CN117515802A - Fixed-frequency central air conditioner daily-load prediction method considering running state of air conditioner - Google Patents
Fixed-frequency central air conditioner daily-load prediction method considering running state of air conditioner Download PDFInfo
- Publication number
- CN117515802A CN117515802A CN202311552798.6A CN202311552798A CN117515802A CN 117515802 A CN117515802 A CN 117515802A CN 202311552798 A CN202311552798 A CN 202311552798A CN 117515802 A CN117515802 A CN 117515802A
- Authority
- CN
- China
- Prior art keywords
- central air
- air conditioner
- air conditioning
- period
- conditioning system
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000004378 air conditioning Methods 0.000 claims abstract description 83
- 238000012544 monitoring process Methods 0.000 claims abstract description 26
- 238000010219 correlation analysis Methods 0.000 claims abstract description 15
- 230000033228 biological regulation Effects 0.000 claims abstract description 11
- 238000010606 normalization Methods 0.000 claims abstract description 4
- 238000012545 processing Methods 0.000 claims abstract description 4
- 230000002068 genetic effect Effects 0.000 claims description 26
- 230000014759 maintenance of location Effects 0.000 claims description 22
- 238000005457 optimization Methods 0.000 claims description 10
- 238000003062 neural network model Methods 0.000 claims description 9
- 238000001816 cooling Methods 0.000 claims description 7
- 230000000644 propagated effect Effects 0.000 claims description 6
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 6
- ATJFFYVFTNAWJD-UHFFFAOYSA-N Tin Chemical compound [Sn] ATJFFYVFTNAWJD-UHFFFAOYSA-N 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 230000036316 preload Effects 0.000 claims 2
- 230000007423 decrease Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0499—Feedforward networks
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/30—Control or safety arrangements for purposes related to the operation of the system, e.g. for safety or monitoring
- F24F11/46—Improving electric energy efficiency or saving
- F24F11/47—Responding to energy costs
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
- F24F11/63—Electronic processing
- F24F11/64—Electronic processing using pre-stored data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Mechanical Engineering (AREA)
- Combustion & Propulsion (AREA)
- Chemical & Material Sciences (AREA)
- Signal Processing (AREA)
- Economics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Water Supply & Treatment (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Public Health (AREA)
- Fuzzy Systems (AREA)
- Physiology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Air Conditioning Control Device (AREA)
Abstract
The invention provides a fixed-frequency central air conditioner daily-load prediction method considering the running state of an air conditioner, which comprises the steps of acquiring n groups of monitoring data in a historical period; carrying out normalization processing on the monitoring data to obtain dimensionless new data, carrying out correlation analysis on the new data influencing the central air conditioning cold demand factors and the central air conditioning system load rate respectively, and distributing weights according to correlation analysis results to obtain a contribution value historical data sequence; establishing an EPGA-BPNN model; obtaining a contribution value data sequence of a set period before a predicted period; inputting the trained EPGA-BPNN model to obtain the central air conditioning system load rate in a prediction period; judging the running state of a central air conditioning system in a prediction period, and formulating a corresponding regulation strategy in the prediction period; the method can improve the accuracy of the daily load prediction of the central air conditioner, can realize the accurate regulation and control of the central air conditioner, and is suitable for the prediction of the load and the operation stage of the air conditioner.
Description
Technical Field
The invention relates to a fixed-frequency central air conditioner daily-load prediction method considering the running state of an air conditioner, and belongs to the technical field of central air conditioner load prediction.
Background
With the development of Chinese economy at a high speed, the building scale in cities is continuously enlarged, and a central air conditioning system is used as one of main energy consumption equipment of the building, so that electricity consumption of the central air conditioning system occupies a large proportion of urban electricity consumption. The load prediction and regulation strategy of the central air conditioner is researched, so that the energy consumption of the central air conditioner can be greatly reduced, and the purposes of reducing the running cost of the air conditioner, saving energy and reducing emission can be well realized.
Central air conditioning load prediction generally utilizes mathematical models and algorithms to predict the load condition of the central air conditioning system over a period of time. Central air conditioning load prediction generally involves a plurality of factors affecting the state of the air conditioner, including indoor and outdoor humiture, wind speed, people flow, building conditions, etc. By collecting and analyzing this data, a predictive model can be built. The method for establishing the prediction model comprises mechanism modeling, data driving modeling and the like. The mechanism modeling is also called white-box modeling, and is to build a model according to the internal structural principle of an object to be researched, the change of each heat transfer medium and each external related factor. The model built based on the mechanism is very cumbersome, a large number of differential equations need to be calculated when designing the controller, and therefore there are great limitations in engineering applications. The data-driven modeling is also called a black box model, and the data-driven model is modeled by utilizing experimental data of a system, does not need a great deal of theoretical knowledge, and only needs to analyze the experimental data and establish a model structure through the relationship between input and output.
