historical ventilation strategies are sorted from high to low according to their corresponding similarities, and the top n ventilation strategies with the highest similarity ranking are extracted;
[0022] According to the set similarity threshold, determine whether n strategies are ventilation strategies that meet the similarity threshold, eliminate ventilation strategies below the similarity threshold, and output the similarity matching results of the working conditions. If the similarity threshold is not reached, no similar strategy is matched this time, and wait for historical strategy accumulation to be evaluated again.
[0023] As a further solution of the present invention: the evaluation and recommendation module scores the ventilation strategies from multiple perspectives through comprehensive evaluation or linear weighting and outputs the comprehensive scores of the strategies, and gives the best recommended strategy according to the similarity of the working conditions and the calculated comprehensive scores;
[0024] The evaluation process of the evaluation and recommendation module is as follows:
[0025] Forward the original data matrix, transform the data corresponding to those extremely small indicators into extremely large indicators, and calculate and process them uniformly;
[0026] x ′ =Mx
[0027] M is the maximum possible value of the indicator x;
[0028] Normalize the positive / negative matrix, eliminate the influence of dimension by normalization, and construct the normalized initial matrix;
[0029] There are n objects to be evaluated, each of which has m indicators or attributes. The original data matrix is constructed as follows:
[0030]
[0031] Construct a weighted normalized matrix, and perform vector normalization on the attributes, that is, subtract the mean from each column element and divide it by the norm of the current column vector, and use the cosine distance metric to obtain the normalized matrix Z:
[0032]
[0033] Calculate the distance between each solution and the optimal solution and the worst solution;
[0034] Optimal Solution Z + It is composed of the maximum value of each column element in Z:
[0035]
[0036] The worst solution Z - It is composed of the minimum value of each column element in Z:
[0037]
[0038] Calculate the degree of proximity between each evaluation object and the optimal solution and the worst solution:
[0039]
[0040] where w j is the weight of the jth attribute, and the weight represents the importance. The indicator weight is determined according to the actual situation. The degree of closeness between each evaluation object and the optimal solution is calculated. i 0≤C_i≤1,C_i→1 indicates that the evaluation object is better:
[0041]
[0042] According to C i The strategies are evaluated by sorting from large to small, and the better ventilation strategy is selected for recommendation after strategy matching.
[0043] As a further solution of the present invention: the trend prediction model training module uses a regression prediction algorithm to construct a regression prediction model of the target indicator based on the actual values and change trends of the external environment temperature and humidity in a period of time before and after the start and stop of the historical ventilation strategy;
[0044] The specific implementation process of the trend prediction calculation module is as follows:
[0045] When the business system interface service module clicks a measurement point outside the security room on the interface, it clicks the trend prediction button in the pop-up dialog box to call the trend prediction interface. Based on the input indicator value of the previous period and the number of points to be predicted, it predicts the change of the indicator in the future period, and returns the result to the business system interface service module in the form of JSON;
[0046] Click the security room location measurement point on the interface, click the trend prediction button in the pop-up dialog box, call the trend prediction interface according to the selected prediction index and pass in the security room and index to predict the changes of the security room location measurement point in the future, and return the results to the business system interface service module in the form of JSON.
[0047] As a further solution of the present invention: the indicators include rapidity, stability, accuracy, timeliness and energy consumption, the single ventilation effect evaluation module stores the indicators and the relevant environmental parameters and temperature and humidity change gradient parameters during operation into the ventilation effect knowledge base, the single ventilation effect evaluation module calculates the single ventilation strategy evaluation indicators according to the internal environment temperature and humidity change data and the temperature and humidity set values from the time when the historical ventilation strategy is turned on to the time when the next ventilation strategy is turned on, and obtains the single ventilation effect evaluation model;
[0048] The specific implementation process of the single ventilation effect evaluation module is as follows:
[0049] Read the ventilation strategy start-stop table regularly. When a new ventilation strategy is detected in the ventilation strategy start-stop table, the ventilation effect index calculation is triggered;
[0050] According to the last ventilation mode start-up and end time, read the internal and external environmental data of the ventilation system to obtain the single ventilation data corresponding to the ventilation strategy and the environmental monitoring data for a period of time before the ventilation strategy is started;
[0051] Calculate the change gradient of each indicator in the period before the ventilation strategy is turned on, and record the initial and final recorded values of each indicator when the ventilation strategy is turned on, as the initial value of the environment when a single ventilation strategy is turned on or off;
[0052] Secondly, the target values of various indicators of this ventilation are obtained through the target value change schedule, and the indicators of rapidity, maintenance duration, accuracy, energy consumption and stability of this ventilation are calculated respectively. The initial recorded values of each indicator when the ventilation strategy is turned on and the ventilation effect data are integrated to obtain the evaluation results of a single ventilation effect.
