A device intelligent energy-saving control method and system

Abstract

The present invention relates to the field of equipment control, and more specifically, it relates to an intelligent energy-saving control method and system for equipment. An intelligent energy-saving control system for equipment includes: an energy consumption value acquisition module, an outdoor temperature acquisition module, a personnel density acquisition module, an energy consumption prediction data construction module, a control parameter set acquisition module, a first energy consumption prediction model management module, a first energy consumption prediction module, a control parameter simulation module, and a control parameter adjustment module. The present invention predicts the predicted energy consumption value required at the next moment based on the current energy consumption value, outdoor temperature, and personnel density, and then simulates the optimal control strategy through the improved sparrow algorithm. The optimal control strategy enables the central air conditioner to achieve the minimum energy consumption output effect under the premise of meeting the output energy consumption, thereby realizing intelligent energy-saving control of the central air conditioner.

Description

Technical Field

[0001] The present invention relates to the field of equipment control, and more specifically, to an equipment intelligent energy-saving control method and system. Background Art

[0002] As an important part of the current building mechanical and electrical system, central air conditioning also accounts for the majority of its energy consumption. During the use of central air conditioning, the terminal load is affected by many factors and there are many components of central air conditioning, so the energy consumption of central air conditioning often exceeds expectations, resulting in energy waste. Summary of the invention

[0003] The present invention predicts the predicted energy consumption value required at the next moment based on the current energy consumption value, outdoor temperature and population density, and then simulates the optimal control strategy through the improved sparrow algorithm. The optimal control strategy can enable the central air-conditioning to achieve the minimum energy output effect while meeting the output energy consumption, thereby realizing intelligent energy-saving control of the central air-conditioning equipment.

[0004] An intelligent energy-saving control method for equipment, comprising:

[0005] Get the current energy consumption value Q of the central air conditioner n , the current outdoor temperature T n and the current density of people in the building P n , where n is the information acquisition number, starting from 1, and is used to represent the nth acquisition of energy consumption value, outdoor temperature, and personnel density information, based on the current energy consumption value Q of the central air conditioner n , the current outdoor temperature T n and the current density of people in the building P n Construct energy consumption prediction data F n , energy consumption forecast data F n The storage format is {Q n , T n , P n};

[0006] Get the current control parameter set H of the central air conditioner n , control parameter set H n The storage format is {h1, h2…h i …h I}, where h i is the i-th control parameter of the central air conditioner, and I is the total number of control parameters;

[0007] The energy consumption prediction data is sent to the trained first energy consumption prediction model for calculation, and the predicted energy consumption value G is output. n ;

[0008] In the current control parameter set H of the central air conditioner n and predicted energy consumption value G n On this basis, the control parameters of the central air conditioner at the next moment are simulated through the improved sparrow algorithm, and the optimal control strategy is output;

[0009] The control parameters of the central air conditioner are adjusted through the optimal control strategy.

[0010] Preferably, the first energy consumption prediction model is established based on a Bi-LSTM model;

[0011] The training of the first energy consumption prediction model specifically includes the following steps: obtaining energy consumption prediction data of historical records, constructing a first training sample with the energy consumption prediction data of each historical record and its energy consumption value at the next moment, and then forming a first training sample set with all the first training samples; sending the first training sample set to the first energy consumption prediction model with initialized parameters for training, calculating a first loss value with the post-positioned energy consumption value in the first training sample as the target, and if the first loss value is within a first preset range, outputting the trained first energy consumption prediction model; otherwise, reversely adjusting the parameters in the first energy consumption prediction model with the first loss value, and continuing iterative training.

[0012] Preferably, the control parameters of the central air conditioner at the next moment are simulated by using the improved sparrow algorithm, which specifically includes the following steps:

[0013] S1: Generate M simulation control strategies E m , simulation control strategy E m The storage format is {e m1 , e m2 …e mi …e mI}, where e mi For the simulation control strategy E m The i-th simulated control parameter in the m Form a population collection;

[0014] S2: Simulation control strategy E within the population set m To strengthen;

[0015] S3: Set the maximum number of iterations W, let w = 1, w is used to record the number of iterations;

[0016] S4: Calculate all simulated control strategies E in the population set m The corresponding fitness δ m , all simulated control strategies E in the population set m According to the fitness δ m Arrange them in order from large to small, and set the fitness δ m The maximum simulation control strategy Em Denoted as the simulation control strategy E best , the fitness δ m Minimum simulation control strategy E m Denoted as the simulation control strategy E worst ; Then, according to the division ratio, all the arranged simulation control strategies E m Divide and set the division ratio artificially, usually 2:8; the former after division form the discoverer set, and the latter after division form the follower set;

[0017] S5: For all simulated control strategies E in the discoverer set m Update, the update formula is as follows:

[0018]