At present, a central air conditioner load prediction model and a regulation strategy research are still in a preliminary stage, the central air conditioner load resource collection accuracy is not high, the regulation refinement level is not enough, and a more accurate prediction method for the central air conditioner load needs to be excavated. Therefore, the fixed-frequency central air conditioner daily-load prediction method considering the running state of the air conditioner is researched, the accurate regulation and control of the central air conditioner load are facilitated, and flexible interaction with power grid dispatching is realized.
The above-mentioned problems are to be considered and solved in the course of predicting daily load of constant-frequency central air conditioner taking into account the running state of air conditioner.
Disclosure of Invention
The invention aims to provide a fixed-frequency central air conditioner daily front load prediction method considering the running state of an air conditioner, which solves the problem that the accuracy of central air conditioner daily front load prediction in the prior art is to be improved.
The technical scheme of the invention is as follows:
a method for predicting daily load of a constant-frequency central air conditioner according to the running state of the air conditioner, which comprises the following steps,
s1, acquiring n groups of monitoring data in a historical period, wherein each group of monitoring data comprises factors influencing the cold energy requirement of a central air conditioner and the total running power of the central air conditioner system;
s2, carrying out normalization processing on the monitored data to obtain dimensionless new data, carrying out correlation analysis on the new data influencing the central air conditioning cold demand factors and the central air conditioning system load rate respectively, and distributing weights according to correlation analysis results to obtain a contribution value historical data sequence of each dimensionless new data to the central air conditioning system load rate;
s3, establishing an elite retention strategy genetic algorithm-based EPGA improved counter propagation neural network model, namely an EPGA-BPNN model, taking a contribution value historical data sequence as input data, and training the EPGA-BPNN model by combining a central air conditioning system load rate of a historical period to obtain a trained EPGA-BPNN model;
s4, acquiring monitoring data of a set period before a predicted period, and obtaining a contribution value data sequence of the set period before the predicted period;
s5, inputting the contribution value data sequence of the set period before the prediction period into the trained EPGA-BPNN model to obtain the central air conditioning system load rate of the prediction period;
s6, judging the running state of the central air conditioning system in the prediction period according to the central air conditioning system load rate in the prediction period obtained in the step S5, and formulating a corresponding regulation strategy in the prediction period according to the running state.
Further, in step S1, n sets of monitoring data of the historical period are obtained, each set of monitoring data includes factors affecting the cooling capacity requirement of the central air conditioner and total running power of the central air conditioning system, specifically, n sets of factors affecting the cooling capacity requirement of the central air conditioner are collected, including indoor temperature T in Indoor humidity H in Outdoor temperature T out Outdoor humidity H out And chilled water temperature T cw The method comprises the steps of carrying out a first treatment on the surface of the And collecting the total running power P of the corresponding central air conditioning system air The method comprises the steps of carrying out a first treatment on the surface of the Obtaining n groups of monitoring data { T ] in ,H in ,T out ,H out ,T cw ,P air }。
Further, in step S2, the monitoring data is normalized to obtain dimensionless new data, specifically, normalized monitoring data:
wherein the sequence x 1 ,x 2 ,x 3 ,x 4 ,x 5 Respectively representing input variables of indoor temperature Tin, indoor humidity Hin, outdoor temperature Tout, outdoor humidity Hout and chilled water temperature Tcw, sequence x 6 Running a total power Pair data sequence for a central air conditioning system, i.e. { x } 1 =T in ,x 2 =H in ,x 3 =T out ,x 4 =H out ,x 5 =T cw ,x 6 =P air Post-conversion to yield the output y 1 ,y 2 ,y 3 ,y 4 ,y 5 ,y 6 ∈[0,1]The output is new data without dimension.