[0053] As a further solution of the present invention: the ventilation effect fitting model training module performs regular scheduling according to the set scheduling cycle, obtains the ventilation system configuration table, and the evaluation result data of the single ventilation effect evaluation table, and for different systems, takes the evaluation result index as the dependent variable, and takes the historical internal and external temperature and humidity indicators and ventilation strategies as independent variables to train the single ventilation effect evaluation model.
[0054] As a further solution of the present invention: the ventilation effect fitting prediction module predicts the temperature rapidity, humidity rapidity, temperature accuracy, humidity accuracy, energy consumption, temperature gradient, humidity gradient, and timeliness indicators under the current environmental conditions and when the equipment is turned on, through the system number, internal and external temperature and humidity indicators and their change gradients, and system start-up gear indicators input by the business system interface service module.
[0055] As a further solution of the present invention: the prior evolution rule model analyzes the results of a single ventilation effect evaluation, and uses the prior evolution rule to evolve the ventilation strategy based on the current ventilation strategy. The prior evolution rule is based on the principle of adjustment in a better direction. The direction of evolution includes adjustment of working conditions, adjustment of quantity, and adjustment of duration. A new ventilation strategy is generated through the heuristic evolution module and provided to the business system interface service module.
[0056] As a further solution of the present invention: the specific implementation process of the heuristic evolution module is as follows:
[0057] Multi-objective optimization:
[0058] The ventilation strategy optimization problem is defined as follows:
[0059] min f(x,y,a)=-g(rbf(x,y,a))
[0060]
[0061] x∈Ω
[0062] a∈A
[0063] in:
[0064] x is the equipment gear, y is the environmental parameter, and a is the corresponding ventilation system;
[0065] rbf(x,y,a) is the ventilation effect prediction result;
[0066] A is the collection of all ventilation systems, Ω is all adjustable equipment;
[0067] x i is the i-th equipment gear to be optimized;
[0068] and The lower and upper limits of the equipment position;
[0069] f is the optimization target and the strategy score of the ventilation strategy effect prediction result;
[0070] The objective function f satisfies all equality and inequality constraints. If a particular objective function is maximized (maxf i), and redefine the problem to minimize its negative value (min-f i );
[0071] Genetic algorithms are used for calculation. During the ventilation optimization process, ventilation strategies are genetically selected. The scores of offspring strategies are higher than those of previous strategies. Ultimately, a strategy that performs best in the evolutionary direction will be selected.
[0072] Compared with the prior art, the beneficial effects of the present invention are: by adopting a ventilation effect knowledge base, a business system interface service module and a ventilation strategy optimization unit to establish a simulation training system, through the operation of the intelligent ventilation strategy optimization system, it can automatically identify changes in the production environment, and match corresponding ventilation strategies according to environmental changes, thereby reducing personnel input costs and incomplete personnel judgment factors; it can evolve the ventilation strategy on the basis of the original ventilation strategy, improve the accuracy of the ventilation strategy, and reduce energy consumption costs; at the same time, it ensures the timeliness and foresight of ventilation strategy switching, avoids large fluctuations in the production environment, and affects normal production. BRIEF DESCRIPTION OF THE DRAWINGS
[0073] Figure 1 A schematic block diagram of the system of the present invention. DETAILED DESCRIPTION
[0074] The technical solution of the present application is further described in detail below in conjunction with specific implementation methods.