[0019] Where j is all the simulated control strategies E in the discoverer set m According to the fitness δ m The corresponding serial number value after arranging in descending order; β is a random number in the interval (0, 1), u1 is the warning value in the interval (0, 1), ST is the safety value; X is a random number that obeys the normal distribution; L is a 1×I matrix, and each element in the matrix is ​​1;

[0020] S6: Calculate all simulated control strategies E in the updated discoverer set m The fitness δ m , the fitness value δ in the discoverer set m The maximum simulation control strategy E m Denoted as the simulation control strategy E part ; All simulated control strategies E for the follower set m Update, the update formula is as follows:

[0021]

[0022] where e worsti For the simulation control strategy E worst The ith simulated control parameter in the set is k, and all simulated control strategies E in the follower set are k. m According to the fitness δ m The corresponding serial number value after arranging in descending order, K is all the simulated control strategies E in the follower set m The total number of parti For the simulation control strategy E part The i-th simulation control parameter in ;

[0023] S7: Select simulation control strategies E one by one from the population set m , for the selected simulation control strategy Em , execute the following content, generate a random number η between 0 and 1 through a random function, and judge "η>P c "Is it true? c is the transformation threshold, if "η>P c " is not true, no operation; otherwise, the selected simulation control strategy E is selected by the following formula m To update:

[0024]

[0025] where e besti For the simulation control strategy E best The i-th item in the simulation control parameter, μ is the step size adjustment coefficient, which is a normally distributed random number with a mean of 0 and a variance of 1, λ is a uniform random number in the interval [-1, 1], and δ best For the simulation control strategy E best The corresponding fitness value, δ worst For the simulation control strategy E worst The corresponding fitness value; σ is a small constant;

[0026] S8: Calculate all simulated control strategies E in the population set m The corresponding fitness δ m , all simulated control strategies E in the population set m According to the fitness δ m Arrange them in order from large to small, and set the fitness δ m The maximum simulation control strategy E m Denoted as the simulation control strategy E best , the fitness δ m Minimum simulation control strategy E m Denoted as the simulation control strategy E worst ; The simulation control strategy E best Stored in the optimal control strategy library, and then all the arranged simulation control strategies E m The former are divided into a discoverer set and the latter are divided into a follower set; and the discoverer set is strengthened to enter S9;

[0027] S9: Determine whether "w<W" is true. If "w<W" is true, assign w+1 to w and return to S5; otherwise, select the fitness value δ in the optimal control strategy library. m The maximum simulation control strategy E m The optimal control strategy is output.

[0028] Preferably, M simulation control strategies E are generated m , specifically including the following steps:

[0029] S1.1: Create an empty temporary set D. The length of temporary set D is I. The storage format of temporary set D is {d1, d2…d i …d I}, select elements d one by one i , for the selected element d i , select a location (e i_max , e i_min ) A random number in the interval d i Assign a value, where e i_max is the maximum value of the ith simulation control parameter, e i_min is the minimum value of the ith simulation control parameter; when all elements in the temporary set D are assigned, the temporary set D is recorded as the simulation control strategy E m , considered as generating a simulated control strategy;

[0030] S1.2: Repeat step S1.1 M times to generate M simulation control strategies E m .

[0031] Preferably, the simulation control strategy E in the population set m Strengthening includes the following steps: Select simulation control strategies E one by one from the population set m , for the selected simulation control strategy E m , the simulation control strategy E is calculated by the following formula m To update:

[0032] e' mi =ε(e i_min +e i_max )-e mi

[0033] Where ε is a random number in the interval (0, 1), e' mi is the updated simulation control strategy E m The i-th simulation control parameter in the simulation control strategy E is updated. m Denoted as the simulation control strategy E' m , the simulated control strategy E m and simulation control strategy E' m The trained second energy consumption prediction model is sent to calculate and the simulation control strategy E is output. m The corresponding simulated energy consumption value R m and simulation control strategy E' m The corresponding simulated energy consumption value R' m ;judge Is it true? ζ is the confidence value. If Not true, continue to select the next simulation control strategy E from the population setm ; Otherwise, calculate the simulation control strategy E respectively m The corresponding fitness δ m and simulation control strategy E' m The corresponding fitness δ' m , if "δ' m >δ m ", the simulation control strategy E' m The simulated control strategy E in the population set m Replace; otherwise, do not simulate the control strategy E in the population set m Replace and continue to select the next simulation control strategy E from the population set m ; When the simulated control strategy E in the population set m After all are selected, the simulation control strategy E in the population set is completed m of strengthening.

[0034] Preferably, the second energy consumption prediction model is established based on BP neural network;

[0035] The training of the second energy consumption prediction model includes the following steps: obtaining a control parameter set of the central air conditioner with historical records; marking the control parameter set according to the actual energy consumption value, recording the marked control parameter set as the second training sample, and forming a second training sample set with all the second training samples; sending the second training sample set to the second energy consumption prediction model with initialized parameters for training, during which the labels corresponding to the second training samples are used as targets, and the second loss value is calculated. If the second loss value is within the second preset range, the trained second energy consumption prediction model is output; otherwise, the parameters in the second energy consumption prediction model are reversely adjusted according to the second loss value, and the iterative training continues.