Further, in step S2, the correlation analysis between the new data affecting the central air conditioning cooling capacity demand factor and the central air conditioning system load rate is performed respectively:
wherein i=1, 2, 6,k =1, 2, n, LR (k) is a central air conditioning system load factor sequence, a parent sequence; y is i (k) The i standardized output sequence is a subsequence; f (LR (k), y i (k) The relevance of the ith subsequence and the corresponding dimension of the parent sequence is represented, the value range is between 0 and 1, 0 represents uncorrelation, 1 represents strong relevance, and the larger the number is, the stronger the representative relevance is; ρ is the resolution factor; p and q are respectively two-stage minimum difference and two-stage maximum difference:
further, in step S2, weights are assigned according to the correlation analysis result, so as to obtain a historical data sequence of contribution values of the dimensionless new data to the load rate of the central air conditioning system, specifically, a weight a is assigned according to the correlation analysis result 1 ,A 2 ,A 3 ,A 4 ,A 5 ,A 6 The normalized data sequence y 1 ,y 2 ,y 3 ,y 4 ,y 5 ,y 6 Multiplying the distributed weight values to obtain a contribution value historical data sequence for the load rate: y is cvi (k)=y i ·A i Wherein i=1, 2,.. 6,k =1, 2,..n.
Further, in step S3, the back propagation neural network model, namely EPGA-BPNN model, which is improved based on the elite retention policy genetic algorithm EPGA comprises a back propagation neural network model, namely BPNN model and a parameter optimization module,
BPNN model: taking a contribution value historical data sequence from w-v moment to w moment in a historical period as input data, carrying out forward propagation, and obtaining a predicted load rate from w+1 moment to w+u moment; the error between the obtained predicted load rate and the corresponding running total power of the central air conditioning system in the historical period is reversely propagated;
parameter optimization module: and carrying out iterative optimization on the network parameters of the BPNN model by adopting an elite retention strategy genetic algorithm EPGA to obtain the optimized network parameters of the BPNN model.
Further, in step S3, the contribution value historical data sequence is used as input data, and after the EPGA-BPNN model is trained by combining the central air conditioning system load rate of the historical period, the trained EPGA-BPNN model is obtained, specifically,
s31, inputting a contribution value historical data sequence from w-v time to w time in a historical period, inputting a BPNN model of the EPGA-BPNN model, and outputting a central air conditioning system load rate predicted value sequence from w+1 time to w+u time;
s32, calculating a Loss function Loss of a prediction error by the total running power of the central air conditioning system in the historical period and the load rate predicted value sequence of the central air conditioning system;
s33, initializing network parameters of the BPNN model, and taking the initialized network parameters as an initial parameter population;
s34, when a Loss function Loss of the prediction error is propagated reversely, iteratively updating the parameter population by using an elite retention strategy genetic algorithm EPGA, and repeating the steps S31-S34 until the Loss function is smaller than a set value or reaches set times, and solving to obtain an optimal network parameter value;
and S35, assigning an optimal network parameter value to the BPNN model to obtain the trained EPGA-BPNN model.
Further, in step S34, the parameter population is iteratively updated using the elite retention policy genetic algorithm EPGA, in particular,
s341, marking parameter population of elite retention strategy genetic algorithm EPGA asWherein x represents population individuals, and k represents population algebra;
s342, taking the Loss function Loss as an fitness function value of an individual in each iteration, and directly copying the individual with the best performance in the population, namely the individual with the highest fitness function value, to the next generation;
s343, old populationThe old population is +.>The other individuals of the population form a new population by the new individuals generated by genetic operator operation and the individuals with the best performance in the population together>For new->Re-calculating the fitness function value of the individual during the re-evolution; the individuals in the population that perform best are taken as the optimal network parameter values.
Further, in step S6, the operation state of the central air conditioning system in the prediction period includes a start-up phase, an operation phase, and a stop phase.
Further, in step S6, the running state of the central air conditioning system in the predicted period is determined according to the central air conditioning system load rate in the predicted period obtained in step S5, specifically,
in the load factor of the predicted period, there are any one of the following three cases that are determined as the start-up phase:
case one: the load rate continuously rises, and the load rate at the starting moment and the load rate at the ending moment rise and change rate are in a first set interval;
and a second case: the load rate of the existing time period in the predicted time period reaches the second set interval;
and a third case: the load rate continuously drops, and the load rate at the starting moment and the load rate at the ending moment drop change rate are in a third set interval;
judging the operation stage when the load rate of the prediction period is kept in a set range and the load rate change rate is in a fourth set interval;
the load factor at the start time and the load factor at the end time fall at a rate within a fifth set interval, and the load factor falls to 0, and the stop phase is determined.
The beneficial effects of the invention are as follows: according to the fixed-frequency central air conditioner daily load prediction method considering the air conditioner running state, the air conditioner running state is considered, the EPGA-BPNN model is adopted to predict the air conditioner load rate, the air conditioner running stage in the prediction period is judged according to the load rate, the accuracy of the central air conditioner daily load prediction can be improved, the accurate regulation and control of the central air conditioner can be realized, and the method is suitable for air conditioner load and running stage prediction.