[0075] As shown in the figure, the present invention provides a technical solution: a ventilation strategy optimization simulation training system, including a ventilation strategy optimization unit, a ventilation effect knowledge base and a business system interface service module, the ventilation strategy optimization unit is bidirectionally connected with the ventilation effect knowledge base and the business system interface service module, and the business system interface service module is bidirectionally connected with the ventilation effect knowledge base;
[0076] The ventilation strategy optimization unit includes a working condition matching module, an evaluation and recommendation module, a trend prediction model training module, a trend prediction calculation module, a single ventilation effect evaluation module, a ventilation effect fitting model training module, a ventilation effect fitting prediction module, a priori evolution rule model, and a heuristic evolution module;
[0077] Working condition matching module: used to identify the internal and external environment of the current ventilation system operation, and call the algorithm to match multiple ventilation strategies from the historical single ventilation effect evaluation results;
[0078] The internal and external environment includes internal temperature, internal humidity, external temperature, and external humidity. The working condition matching module matches multiple ventilation strategies and corresponding historical operating effects of similar working conditions from the ventilation effect knowledge base according to the similarity between the current working condition and the historical working condition, and provides them to the evaluation and recommendation module;
[0079] The specific implementation process of the working condition matching module is as follows:
[0080] When a ventilation strategy recommendation request is received, the working condition matching module is triggered to obtain the current real-time working condition data and environmental monitoring data, followed by access to the single ventilation effect evaluation table data;
[0081] Clean the data, check whether there are null values or abnormal values in the acquired data, process the abnormalities before calling the algorithm, replace the missing values with the mean of ventilation data, replace the time series data with the mean before and after the missing values, and process the abnormal values with outliers, and replace the outliers with the n-sigma rule;
[0082] Filter out the key fields of real-time working condition data and ventilation effect knowledge base data, and standardize the field data to eliminate the impact of different dimensions on ventilation strategy matching;
[0083] Based on the standardized data, the similarity between the current working condition and the historical ventilation strategy data is calculated through a similarity matching algorithm, and the historical ventilation strategies are sorted from high to low according to their corresponding similarities, and the top n ventilation strategies with the highest similarity ranking are extracted;
[0084] According to the set similarity threshold, determine whether n strategies are ventilation strategies that meet the similarity threshold, eliminate ventilation strategies below the similarity threshold, and output the similarity matching results of the working conditions. If the similarity threshold is not reached, no similar strategy is matched this time, and wait for historical strategy accumulation to be evaluated again.
[0085] Evaluation and recommendation module: Based on the operating effects of multiple ventilation strategies provided by the working condition matching module and the evolution strategies and effects output by the ventilation strategy evolution module;
[0086] The evaluation and recommendation module scores the ventilation strategies from multiple perspectives through comprehensive evaluation or linear weighting and outputs the comprehensive scores of the strategies, and gives the best recommended strategy according to the similarity of the working conditions and the calculated comprehensive scores;
[0087] The evaluation process of the evaluation and recommendation module is as follows:
[0088] Forward the original data matrix, transform the data corresponding to those extremely small indicators into extremely large indicators, and calculate and process them uniformly;
[0089] x ′ =Mx
[0090] M is the maximum possible value of the indicator x;
[0091] Normalize the positive / negative matrix, eliminate the influence of dimension by normalization, and construct the normalized initial matrix;
[0092] There are n objects to be evaluated, each of which has m indicators or attributes. The original data matrix is constructed as follows:
[0093]
[0094] Construct a weighted normalized matrix, and perform vector normalization on the attributes, that is, subtract the mean from each column element and divide it by the norm of the current column vector, and use the cosine distance metric to obtain the normalized matrix Z:
[0095]
[0096] Calculate the distance between each solution and the optimal solution and the worst solution;
[0097] Optimal Solution Z + It is composed of the maximum value of each column element in Z:
[0098]
[0099] The worst solution Z - It is composed of the minimum value of each column element in Z:
[0100]
[0101] Calculate the degree of proximity between each evaluation object and the optimal solution and the worst solution:
[0102]
[0103] where w j is the weight of the jth attribute, and the weight represents the importance. The indicator weight is determined according to the actual situation. The degree of closeness between each evaluation object and the optimal solution is calculated. i 0≤C_i≤1,C_i→1 indicates that the evaluation object is better:
[0104]
[0105] According to C i The strategies are evaluated by sorting from large to small, and the better ventilation strategy is selected for recommendation after strategy matching.
[0106] Trend prediction model training module: build trend prediction models based on real-time environmental monitoring data;
[0107] The trend prediction model training module uses a regression prediction algorithm to build a regression prediction model for the target indicator based on the actual values and change trends of the external environment temperature and humidity in a period of time before and after the historical ventilation strategy is started and stopped.
[0108] Trend prediction calculation module: uses the trained trend prediction model to make real-time trend predictions on indoor temperature and humidity indicators based on real-time environmental monitoring data, and is used to predict temperature and humidity changes in the future.