[0036] Preferably, the simulation control strategy E is calculated m The corresponding fitness δ m , specifically including the following steps: Calculate the simulation control strategy E m With the control parameter set H n The similarity between m ;Simulation control strategy E m The corresponding fitness δ m The calculation formula is as follows:

[0037]

[0038] α1 and α2 are the first weight coefficient and the second weight coefficient respectively, and α1+α2=1.

[0039] Preferably, the discovery set is strengthened, specifically comprising the following steps: selecting simulation control strategies E one by one from the discovery set m, for the selected simulation control strategy E m , do the following:

[0040] S8.1: Traverse all simulated control strategies E in the population set m , select all simulation control strategies E m The maximum value of the ith simulated control parameter in e i_max , select all simulation control strategies E m The minimum value of the ith simulation control parameter in e i_min ;

[0041] S8.2: The simulation control strategy E is calculated by the following formula m To update:

[0042] e' mi =ε(e i_min +e i_max )-e mi

[0043] If the updated simulation control strategy E' m Meet the following conditions Output update times f, enter S8.3; otherwise continue to simulate control strategy E m Perform an update and add one to the update number f;

[0044] S8.3: Determine whether "f>φ" is true, where φ is the update threshold, which is set by the user in advance. If "f>φ" is true, enter S8.4; if "f>φ" is not true, the updated simulation control strategy E' m The simulated control strategy E selected from the discoverer set m Make a replacement;

[0045] S8.4: Select the simulation control strategy E m Remove from the discoverer set and select fitness value δ from the follower set m The maximum simulation control strategy E m Deposit to finders collection.

[0046] Preferably, the simulation control strategy E is calculated m With the control parameter set H n The similarity between m The cosine similarity algorithm is used.

[0047] An intelligent energy-saving control system for equipment, comprising:

[0048] Energy consumption value acquisition module, used to obtain the current energy consumption value of the central air conditioner;

[0049] Outdoor temperature acquisition module, used to obtain the current outdoor temperature;

[0050] A personnel density acquisition module is used to obtain the current personnel density;

[0051] An energy consumption prediction data building module is used to build energy consumption prediction data based on the current energy consumption value of the central air conditioner, the current outdoor temperature and the current density of people in the building;

[0052] A control parameter set acquisition module is used to obtain the current control parameter set of the central air conditioner;

[0053] A first energy consumption prediction model management module, used for training and storing the first energy consumption prediction model;

[0054] A first energy consumption prediction module is used to send energy consumption prediction data to a trained first energy consumption prediction model for calculation and output a predicted energy consumption value;

[0055] The control parameter simulation module is used to simulate the control parameters of the central air conditioner at the next moment through the improved sparrow algorithm and output the optimal control strategy;

[0056] The control parameter adjustment module is used to adjust the control parameters of the central air conditioner through the optimal control strategy.

[0057] The present invention has the following advantages:

[0058] 1. The present invention predicts the predicted energy consumption value required at the next moment based on the current energy consumption value, outdoor temperature and personnel density, and then simulates the optimal control strategy through the improved sparrow algorithm. The optimal control strategy can enable the central air-conditioning to achieve the minimum energy output effect while meeting the output energy consumption, thereby realizing intelligent energy-saving control of the central air-conditioning equipment.

[0059] 2. The present invention strengthens the simulation control strategy within the primary population set so that the primary population set as a whole has a higher fitness value, can converge faster in subsequent iterative simulations, reduce the time of simulation operations, and improve efficiency.

[0060] 3. The present invention further strengthens the discoverer set so that the simulation control strategy E m The quality of the solution is improved, and the convergence speed is further increased; at the same time, the number of updates is limited to avoid falling into the local optimal solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a schematic diagram of the structure of the equipment intelligent energy-saving control system used in an embodiment of the present invention. DETAILED DESCRIPTION

[0062] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0063] Embodiment 1: A device intelligent energy-saving control method, comprising:

[0064] Get the current energy consumption value Q of the central air conditioner n , the current outdoor temperature T n and the current density of people in the building P n , where n is the information acquisition number, counting from 1, and is used to represent the nth acquisition of energy consumption value, outdoor temperature, population density and other information; it should be noted that the energy consumption value of the central air conditioner as an electromechanical device in the building is the total amount of energy consumed within a preset time, which is obtained through the energy consumption monitoring equipment installed at the central air conditioner. The preset time is set manually, which represents the frequency of adjusting the central air conditioner. The current population density in the building refers to the ratio of the total number of people in the building to the total size of the space in the building, which can affect the energy consumption of the central air conditioner in the building. It can be further supplemented that the total number of people in the building can be obtained in the following way: a camera is set at the door of the building, and the people entering and leaving the building are counted through image recognition technology, and then the total number of people in the building is calculated; based on the current energy consumption value Q of the central air conditioner n , the current outdoor temperature T n and the current density of people in the building P n Construct energy consumption prediction data F n , energy consumption forecast data F n The storage format is {Q n , T n , P n};