Drawings
FIG. 1 is a schematic flow chart of a method for predicting daily load of a constant-frequency central air conditioner according to an embodiment of the invention, which takes into consideration the running state of the air conditioner;
FIG. 2 is a schematic diagram illustrating iterative updating of parameter populations using elite retention strategy genetic algorithm EPGA in an embodiment;
fig. 3 is a schematic diagram of a trend of load factor change corresponding to an operation state of a central air conditioning system in an embodiment.
Detailed Description
Preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.
Examples
A fixed frequency central air conditioner daily load prediction method considering the running state of an air conditioner, as shown in figure 1, comprises the following steps,
s1, acquiring n groups of monitoring data of a historical period, wherein each group of monitoring data comprises factors influencing the cold energy requirement of a central air conditioner and the total running power of the central air conditioner system.
In step S1, n sets of monitoring data of the history period are acquired, each set of monitoring data including the influenceThe factors of the cold energy demand of the central air conditioner and the total running power of the central air conditioning system are specifically, n groups of factors affecting the cold energy demand of the central air conditioner are collected and comprise the indoor temperature T in Indoor humidity H in Outdoor temperature T out Outdoor humidity H out And chilled water temperature T cw Collecting through a temperature and humidity sensor; and collecting the total running power P of the corresponding central air conditioning system air The method comprises the steps of carrying out a first treatment on the surface of the Obtaining n groups of monitoring data { T ] in ,H in ,T out ,H out ,T cw ,P air }。
S2, carrying out normalization processing on the monitored data to obtain dimensionless new data, carrying out correlation analysis on the new data influencing the central air conditioning cold demand factors and the central air conditioning system load rate respectively, and distributing weights according to correlation analysis results to obtain a contribution value historical data sequence of each dimensionless new data to the central air conditioning system load rate.
In step S2, the monitoring data is normalized to obtain dimensionless new data, specifically, normalized monitoring data:
wherein the sequence x 1 ,x 2 ,x 3 ,x 4 ,x 5 Respectively representing input variables of indoor temperature Tin, indoor humidity Hin, outdoor temperature Tout, outdoor humidity Hout and chilled water temperature Tcw, sequence x 6 Running a total power Pair data sequence for a central air conditioning system, i.e. { x } 1 =T in ,x 2 =H in ,x 3 =T out ,x 4 =H out ,x 5 =T cw ,x 6 =P air Post-conversion to yield the output y 1 ,y 2 ,y 3 ,y 4 ,y 5 ,y 6 ∈[0,1]The output is new data without dimension.
In step S2, the new data affecting the cooling capacity demand factor of the central air conditioner are respectively analyzed in correlation with the load factor of the central air conditioner system:
wherein i=1, 2, 6,k =1, 2, n, LR (k) is a central air conditioning system load factor sequence, a parent sequence; y is i (k) The i standardized output sequence is a subsequence; f (LR (k), y i (k) The relevance of the ith subsequence and the corresponding dimension of the parent sequence is represented, the value range is between 0 and 1, 0 represents uncorrelation, 1 represents strong relevance, and the larger the number is, the stronger the representative relevance is; ρ is the resolution factor; p and q are respectively two-stage minimum difference and two-stage maximum difference:
in step S2, weight is allocated according to the correlation analysis result to obtain a historical data sequence of contribution value of each dimensionless new data to the load rate of the central air conditioning system, specifically, weight A is allocated according to the correlation analysis result 1 ,A 2 ,A 3 ,A 4 ,A 5 ,A 6 The normalized data sequence y 1 ,y 2 ,y 3 ,y 4 ,y 5 ,y 6 Multiplying the distributed weight values to obtain a contribution value historical data sequence for the load rate: y is cvi (k)=y i ·A i Wherein i=1, 2,.. 6,k =1, 2,..n.
S3, establishing an elite retention strategy genetic algorithm-based EPGA improved counter propagation neural network model, namely an EPGA-BPNN model, taking a contribution value historical data sequence as input data, and training the EPGA-BPNN model by combining a central air conditioning system load rate in a historical period to obtain a trained EPGA-BPNN model.
In step S3, the back propagation neural network model (EPGA-BPNN model) improved based on elite retention policy genetic algorithm (EPGA) comprises a back propagation neural network model (BPNN model) and a parameter optimization module,
BPNN model: taking a contribution value historical data sequence from w-v moment to w moment in a historical period as input data, carrying out forward propagation, and obtaining a predicted load rate from w+1 moment to w+u moment; the error between the obtained predicted load rate and the corresponding running total power of the central air conditioning system in the historical period is reversely propagated;
parameter optimization module: and carrying out iterative optimization on the network parameters of the BPNN model by adopting an elite retention strategy genetic algorithm EPGA to obtain the optimized network parameters of the BPNN model.