[0109] The specific implementation process of the trend prediction calculation module is as follows:
[0110] When the business system interface service module clicks a measurement point outside the security room on the interface, it clicks the trend prediction button in the pop-up dialog box to call the trend prediction interface. Based on the input indicator value of the previous period and the number of points to be predicted, it predicts the change of the indicator in the future period, and returns the result to the business system interface service module in the form of JSON;
[0111] Click the security room location measurement point on the interface, click the trend prediction button in the pop-up dialog box, call the trend prediction interface according to the selected prediction index and pass in the security room and index to predict the changes of the security room location measurement point in the future, and return the results to the business system interface service module in the form of JSON.
[0112] Single ventilation effect evaluation module: used to calculate the index reflecting the ventilation effect after each ventilation, and store the evaluation results in the ventilation effect knowledge base;
[0113] The indicators include rapidity, stability, accuracy, timeliness and energy consumption. The single ventilation effect evaluation module stores the indicators and the relevant environmental parameters and temperature and humidity change gradient parameters during operation into the ventilation effect knowledge base. The single ventilation effect evaluation module calculates the single ventilation strategy evaluation indicators according to the internal environment temperature and humidity change data and the temperature and humidity setting values from the time when the historical ventilation strategy is turned on to the time when the next ventilation strategy is turned on, and obtains a single ventilation effect evaluation model.
[0114] The specific implementation process of the single ventilation effect evaluation module is as follows:
[0115] Read the ventilation strategy start-stop table regularly. When a new ventilation strategy is detected in the ventilation strategy start-stop table, the ventilation effect index calculation is triggered;
[0116] According to the last ventilation mode start-up and end time, read the internal and external environmental data of the ventilation system to obtain the single ventilation data corresponding to the ventilation strategy and the environmental monitoring data for a period of time before the ventilation strategy is started;
[0117] Calculate the gradient of each indicator change in the period before the ventilation strategy is turned on, and record the initial and final recorded values of each indicator when the ventilation strategy is turned on, as the initial value of the environment when a single ventilation strategy is turned on or off;
[0118] Secondly, the target values of various indicators of this ventilation are obtained through the target value change schedule, and the indicators of rapidity, maintenance duration, accuracy, energy consumption and stability of this ventilation are calculated respectively. The initial recorded values of each indicator when the ventilation strategy is turned on and the ventilation effect data are integrated to obtain the evaluation results of a single ventilation effect.
[0119] Ventilation effect fitting model training module: Based on the historical single ventilation effect evaluation result data, historical internal and external temperature and humidity start and stop data, and historical ventilation strategy data, the RBF neural network algorithm is used to train the ventilation effect fitting model;
[0120] The ventilation effect fitting model training module performs regular scheduling according to the set scheduling cycle, obtains the ventilation system configuration table and the evaluation result data of the single ventilation effect evaluation table, and trains the single ventilation effect evaluation model for different systems with the evaluation result index as the dependent variable and the historical internal and external temperature and humidity indicators and ventilation strategies as the independent variables.
[0121] Ventilation effect fitting prediction module: by inputting new independent variables, the ventilation effect fitting model gives the ventilation effect fitting results, and the ventilation effect fitting model has the function of regular update;
[0122] The ventilation effect fitting prediction module predicts the temperature rapidity, humidity rapidity, temperature accuracy, humidity accuracy, energy consumption, temperature gradient, humidity gradient, and timeliness indicators under the current environmental conditions and when the equipment is turned on, based on the system number, internal and external temperature and humidity indicators and their change gradients, and system start-up gear indicators input by the business system interface service module.
[0123] Prior evolution rule model: by inputting the system number, the temperature and humidity indicators inside and outside the system and their change gradients, the ventilation strategy and the operation results of the ventilation strategy, the prior evolution rule is called, the number of equipment turned on and the opening gear are adjusted according to the operation effect of the current ventilation strategy, and the adjusted ventilation strategy is output;
[0124] The prior evolution rule model analyzes the results of a single ventilation effect evaluation, and uses the prior evolution rule to evolve the ventilation strategy based on the current ventilation strategy. The prior evolution rule is based on the principle of adjustment in a better direction. The direction of evolution includes adjustment of working conditions, quantity, and duration. A new ventilation strategy is generated through the heuristic evolution module and provided to the business system interface service module.