[0065] Get the current control parameter set H of the central air conditioner n , control parameter set H n The storage format is {h1, h2…h i …h I}, where h i is the i-th control parameter of the central air conditioner, and I is the total number of control parameters. It should be noted that the central air conditioner includes several components, such as water pumps and fans at different locations, and each component corresponds to a control parameter, which can be the speed of the engine inside the water pump or fan;

[0066] The energy consumption prediction data is sent to the trained first energy consumption prediction model for calculation, and the predicted energy consumption value G is output. n ;

[0067] In the current control parameter set H of the central air conditioner nand predicted energy consumption value G n On this basis, the control parameters of the central air conditioner at the next moment are simulated through the improved sparrow algorithm, and the optimal control strategy is output;

[0068] The control parameters of the central air conditioner are adjusted through the optimal control strategy.

[0069] This application predicts the predicted energy consumption value required at the next moment based on the current energy consumption value, outdoor temperature and population density, and then simulates the optimal control strategy through the improved sparrow algorithm. The optimal control strategy can enable the central air-conditioning to achieve the minimum energy output while meeting the output energy consumption, thereby realizing intelligent energy-saving control of the central air-conditioning equipment.

[0070] The first energy consumption prediction model is established based on the Bi-LSTM model. As a time series data prediction neural network, the Bi-LSTM model can well learn the changing trends between time series data, and can significantly improve the prediction accuracy by learning the information connection between the context. The energy consumption value of the central air conditioner has a certain change pattern in time series, and the Bi-LSTM model can be used to predict the energy consumption value.

[0071] The training of the first energy consumption prediction model specifically includes the following steps: obtaining energy consumption prediction data of historical records. It should be noted that the energy consumption prediction data of historical records are data recorded by staff on the normally operating central air conditioners; constructing a first training sample with each historical record of energy consumption prediction data and its energy consumption value at the next moment, and then forming all the first training samples into a first training sample set; sending the first training sample set to the first energy consumption prediction model with initialized parameters for training, and calculating a first loss value with the post-placed energy consumption value in the first training sample as the target. If the first loss value is within a first preset range, which is artificially set, output the trained first energy consumption prediction model; otherwise, reversely adjust the parameters in the first energy consumption prediction model with the first loss value, and continue iterative training.

[0072] The improved sparrow algorithm is used to simulate the control parameters of the central air conditioner at the next moment, which specifically includes the following steps:

[0073] S1: Generate M simulation control strategies E m , simulation control strategy E m The storage format is {e m1 , e m2 …e mi …e mI}, where e mi For the simulation control strategy E m The i-th simulated control parameter in the m Form a population collection;

[0074] Generate M simulation control strategies E m , specifically including the following steps:

[0075] S1.1: Create an empty temporary set D. The length of temporary set D is I. The storage format of temporary set D is {d1, d2…d i …d I}, select elements d one by one i , for the selected element d i , select a location (e i_max , e i_min ) A random number in the interval d i Assign a value, where e i_max is the maximum value of the ith simulation control parameter, e i_min is the minimum value of the ith simulation control parameter. In the initial state, e i_max and e i_min It is set in advance by the operator; when all elements in the temporary set D are assigned, the temporary set D is recorded as the simulation control strategy E m , considered as generating a simulated control strategy;

[0076] S1.2: Repeat step S1.1 M times to generate M simulation control strategies E m ;

[0077] S2: Simulate the control strategy E within the population set m Strengthen and improve the quality of the initial population set;

[0078] The simulated control strategy E in the population set m Strengthening includes the following steps: Select simulation control strategies E one by one from the population set m , for the selected simulation control strategy E m , the simulation control strategy E is calculated by the following formula m To update:

[0079] e' mi =ε(e i_min +e i_max )-e mi

[0080] Where ε is a random number in the interval (0, 1), e' mi is the updated simulation control strategy E m The i-th simulation control parameter in the simulation control strategy E is updated. m Denoted as the simulation control strategy E' m , the simulated control strategy E m and simulation control strategy E'm The trained second energy consumption prediction model is sent to calculate and the simulation control strategy E is output. m The corresponding simulated energy consumption value R m and simulation control strategy E' m The corresponding simulated energy consumption value R' m ;judge Is it true? ζ is the confidence value, which is set manually. If This is not true, indicating that the simulation control strategy E m The corresponding simulated energy consumption value cannot meet the demand of the predicted energy consumption value, and the next simulated control strategy E is selected from the population set. m ; Otherwise, calculate the simulation control strategy E respectively m The corresponding fitness δ m and simulation control strategy E' m The corresponding fitness δ' m , if "δ' m >δ m ", indicating the updated simulation control strategy E' m The fitness is higher and it is more valuable as the first generation population. The simulated control strategy E' m The simulated control strategy E in the population set m Replace; otherwise, the updated simulation control strategy E' m The fitness is lower, and the simulation control strategy E in the population set is not m Replace and continue to select the next simulation control strategy E from the population set m ; When the simulated control strategy E in the population set m After all are selected, the simulation control strategy E in the population set is completed m Strengthening; This application strengthens the simulation control strategy within the initial population set, so that the initial population set as a whole has a higher fitness value, can converge faster in subsequent iterative simulations, reduce the time of simulation operations, and improve efficiency.