In step S3, the contribution value historical data sequence is used as input data, and after the EPGA-BPNN model is trained by combining the central air conditioning system load rate of the historical period, the trained EPGA-BPNN model is obtained, specifically,
s31, inputting a contribution value historical data sequence from w-v time to w time in a historical period, inputting a BPNN model of the EPGA-BPNN model, and outputting a central air conditioning system load rate predicted value sequence from w+1 time to w+u time;
s32, calculating a Loss function Loss of a prediction error by the total running power of the central air conditioning system in the historical period and the load rate predicted value sequence of the central air conditioning system;
s33, initializing network parameters of the BPNN model, and taking the initialized network parameters as an initial parameter population;
s34, when a Loss function Loss of the prediction error is propagated reversely, iteratively updating the parameter population by using an elite retention strategy genetic algorithm EPGA, and repeating the steps S31-S34 until the Loss function is smaller than a set value or reaches set times, and solving to obtain an optimal network parameter value;
in step S34, during the back propagation of the Loss function Loss of the prediction error, the elite retention policy genetic algorithm EPGA is used to update the parameter population iteratively, specifically as shown in fig. 2:
s341, marking parameter population of elite retention strategy genetic algorithm EPGA asWherein x represents population individuals, and k represents population algebra;
s342, taking the Loss function Loss as an fitness function value of an individual in each iteration, and directly copying the individual with the best performance in the population, namely the individual with the highest fitness function value, to the next generation;
s343, old populationThe old population is +.>The other individuals of the population form a new population by the new individuals generated by genetic operator operation and the individuals with the best performance in the population together>For new->Re-calculating the fitness function value of the individual during the re-evolution; the individuals in the population that perform best are taken as the optimal network parameter values.
And S35, assigning an optimal network parameter value to the BPNN model to obtain the trained EPGA-BPNN model.
In step S3, the parameter optimization module adopts an elite retention policy genetic algorithm EPGA to solve the optimal network parameters for the BPNN model. Each individual in the EPGA algorithm population represents a set of network parameters of the BPNN model, including all weights and thresholds of the BPNN model, and when the elite retention policy genetic algorithm EPGA iterates, the mean square error function Loss of the BPNN output value and the true value is used as an fitness function for evaluating the individual performance.
In step S3, the population of elite retention strategy genetic algorithm EPGA is marked as(where x represents population individuals and k represents population algebra), EPGA employs +.>Searching optimal individuals in an iterative updating mode, and directly copying the individuals with the optimal performance in the population to the next generation in each iteration; old population->On the premise of keeping the whole scale unchanged, keeping +.>A group of network parameters with the best performance, and a new population is formed together with new individuals generated by other individuals through genetic operator operation>At->The fitness function value of the individual is recalculated when evolving again.
S4, acquiring monitoring data of a set period before the prediction period, and obtaining a contribution value data sequence of the set period before the prediction period.
In step S4, the same method as in step S2 is used to obtain the contribution value data sequence of the set period before the predicted period for the monitored data of the set period before the predicted period.
S5, inputting the contribution value data sequence of the set period before the prediction period into the trained EPGA-BPNN model to obtain the central air conditioning system load rate of the prediction period.
In step S5, the trained EPGA-BPNN model is input by using the contribution value data sequence from the T-n time to the T time, and the air conditioner load rate LR from the T+1 time to the T+m time in the prediction period is predicted fv (j) Where j=1, 2,..m.
In step S5, the central air conditioning system load rate=the central air conditioning system load amount/the sum of rated powers of the respective devices of the central air conditioning system. Since the sum of rated powers of the respective devices in the central air conditioning system is a constant value, the prediction of the load factor of the central air conditioning system is also called central air conditioning load prediction.
S6, judging the running state of the central air conditioning system in the prediction period according to the central air conditioning system load rate in the prediction period obtained in the step S5, and formulating a corresponding regulation strategy in the prediction period according to the running state.
In step S6, the operation state of the central air conditioning system of the prediction period includes a start-up phase, an operation phase, and a stop phase.