[0125] Heuristic evolution module: Based on the current strategy, the heuristic evolution algorithm is used to evolve the current strategy in the form of multiple rounds of mutation, crossover, and evaluation. The evolution algorithm is used to randomly, mutate, and cross-expand the initial strategy to form multiple new strategies, which are then substituted into the RBF ventilation effect fitting model to fit the ventilation effect results;
[0126] The specific implementation process of the heuristic evolution module is as follows:
[0127] Multi-objective optimization:
[0128] The ventilation strategy optimization problem is defined as follows:
[0129] min f(x,y,a)=-g(rbf(x,y,a))
[0130]
[0131] x∈Ω
[0132] a∈A
[0133] in:
[0134] x is the equipment gear, y is the environmental parameter, and a is the corresponding ventilation system;
[0135] rbf(x,y,a) is the ventilation effect prediction result;
[0136] A is the collection of all ventilation systems, Ω is all adjustable equipment;
[0137] x i is the i-th equipment gear to be optimized;
[0138] and The lower and upper limits of the equipment position;
[0139] f is the optimization target and the strategy score of the ventilation strategy effect prediction result;
[0140] The objective function f satisfies all equality and inequality constraints. If a particular objective function is maximized (maxf i ), and redefine the problem to minimize its negative value (min-f i );
[0141] Genetic algorithms are used for calculation. During the ventilation optimization process, ventilation strategies are genetically selected. The scores of offspring strategies are higher than those of previous strategies. Ultimately, a strategy that performs best in the evolutionary direction will be selected.
[0142] From the above we know:
[0143] By using the ventilation effect knowledge base, business system interface service module and ventilation strategy optimization unit to establish a simulation training system, the operation of the system can be optimized through intelligent ventilation strategy, which can automatically identify changes in the production environment and match the corresponding ventilation strategy according to the environmental changes, thus reducing the cost of personnel input and the situation where personnel make incomplete judgments.
[0144] Ability to evolve ventilation strategies based on the original ones, improve the accuracy of ventilation strategies, and reduce energy consumption costs;
[0145] At the same time, ensure the timeliness and foresight of ventilation strategy switching to avoid large fluctuations in the production environment and affect normal production.
[0146] In the description of the present invention, it is necessary to understand that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.
[0147] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.
[0148] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, an electrical connection, or a communication; it can be a direct connection, or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0149] In the present invention, unless otherwise clearly specified and limited, the first feature "on" or "under" the second feature may be that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. In the description of this specification, the description with reference to the terms "one scheme", "some schemes", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the scheme or example are included in at least one scheme or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same scheme or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more schemes or examples in a suitable manner.
Claims
1. A ventilation strategy optimization simulation training system, comprising a ventilation strategy optimization unit, a ventilation effect knowledge base and a business system interface service module, characterized in that: The ventilation strategy optimization unit is bidirectionally connected to the ventilation effect knowledge base and the business system interface service module by signal, and the business system interface service module is bidirectionally connected to the ventilation effect knowledge base; The ventilation strategy optimization unit includes a working condition matching module, an evaluation and recommendation module, a trend prediction model training module, a trend prediction calculation module, a single ventilation effect evaluation module, a ventilation effect fitting model training module, a ventilation effect fitting prediction module, a priori evolution rule model, and a heuristic evolution module; Working condition matching module: used to identify the internal and external environment of the current ventilation system operation, and call the algorithm to match multiple ventilation strategies from the historical single ventilation effect evaluation results; Evaluation and recommendation module: Based on the operating effects of multiple ventilation strategies provided by the working condition matching module and the evolution strategies and effects output by the ventilation strategy evolution module; Trend prediction model training module: build trend prediction models based on real-time environmental monitoring data; Trend prediction calculation module: uses the trained trend prediction model to make real-time trend predictions on indoor temperature and humidity indicators based on real-time environmental monitoring data, and is used to predict temperature and humidity changes in the future. Single ventilation effect evaluation module: used to calculate the index reflecting the ventilation effect after each ventilation, and store the evaluation results in the ventilation effect knowledge base; Ventilation effect fitting model training module: Based on the historical single ventilation effect evaluation result data, historical internal and external temperature and humidity start and stop data, and historical ventilation strategy data, the RBF neural network algorithm is used to train the ventilation effect fitting model; Ventilation effect fitting prediction module: by inputting new independent variables, the ventilation effect fitting model gives the ventilation effect fitting results, and the ventilation effect fitting model has the function of regular update; Prior evolution rule model: by inputting the system number, the temperature and humidity indicators inside and outside the system and their change gradients, the ventilation strategy and the operation results of the ventilation strategy, the prior evolution rule is called, the number of equipment turned on and the opening gear are adjusted according to the operation effect of the current ventilation strategy, and the adjusted ventilation strategy is output; Heuristic evolution module: According to the current strategy, the heuristic evolution algorithm is used to evolve the current strategy through multiple rounds of mutation, crossover and evaluation. The evolution algorithm is used to randomly, mutate and cross-expand the initial strategy to form multiple new strategies, which are then substituted into the RBF ventilation effect fitting model to fit the ventilation effect results.