[0081] The second energy consumption prediction model is established based on the BP neural network, and the training of the second energy consumption prediction model includes the following steps: obtaining a control parameter set of the central air conditioner recorded in history. It should be noted that the control parameter set in the historical record is a record of the actual operating parameters of different components of the central air conditioner in normal operation through sensors; marking the control parameter set according to the actual energy consumption value, recording the marked control parameter set as the second training sample, and forming all the second training samples into a second training sample set; sending the second training sample set to the second energy consumption prediction model with initialized parameters for training, during which the label corresponding to the second training sample is used as the target, and the second loss value is calculated. If the second loss value is within the second preset range, the second preset range is set manually, and the trained second energy consumption prediction model is output; otherwise, the parameters in the second energy consumption prediction model are adjusted in reverse according to the second loss value, and the iterative training continues.

[0082] Calculation simulation control strategy E m The corresponding fitness δ m , specifically including the following steps: Calculate the simulation control strategy E m With the control parameter set H n The similarity between m , the similarity calculation uses the cosine similarity algorithm; the simulation control strategy E m The corresponding fitness δ m The calculation formula is as follows:

[0083]

[0084] Among them, α1 and α2 are the first weight coefficient and the second weight coefficient respectively, and α1+α2=1, α1 and α2 are used to characterize the importance of simulated energy consumption value and similarity respectively, α1 and α2 are set artificially, generally α1 is 0.75, α2 is 0.25; the fitness value represents a lower energy consumption value and less control parameter changes.

[0085] S3: Set the maximum number of iterations W, let w = 1, w is used to record the number of iterations;

[0086] S4: Calculate all simulated control strategies E in the population set m The corresponding fitness δ m , all simulated control strategies E in the population set m According to the fitness δ m Arrange them in order from large to small, and set the fitness δ m The maximum simulation control strategy E m Denoted as the simulation control strategy E best , the fitness δ m Minimum simulation control strategy E m Denoted as the simulation control strategy E worst; Then, according to the division ratio, all the arranged simulation control strategies E m Divide and set the division ratio artificially, usually 2:8; the former after division form the discoverer set, and the latter after division form the follower set;

[0087] S5: For all simulated control strategies E in the discoverer set m Update, the update formula is as follows:

[0088]

[0089] Where j is all the simulated control strategies E in the discoverer set m According to the fitness δ m The corresponding serial number value after arranging in descending order; β is a random number in the interval (0, 1), generated by a random function, u1 is a warning value in the interval (0, 1), generated by a random function, ST is a safety value, set in advance by the user; X is a random number that obeys a normal distribution; L is a 1×I matrix, and each element in the matrix is ​​1;

[0090] S6: Calculate all simulated control strategies E in the updated discoverer set m The fitness δ m , the fitness value δ in the discoverer set m The maximum simulation control strategy E m Denoted as the simulation control strategy E part ; All simulated control strategies E for the follower set m Update, the update formula is as follows:

[0091]

[0092] Where eworsti is the simulation control strategy E worst The ith simulated control parameter in the set, k is all simulated control strategies E in the follower set m According to the fitness δ m The corresponding serial number value after arranging in descending order, K is all the simulated control strategies E in the follower set m The total number of; eparti is the simulation control strategy E part The ith analog control parameter in ;

[0093] S7: Select simulation control strategies E one by one from the population set m , for the selected simulation control strategy E m , execute the following content, generate a random number η between 0 and 1 through a random function, and judge "η>P c "Is it true? cis the transformation threshold, set by the user, usually 0.7. c " is not true, no operation; otherwise, the selected simulation control strategy E is selected by the following formula m To update:

[0094]

[0095] Where ebesti is the simulation control strategy E best The i-th item in the simulation control parameter, μ is the step size adjustment coefficient, which is a normally distributed random number with a mean of 0 and a variance of 1, λ is a uniform random number in the interval [-1, 1], and δ best For the simulation control strategy E best The corresponding fitness value, δ worst For the simulation control strategy E worst The corresponding fitness value; σ is a small constant;