In step S6, the running state of the central air conditioning system in the predicted period is determined according to the central air conditioning system load rate in the predicted period obtained in step S5, specifically,
in the load factor of the predicted period, there are any one of the following three cases that are determined as the start-up phase:
case one: the load factor is continuously increased, and the load factor at the starting moment and the load factor at the ending moment are increased and changed at a first set interval such as [180, 570], wherein the interval unit is%/1 h;
and a second case: the load rate of the existing time period in the predicted time period reaches the second set interval such as [80,100], and the interval unit is;
and a third case: the load rate continuously drops, and the load rate at the starting moment and the load rate at the ending moment drop change rates are in a third set interval such as [20,34], and the interval unit is that;
when the load rate of the predicted period is maintained within a set range such as 50% to 80%, and the load rate change rate is within a fourth set interval such as [0,14], the interval unit is%/1 h, and the operation stage is determined;
the load factor at the start time and the load factor at the end time decrease in the rate of change are within a fifth set interval [300, 420], the interval unit is%/1 h, and the load factor decreases to 0, and the stop phase is determined.
According to the fixed-frequency central air conditioner daily load prediction method considering the air conditioner running state, the air conditioner running state is considered, the EPGA-BPNN model is adopted to predict the air conditioner load rate, the air conditioner running stage in the prediction period is judged according to the load rate, the accuracy of the central air conditioner daily load prediction can be improved, the accurate regulation and control of the central air conditioner can be realized, and the method is suitable for air conditioner load and running stage prediction.
Although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that the foregoing embodiments may be modified or equivalents substituted for some of the features thereof. All equivalent structures made by the content of the specification and the drawings of the invention are directly or indirectly applied to other related technical fields, and are also within the scope of the invention.
Claims (10)
1. A fixed frequency central air conditioner daily load prediction method considering the running state of an air conditioner is characterized in that: comprises the steps of,
s1, acquiring n groups of monitoring data in a historical period, wherein each group of monitoring data comprises factors influencing the cold energy requirement of a central air conditioner and the total running power of the central air conditioner system;
s2, carrying out normalization processing on the monitored data to obtain dimensionless new data, carrying out correlation analysis on the new data influencing the central air conditioning cold demand factors and the central air conditioning system load rate respectively, and distributing weights according to correlation analysis results to obtain a contribution value historical data sequence of each dimensionless new data to the central air conditioning system load rate;
s3, establishing an elite retention strategy genetic algorithm-based EPGA improved counter propagation neural network model, namely an EPGA-BPNN model, taking a contribution value historical data sequence as input data, and training the EPGA-BPNN model by combining a central air conditioning system load rate of a historical period to obtain a trained EPGA-BPNN model;
s4, acquiring monitoring data of a set period before a predicted period, and obtaining a contribution value data sequence of the set period before the predicted period;
s5, inputting the contribution value data sequence of the set period before the prediction period into the trained EPGA-BPNN model to obtain the central air conditioning system load rate of the prediction period;
s6, judging the running state of the central air conditioning system in the prediction period according to the central air conditioning system load rate in the prediction period obtained in the step S5, and formulating a corresponding regulation strategy in the prediction period according to the running state.
2. As claimed inThe fixed-frequency central air conditioner daily-load prediction method considering the running state of the air conditioner is characterized by comprising the following steps of: in step S1, n sets of monitoring data of the historical period are obtained, each set of monitoring data includes factors affecting the cooling capacity requirement of the central air conditioner and total running power of the central air conditioning system, specifically, n sets of factors affecting the cooling capacity requirement of the central air conditioner are collected, including indoor temperature T in Indoor humidity H in Outdoor temperature T out Outdoor humidity H out And chilled water temperature T cw The method comprises the steps of carrying out a first treatment on the surface of the And collecting the total running power P of the corresponding central air conditioning system air The method comprises the steps of carrying out a first treatment on the surface of the Obtaining n groups of monitoring data { T ] in ,H in ,T out ,H out ,T cw ,P air }。
3. The fixed frequency central air conditioner daily pre-load prediction method considering the running state of the air conditioner as claimed in claim 1, wherein: in step S2, the monitoring data is normalized to obtain dimensionless new data, specifically, normalized monitoring data:
wherein the sequence x 1 ,x 2 ,x 3 ,x 4 ,x 5 Respectively representing input variables of indoor temperature Tin, indoor humidity Hin, outdoor temperature Tout, outdoor humidity Hout and chilled water temperature Tcw, sequence x 6 Running a total power Pair data sequence for a central air conditioning system, i.e. { x } 1 =T in ,x 2 =H in ,x 3 =T out ,x 4 =H out ,x 5 =T cw ,x 6 =P air Post-conversion to yield the output y 1 ,y 2 ,y 3 ,y 4 ,y 5 ,y 6 ∈[0,1]The output is new data without dimension.