2. A ventilation strategy optimization simulation training system according to claim 1, characterized in that: The internal and external environment includes internal temperature, internal humidity, external temperature, and external humidity. The working condition matching module matches multiple ventilation strategies and corresponding historical operating effects of similar working conditions from the ventilation effect knowledge base according to the similarity between the current working condition and the historical working condition, and provides them to the evaluation and recommendation module; The specific implementation process of the working condition matching module is as follows: When a ventilation strategy recommendation request is received, the working condition matching module is triggered to obtain the current real-time working condition data and environmental monitoring data, followed by access to the single ventilation effect evaluation table data; Clean the data, check whether there are null values or abnormal values in the acquired data, process the abnormalities before calling the algorithm, replace the missing values with the mean of ventilation data, replace the time series data with the mean before and after the missing values, and process the abnormal values with outliers, and replace the outliers with the n-sigma rule; Filter out the key fields of real-time working condition data and ventilation effect knowledge base data, and standardize the field data to eliminate the impact of different dimensions on ventilation strategy matching; Based on the standardized data, the similarity between the current working condition and the historical ventilation strategy data is calculated through a similarity matching algorithm, and the historical ventilation strategies are sorted from high to low according to their corresponding similarities, and the top n ventilation strategies with the highest similarity ranking are extracted; According to the set similarity threshold, determine whether n strategies are ventilation strategies that meet the similarity threshold, eliminate ventilation strategies below the similarity threshold, and output the similarity matching results of the working conditions. If the similarity threshold is not reached, no similar strategy is matched this time, and wait for historical strategy accumulation to be evaluated again.
3. A ventilation strategy optimization simulation training system according to claim 1, characterized in that: The evaluation and recommendation module scores the ventilation strategies from multiple perspectives through comprehensive evaluation or linear weighting and outputs the comprehensive scores of the strategies, and gives the best recommended strategy according to the similarity of the working conditions and the calculated comprehensive scores; The evaluation process of the evaluation and recommendation module is as follows: Forward the original data matrix, transform the data corresponding to those extremely small indicators into extremely large indicators, and calculate and process them uniformly; x ′ =M-x M is the maximum possible value of the indicator x; Normalize the positive / negative matrix, eliminate the influence of dimension by normalization, and construct the normalized initial matrix; There are n objects to be evaluated, each of which has m indicators or attributes. The original data matrix is constructed as follows: Construct a weighted normalized matrix, and perform vector normalization on the attributes, that is, subtract the mean from each column element and divide it by the norm of the current column vector, and use the cosine distance metric to obtain the normalized matrix Z: Calculate the distance between each solution and the optimal solution and the worst solution; Optimal Solution Z + It is composed of the maximum value of each column element in Z: The worst solution Z - It is composed of the minimum value of each column element in Z: Calculate the degree of proximity between each evaluation object and the optimal solution and the worst solution: where w j is the weight of the jth attribute, and the weight represents the importance. The indicator weight is determined according to the actual situation. The degree of closeness between each evaluation object and the optimal solution is calculated. i 0≤C_i≤1,C_i→1 indicates that the evaluation object is better: According to C i The strategies are evaluated by sorting from large to small, and the better ventilation strategy is selected for recommendation after strategy matching.
4. A ventilation strategy optimization simulation training system according to claim 1, characterized in that: The trend prediction model training module uses a regression prediction algorithm to build a regression prediction model of the target indicator based on the actual values and change trends of the external environment temperature and humidity in a period of time before and after the start and stop of the historical ventilation strategy; The specific implementation process of the trend prediction calculation module is as follows: When the business system interface service module clicks a measurement point outside the security room on the interface, it clicks the trend prediction button in the pop-up dialog box to call the trend prediction interface. Based on the input indicator value of the previous period and the number of points to be predicted, it predicts the change of the indicator in the future period, and returns the result to the business system interface service module in the form of JSON; Click the security room location measurement point on the interface, click the trend prediction button in the pop-up dialog box, call the trend prediction interface according to the selected prediction index and pass in the security room and index to predict the changes of the security room location measurement point in the future, and return the results to the business system interface service module in the form of JSON.