[0096] S8: Calculate all simulated control strategies E in the population set m The corresponding fitness δ m , all simulated control strategies E in the population set m According to the fitness δ m Arrange them in order from large to small, and set the fitness δ m The maximum simulation control strategy E m Denoted as the simulation control strategy E best , the fitness δ m Minimum simulation control strategy E m Denoted as the simulation control strategy E worst ; The simulation control strategy E best Stored in the optimal control strategy library, and then all the arranged simulation control strategies E m The former are divided into a discoverer set and the latter are divided into a follower set; and the discoverer set is strengthened to enter S9;

[0097] Strengthening the discoverer set includes the following steps: selecting simulation control strategies E from the discoverer set one by one m , for the selected simulation control strategy E m , do the following:

[0098] S8.1: Traverse all simulated control strategies E in the population set m , select all simulation control strategies E m The maximum value of the ith simulated control parameter in e i_max , select all simulation control strategies E m The minimum value of the ith simulation control parameter in ei_min ;

[0099] S8.2: The simulation control strategy E is calculated by the following formula m To update:

[0100] e' mi =ε(e i_min +e i_max )-e mi

[0101] If the updated simulation control strategy E'm satisfies the following conditions Output update times f, enter S8.3; otherwise continue to simulate control strategy E m Perform an update and add one to the update number f; it should be noted that if during the update process, the preset update number, for example, 100 times, is exceeded and the condition is not met, the operation of step S8.4 is directly executed;

[0102] S8.3: Determine whether "f>φ" is true, where φ is the update threshold, which is set by the user in advance. If "f>φ" is true, it means that the number of updates is too many. At this time, the simulation control strategy E m There is a suspicion of falling into the optimal solution, and enter S8.4; if "f>φ" does not hold, it means that the number of updates is in line with expectations. At this time, the simulation control strategy E m It also has a higher fitness value, and the updated simulation control strategy E'm is compared with the simulation control strategy E selected in the discoverer set. m Make a replacement;

[0103] S8.4: Select the simulation control strategy E m Remove from the discoverer set and select fitness value δ from the follower set m The maximum simulation control strategy E m Stored in the discoverer set; This application further strengthens the discoverer set so that the simulation control strategy E m The quality of the solution is improved, and the convergence speed is further increased; at the same time, the number of updates is limited to avoid falling into the local optimal solution.

[0104] S9: Determine whether "w<W" is true. If "w<W" is true, it means that the maximum number of iterations has not been reached. Assign w+1 to w and return to S5; otherwise, select the fitness value δ in the optimal control strategy library. m The maximum simulation control strategy E m The optimal control strategy is output.

[0105] Example 2: An intelligent energy-saving control system for equipment, see Figure 1 ,include:

[0106] Energy consumption value acquisition module, used to obtain the current energy consumption value of the central air conditioner;

[0107] Outdoor temperature acquisition module, used to obtain the current outdoor temperature;

[0108] A personnel density acquisition module is used to obtain the current personnel density;

[0109] An energy consumption prediction data building module is used to build energy consumption prediction data based on the current energy consumption value of the central air conditioner, the current outdoor temperature and the current density of people in the building;

[0110] A control parameter set acquisition module is used to obtain the current control parameter set of the central air conditioner;

[0111] A first energy consumption prediction model management module, used for training and storing the first energy consumption prediction model;

[0112] A first energy consumption prediction module is used to send energy consumption prediction data to a trained first energy consumption prediction model for calculation and output a predicted energy consumption value;

[0113] The control parameter simulation module is used to simulate the control parameters of the central air conditioner at the next moment through the improved sparrow algorithm and output the optimal control strategy;

[0114] The control parameter adjustment module is used to adjust the control parameters of the central air conditioner through the optimal control strategy.