4. A fixed frequency central air conditioning daily load prediction method taking into account the running state of an air conditioner as claimed in any one of claims 1 to 3, wherein: in step S2, the new data affecting the cooling capacity demand factor of the central air conditioner are respectively analyzed in correlation with the load factor of the central air conditioner system:
wherein i=1, 2, 6,k =1, 2, n, LR (k) is a central air conditioning system load factor sequence, a parent sequence; y is i (k) The i standardized output sequence is a subsequence; f (LR (k), y i (k) The relevance of the ith subsequence and the corresponding dimension of the parent sequence is represented, the value range is between 0 and 1, 0 represents uncorrelation, 1 represents strong relevance, and the larger the number is, the stronger the representative relevance is; ρ is the resolution factor; p and q are respectively two-stage minimum difference and two-stage maximum difference:
5. a fixed frequency central air conditioning daily load prediction method taking into account the running state of an air conditioner as claimed in any one of claims 1 to 3, wherein: in step S2, weight is allocated according to the correlation analysis result to obtain a historical data sequence of contribution value of each dimensionless new data to the load rate of the central air conditioning system, specifically, weight A is allocated according to the correlation analysis result 1 ,A 2 ,A 3 ,A 4 ,A 5 ,A 6 The normalized data sequence y 1 ,y 2 ,y 3 ,y 4 ,y 5 ,y 6 Multiplying the distributed weight values to obtain a contribution value historical data sequence for the load rate: y is cvi (k)=y i ·A i Wherein i=1, 2,.. 6,k =1, 2,..n.
6. A fixed frequency central air conditioning daily load prediction method taking into account the running state of an air conditioner as claimed in any one of claims 1 to 3, wherein: in step S3, the back propagation neural network model (EPGA-BPNN model) improved based on elite retention policy genetic algorithm (EPGA) comprises a back propagation neural network model (BPNN model) and a parameter optimization module,
BPNN model: taking a contribution value historical data sequence from w-v moment to w moment in a historical period as input data, carrying out forward propagation, and obtaining a predicted load rate from w+1 moment to w+u moment; the error between the obtained predicted load rate and the corresponding running total power of the central air conditioning system in the historical period is reversely propagated;
parameter optimization module: and carrying out iterative optimization on the network parameters of the BPNN model by adopting an elite retention strategy genetic algorithm EPGA to obtain the optimized network parameters of the BPNN model.
7. A fixed frequency central air conditioning daily load prediction method taking into account the running state of an air conditioner as claimed in any one of claims 1 to 3, wherein: in step S3, the contribution value historical data sequence is used as input data, and after the EPGA-BPNN model is trained by combining the central air conditioning system load rate of the historical period, the trained EPGA-BPNN model is obtained, specifically,
s31, inputting a contribution value historical data sequence from w-v time to w time in a historical period, inputting a BPNN model of the EPGA-BPNN model, and outputting a central air conditioning system load rate predicted value sequence from w+1 time to w+u time;
s32, calculating a Loss function Loss of a prediction error by the total running power of the central air conditioning system in the historical period and the load rate predicted value sequence of the central air conditioning system;
s33, initializing network parameters of the BPNN model, and taking the initialized network parameters as an initial parameter population;
s34, when a Loss function Loss of the prediction error is propagated reversely, iteratively updating the parameter population by using an elite retention strategy genetic algorithm EPGA, and repeating the steps S31-S34 until the Loss function is smaller than a set value or reaches set times, and solving to obtain an optimal network parameter value;
and S35, assigning an optimal network parameter value to the BPNN model to obtain the trained EPGA-BPNN model.
8. The method for predicting the daily load of the constant-frequency central air conditioner according to claim 7, wherein the method is characterized in that: in step S34, the parameter population is iteratively updated using the elite retention policy genetic algorithm EPGA, in particular,
s341, marking parameter population of elite retention strategy genetic algorithm EPGA asWherein x represents population individuals, and k represents population algebra;
s342, taking the Loss function Loss as an fitness function value of an individual in each iteration, and directly copying the individual with the best performance in the population, namely the individual with the highest fitness function value, to the next generation;
s343, old populationThe old population is +.>The other individuals of the population form a new population by the new individuals generated by genetic operator operation and the individuals with the best performance in the population together>For new->Re-calculating the fitness function value of the individual during the re-evolution; the individuals in the population that perform best are taken as the optimal network parameter values.
9. A fixed frequency central air conditioning daily load prediction method taking into account the running state of an air conditioner as claimed in any one of claims 1 to 3, wherein: in step S6, the operation state of the central air conditioning system of the prediction period includes a start-up phase, an operation phase, and a stop phase.