5. A ventilation strategy optimization simulation training system according to claim 1, characterized in that: The indicators include rapidity, stability, accuracy, timeliness and energy consumption. The single ventilation effect evaluation module stores the indicators and the relevant environmental parameters and temperature and humidity change gradient parameters during operation into the ventilation effect knowledge base. The single ventilation effect evaluation module calculates the single ventilation strategy evaluation indicators according to the internal environment temperature and humidity change data and the temperature and humidity setting values from the time when the historical ventilation strategy is turned on to the time when the next ventilation strategy is turned on, and obtains a single ventilation effect evaluation model. The specific implementation process of the single ventilation effect evaluation module is as follows: Read the ventilation strategy start-stop table regularly. When a new ventilation strategy is detected in the ventilation strategy start-stop table, the ventilation effect index calculation is triggered; According to the last ventilation mode start-up and end time, read the internal and external environmental data of the ventilation system to obtain the single ventilation data corresponding to the ventilation strategy and the environmental monitoring data for a period of time before the ventilation strategy is started; Calculate the change gradient of each indicator in the period before the ventilation strategy is turned on, and record the initial and final recorded values of each indicator when the ventilation strategy is turned on, as the initial value of the environment when a single ventilation strategy is turned on or off; Secondly, the target values of various indicators of this ventilation are obtained through the target value change schedule, and the indicators of rapidity, maintenance duration, accuracy, energy consumption and stability of this ventilation are calculated respectively. The initial recorded values of each indicator when the ventilation strategy is turned on and the ventilation effect data are integrated to obtain the evaluation results of a single ventilation effect.
6. A ventilation strategy optimization simulation training system according to claim 5, characterized in that: The ventilation effect fitting model training module performs regular scheduling according to the set scheduling cycle, obtains the ventilation system configuration table and the evaluation result data of the single ventilation effect evaluation table, and trains the single ventilation effect evaluation model for different systems with the evaluation result index as the dependent variable and the historical internal and external temperature and humidity indicators and ventilation strategies as the independent variables.
7. A ventilation strategy optimization simulation training system according to claim 1, characterized in that: The ventilation effect fitting prediction module predicts the temperature rapidity, humidity rapidity, temperature accuracy, humidity accuracy, energy consumption, temperature gradient, humidity gradient, and timeliness indicators under the current environmental conditions and when the equipment is turned on, based on the system number, internal and external temperature and humidity indicators and their change gradients, and system start-up gear indicators input by the business system interface service module.
8. A ventilation strategy optimization simulation training system according to claim 1, characterized in that: The prior evolution rule model analyzes the results of a single ventilation effect evaluation, and uses the prior evolution rule to evolve the ventilation strategy based on the current ventilation strategy. The prior evolution rule is based on the principle of adjustment in a better direction. The direction of evolution includes adjustment of working conditions, quantity, and duration. A new ventilation strategy is generated through the heuristic evolution module and provided to the business system interface service module.
9. A ventilation strategy optimization simulation training system according to claim 1, characterized in that: The specific implementation process of the heuristic evolution module is as follows: Multi-objective optimization: The ventilation strategy optimization problem is defined as follows: min f(x,y,a)=-g(rbf(x,y,a)) x∈Ω a∈A in: x is the equipment gear, y is the environmental parameter, and a is the corresponding ventilation system; rbf(x,y,a) is the ventilation effect prediction result; A is the collection of all ventilation systems, Ω is all adjustable equipment; x i is the i-th equipment gear to be optimized; and The lower and upper limits of the equipment position; f is the optimization target and the strategy score of the ventilation strategy effect prediction result; The objective function f satisfies all equality and inequality constraints. If a particular objective function is maximized (maxf i ), and redefine the problem to minimize its negative value (min-f i ); Genetic algorithms are used for calculation. During the ventilation optimization process, ventilation strategies are genetically selected. The scores of offspring strategies are higher than those of previous strategies. Ultimately, a strategy that performs best in the evolutionary direction will be selected.