[0115] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A device intelligent energy-saving control method, characterized in that: include: Get the current energy consumption value Q of the central air conditioner n , the current outdoor temperature T n and the current density of people in the building P n , where n is the information acquisition number, starting from 1, and is used to represent the nth acquisition of energy consumption value, outdoor temperature, and personnel density information, based on the current energy consumption value Q of the central air conditioner n , the current outdoor temperature T n and the current density of people in the building P n Construct energy consumption prediction data F n , energy consumption forecast data F n The storage format is {Q n , T n , P n }; Get the current control parameter set H of the central air conditioner n , control parameter set H n The storage format is {h1, h2…h i …h I }, where h i is the i-th control parameter of the central air conditioner, and I is the total number of control parameters; The energy consumption prediction data is sent to the trained first energy consumption prediction model for calculation, and the predicted energy consumption value G is output. n ; In the current control parameter set H of the central air conditioner n and predicted energy consumption value G n On this basis, the control parameters of the central air conditioner at the next moment are simulated through the improved sparrow algorithm, and the optimal control strategy is output; Adjust the control parameters of central air conditioning through optimal control strategy; The improved sparrow algorithm is used to simulate the control parameters of the central air conditioner at the next moment, which specifically includes the following steps: S1: Generate M simulation control strategies E m , simulation control strategy E m The storage format is {e m1 , e m2 …e mi …e mI }, where e mi For the simulation control strategy E m The i-th simulated control parameter in the m Form a population collection; S2: Simulation control strategy E within the population set m To strengthen; S3: Set the maximum number of iterations W, let w = 1, w is used to record the number of iterations; S4: Calculate all simulated control strategies E in the population set m The corresponding fitness δ m , all simulated control strategies E in the population set m According to the fitness δ m Arrange them in order from large to small, and set the fitness δ m The maximum simulation control strategy E m Denoted as the simulation control strategy E best , the fitness δ m Minimum simulation control strategy E m Denoted as the simulation control strategy E worst ; Then, according to the division ratio, all the arranged simulation control strategies E m Divide and set the division ratio artificially, usually 2:8; the former after division form the discoverer set, and the latter after division form the follower set; S5: For all simulated control strategies E in the discoverer set m Update, the update formula is as follows: Where j is all the simulated control strategies E in the discoverer set m According to the fitness δ m The corresponding serial number value after arranging in descending order; β is a random number in the interval (0, 1), u1 is the warning value in the interval (0, 1), and ST is the safety value; η is a random number that follows a normal distribution; L is a 1×I matrix, and each element in the matrix is ​​1; S6: Calculate all simulated control strategies E in the updated discoverer set m The fitness δ m , the fitness value δ in the discoverer set m The maximum simulation control strategy E m Denoted as the simulation control strategy E part ; All simulated control strategies E for the follower set m Update, the update formula is as follows: where e worsti For the simulation control strategy E worst The ith simulated control parameter in the set, k is all simulated control strategies E in the follower set m According to the fitness δ m The corresponding serial number value after arranging in descending order, K is all the simulated control strategies E in the follower set m The total number of parti For the simulation control strategy E part The ith analog control parameter in ; S7: Select simulation control strategies E one by one from the population set m , for the selected simulation control strategy E m , execute the following content, generate a random number η between 0 and 1 through a random function, and judge "η>P c "Is it true? c is the transformation threshold, if "η>P c " is not true, no operation; otherwise, the selected simulation control strategy E is selected by the following formula m To update: where e besti For the simulation control strategy E best The i-th item in the simulation control parameter, μ is the step size adjustment coefficient, which is a normally distributed random number with a mean of 0 and a variance of 1, λ is a uniform random number in the interval [1, 1], and δ best For the simulation control strategy E best The corresponding fitness value, δ worst For the simulation control strategy E worst The corresponding fitness value; σ is a small constant; S8: Calculate all simulated control strategies E in the population set m The corresponding fitness δ m , all simulated control strategies E in the population set m According to the fitness δ m Arrange them in order from large to small, and set the fitness δ m The maximum simulation control strategy E m Denoted as the simulation control strategy E best , the fitness δ m Minimum simulation control strategy E m Denoted as the simulation control strategy E worst ; The simulation control strategy E best Stored in the optimal control strategy library, and then all the arranged simulation control strategies E m Divide the former into a discoverer set and the latter into a follower set; And strengthen the discoverer set and enter S9; S9: Determine whether "w<W" is true. If "w<W" is true, assign w+1 to w and return to S5; Otherwise, select the fitness value δ in the optimal control strategy library m The maximum simulation control strategy E m The optimal control strategy is output.

2. The device intelligent energy-saving control method according to claim 1, characterized in that: The first energy consumption prediction model is established based on the Bi-LSTM model; The training of the first energy consumption prediction model specifically includes the following steps: obtaining energy consumption prediction data of historical records, constructing a first training sample with each energy consumption prediction data of the historical records and its energy consumption value at the next moment, and then forming all the first training samples into a first training sample set; The first training sample set is sent to the first energy consumption prediction model with initialized parameters for training, and the first loss value is calculated with the post-placed energy consumption value in the first training sample as the target. If the first loss value is within the first preset range, the trained first energy consumption prediction model is output; otherwise, the parameters in the first energy consumption prediction model are reversely adjusted with the first loss value, and the iterative training continues.

3. The device intelligent energy-saving control method according to claim 2 is characterized in that: Generate M simulation control strategies E m , specifically including the following steps: S1.1: Create an empty temporary set D. The length of temporary set D is I. The storage format of temporary set D is {d1, d2…d i …d I }, select elements d one by one i , for the selected element d i , select a location (e i_max , e i_min ) A random number in the interval d i Assign a value, where e i_max is the maximum value of the ith simulation control parameter, e i_min is the minimum value of the ith simulation control parameter; When all elements in the temporary set D are assigned values, the temporary set D is recorded as the simulation control strategy E m , considered as generating a simulated control strategy; S1.2: Repeat step S1.1 M times to generate M simulation control strategies E m .

4. The device intelligent energy-saving control method according to claim 3 is characterized in that: The simulated control strategy E in the population set m Strengthening includes the following steps: Select simulation control strategies E one by one from the population set m , for the selected simulation control strategy E m , the simulation control strategy E is calculated by the following formula m To update: And' mi =ε(e i_min +e i_max )-And mi Where ε is a random number in the interval (0, 1), e′ mi is the updated simulation control strategy E m The i-th simulation control parameter in the simulation control strategy E is updated. m Denoted as the simulation control strategy E′ m , the simulated control strategy E m and simulation control strategy E′ m The trained second energy consumption prediction model is sent to calculate and the simulation control strategy E is output. m The corresponding simulated energy consumption value R m and simulation control strategy E′ m The corresponding simulated energy consumption value R′ m ;judge Is it true? ζ is the confidence value. If Not true, continue to select the next simulation control strategy E from the population set m ; Otherwise, calculate the simulation control strategy E respectively m The corresponding fitness δ m and simulation control strategy E′ m The corresponding fitness δ′ m , if "δ′ m >δ m ”, the simulated control strategy E′ m The simulated control strategy E in the population set m Replace; otherwise, do not simulate the control strategy E in the population set m Replace and continue to select the next simulation control strategy E from the population set m ; When the simulated control strategy E in the population set m After all are selected, the simulation control strategy E in the population set is completed m Strengthening.

5. The device intelligent energy-saving control method according to claim 4, characterized in that: The second energy consumption prediction model is established based on BP neural network; The training of the second energy consumption prediction model includes the following steps: obtaining a control parameter set of the central air conditioner recorded in history; The control parameter set is labeled by the actual energy consumption value, the labeled control parameter set is recorded as a second training sample, and all the second training samples are combined into a second training sample set; The second training sample set is sent to the second energy consumption prediction model with initialized parameters for training. During the training, the labels corresponding to the second training samples are used as targets, and the second loss value is calculated. If the second loss value is within the second preset range, the trained second energy consumption prediction model is output; otherwise, the parameters in the second energy consumption prediction model are reversely adjusted according to the second loss value, and the iterative training continues.

6. The device intelligent energy-saving control method according to claim 5, characterized in that: Calculation simulation control strategy E m The corresponding fitness δ m , specifically including the following steps: Calculate the simulation control strategy E m With the control parameter set H n The similarity between m ;Simulation control strategy E m The corresponding fitness δ m The calculation formula is as follows: α1 and α2 are the first weight coefficient and the second weight coefficient respectively, and α1+α2=1.

7. The device intelligent energy-saving control method according to claim 6, characterized in that: Strengthening the discoverer set includes the following steps: selecting simulation control strategies E from the discoverer set one by one m , for the selected simulation control strategy E m , do the following: S8.1: Traverse all simulated control strategies E in the population set m , select all simulation control strategies E m The maximum value of the ith simulated control parameter in e i_max , select all simulation control strategies E m The minimum value of the ith simulation control parameter in e i_min ; S8.2: The simulation control strategy E is calculated by the following formula m To update: And' mi =ε(e i_min +e i_max )-And mi If the updated simulation control strategy E′ m Meet the following conditions Output the update number f and enter S8.3; otherwise, continue to update the simulation control strategy Em and add one to the update number f; S8.3: Determine whether "f>φ" is true, where φ is the update threshold, which is set in advance by the user. If "f>φ" is true, enter S8.4; if "f>φ" is not true, the updated simulation control strategy E′ is m The simulated control strategy E selected from the discoverer set m Make a replacement; S8.4: Select the simulation control strategy E m Remove from the discoverer set and select fitness value δ from the follower set m The maximum simulation control strategy E m Deposit to finders collection.

8. The device intelligent energy-saving control method according to claim 6, characterized in that: Calculation simulation control strategy E m With the control parameter set H n The similarity between m The cosine similarity algorithm is used.

9. An intelligent energy-saving control system for equipment, characterized in that: The system is applied to a device intelligent energy-saving control method according to any one of claims 1 to 8, comprising: Energy consumption value acquisition module, used to obtain the current energy consumption value of the central air conditioner; Outdoor temperature acquisition module, used to obtain the current outdoor temperature; A personnel density acquisition module is used to obtain the current personnel density; An energy consumption prediction data building module is used to build energy consumption prediction data based on the current energy consumption value of the central air conditioner, the current outdoor temperature and the current density of people in the building; A control parameter set acquisition module is used to obtain the current control parameter set of the central air conditioner; A first energy consumption prediction model management module, used for training and storing the first energy consumption prediction model; A first energy consumption prediction module is used to send energy consumption prediction data to a trained first energy consumption prediction model for calculation and output a predicted energy consumption value; The control parameter simulation module is used to simulate the control parameters of the central air conditioner at the next moment through the improved sparrow algorithm and output the optimal control strategy; The control parameter adjustment module is used to adjust the control parameters of the central air conditioner through the optimal control strategy.

Images