10. The fixed frequency central air conditioner daily pre-load prediction method considering the running state of the air conditioner as claimed in claim 9, wherein: in step S6, the running state of the central air conditioning system in the predicted period is determined according to the central air conditioning system load rate in the predicted period obtained in step S5, specifically,
in the load factor of the predicted period, there are any one of the following three cases that are determined as the start-up phase:
case one: the load rate continuously rises, and the load rate at the starting moment and the load rate at the ending moment rise and change rate are in a first set interval;
and a second case: the load rate of the existing time period in the predicted time period reaches the second set interval;
and a third case: the load rate continuously drops, and the load rate at the starting moment and the load rate at the ending moment drop change rate are in a third set interval;
judging the operation stage when the load rate of the prediction period is kept in a set range and the load rate change rate is in a fourth set interval;
the load factor at the start time and the load factor at the end time fall at a rate within a fifth set interval, and the load factor falls to 0, and the stop phase is determined.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311552798.6A CN117515802A (en) | 2023-11-20 | 2023-11-20 | Fixed-frequency central air conditioner daily-load prediction method considering running state of air conditioner |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311552798.6A CN117515802A (en) | 2023-11-20 | 2023-11-20 | Fixed-frequency central air conditioner daily-load prediction method considering running state of air conditioner |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117515802A true CN117515802A (en) | 2024-02-06 |
Family
ID=89762288
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311552798.6A Pending CN117515802A (en) | 2023-11-20 | 2023-11-20 | Fixed-frequency central air conditioner daily-load prediction method considering running state of air conditioner |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117515802A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117968208A (en) * | 2024-03-29 | 2024-05-03 | 中建安装集团有限公司 | Environment system control method and control system |
-
2023
- 2023-11-20 CN CN202311552798.6A patent/CN117515802A/en active Pending
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117968208A (en) * | 2024-03-29 | 2024-05-03 | 中建安装集团有限公司 | Environment system control method and control system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Jallal et al. | A hybrid neuro-fuzzy inference system-based algorithm for time series forecasting applied to energy consumption prediction | |
CN111695793B (en) | Method and system for evaluating energy utilization flexibility of comprehensive energy system | |
Yao et al. | Hourly cooling load prediction by a combined forecasting model based on analytic hierarchy process | |
CN115095907B (en) | Intelligent heat supply energy-saving regulation and control method and system based on deep learning and storage medium | |
Moradzadeh et al. | Heating and cooling loads forecasting for residential buildings based on hybrid machine learning applications: A comprehensive review and comparative analysis | |
CN105631528B (en) | Multi-target dynamic optimal power flow solving method based on NSGA-II and approximate dynamic programming | |
CN117515802A (en) | Fixed-frequency central air conditioner daily-load prediction method considering running state of air conditioner | |
CN114692265A (en) | Zero-carbon building optimization design method based on deep reinforcement learning | |
Ma et al. | Reducing the consumption of household systems using hybrid deep learning techniques | |
CN114777192B (en) | Secondary network heat supply autonomous optimization regulation and control method based on data association and deep learning | |
CN117354171B (en) | Platform health condition early warning method and system based on Internet of things platform | |
CN113344192A (en) | Enterprise-level motor system energy-saving optimization automatic control method and system | |
CN113048807A (en) | Air cooling unit backpressure abnormity detection method | |
CN115983430A (en) | Method and system for managing and optimizing comprehensive energy system | |
CN115879369A (en) | Coal mill fault early warning method based on optimized LightGBM algorithm | |
Vasanthkumar et al. | Improving energy consumption prediction for residential buildings using Modified Wild Horse Optimization with Deep Learning model | |
CN114429248A (en) | Transformer apparent power prediction method | |
Godahewa et al. | Simulation and optimisation of air conditioning systems using machine learning | |
CN108108837A (en) | A kind of area new energy power supply structure optimization Forecasting Methodology and system | |
CN117313351A (en) | Safe data center optimized cold prediction method and system | |
CN116578134A (en) | Universal base station temperature control method and system based on reinforcement learning | |
Kaligambe et al. | Indoor Room Temperature and Relative Humidity Estimation in a Commercial Building Using the XGBoost Machine Learning Algorithm | |
CN114091767A (en) | Method and system for robust random planning of data-driven comprehensive energy system | |
CN115264768A (en) | Building air conditioner energy-saving control method and system based on Internet of things | |
CN113807019A (en) | MCMC wind power simulation method based on improved scene classification and coarse grain removal |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |