CN102997374A - Method and device for forecasting air-conditioning load and air-conditioner - Google Patents

Abstract

The invention relates to the technical field of air conditioners and provides a method and a device for forecasting air-conditioning load and an air-conditioner. The method comprises steps of acquiring the parameter amount affecting the air-conditioning system load; inputting parameters to an air-conditioning load forecasting neuronic network model which is trained in advance; and acquiring the prediction value of the air-conditioning load. The air-conditioning hourly load of the air-conditioning system can be predicted accurately, the refrigerating capacity changes with the air-conditioning load, the energy is not wasted, the operation and the adjustment of a water cooling unit has a basis, and the water cooling unit can operate at the best efficiency state.

Description

A kind of Air-conditioning Load Prediction method, device and air-conditioning

Technical field

The invention belongs to air-conditioning technical field, relate in particular to a kind of Air-conditioning Load Prediction method, device and air-conditioning.

Background technology

The method of operation of traditional air-conditioning system, it mainly is the deviation by direct relatively room temperature and design temperature, thereby regulate the cold that supplies of air conditioner unit, wherein, in the air-conditioning system adjustment process, only air conditioner load is enclosed in the close loop control circuit as interference volume, and does not consider the dynamic change situation of air conditioner load.Owing to do not consider the dynamic change situation of air conditioner load, the inconsistent situation of variation that often causes refrigerating capacity and air conditioner load, for example, refrigerating capacity may be occurred and the required cold of air conditioner load can't be satisfied, perhaps refrigerating capacity is greater than the situation of the required cold of air conditioner load, particularly, and when refrigerating capacity during greater than the required cold of air conditioner load, waste a large amount of energy, do not reach energy-conservation effect.In addition, because the refrigerating capacity of handpiece Water Chilling Units is in passive adjustment state, can't knows next refrigerating capacity constantly, thereby can't guarantee handpiece Water Chilling Units that cooling, chilled water pump are under the best energy efficiency state and move.

To sum up, the air-conditioning system adjustment process of prior art is not considered the dynamic change situation of air conditioner load, may cause the variation of refrigerating capacity and air conditioner load inconsistent, and air-conditioning system equipment can't move under best energy efficiency state.

Summary of the invention

The purpose of the embodiment of the invention is to provide a kind of Air-conditioning Load Prediction method, the air-conditioning system adjustment process that is intended to solve prior art is not considered the dynamic change situation of air conditioner load, may cause the variation of refrigerating capacity and air conditioner load inconsistent, and the problem that can't under best energy efficiency state, move of air-conditioning system equipment.

To achieve these goals, the embodiment of the invention provides following technical scheme:

The embodiment of the invention is achieved in that a kind of Air-conditioning Load Prediction method, and described method comprises:

Obtain the parameter amount that affects the air-conditioning system load;

With the Air-conditioning Load Prediction neural network model of described parameter amount input training in advance, wherein, described Air-conditioning Load Prediction neural network model comprises: input layer, intermediate layer, feedback layer and output layer;

Obtain the predicted value of the air conditioner load of described Air-conditioning Load Prediction neural network model output.

The embodiment of the invention also provides a kind of Air-conditioning Load Prediction system, and described system comprises:

Parameter amount acquiring unit is used for obtaining the parameter amount that affects the air-conditioning system load;

Parameter amount input block is used for the Air-conditioning Load Prediction neural network model with described parameter amount input training in advance, and wherein, described Air-conditioning Load Prediction neural network model comprises: input layer, intermediate layer, feedback layer and output layer;

The predicted value acquiring unit is for the predicted value of the air conditioner load that obtains described Air-conditioning Load Prediction neural network model output.

The embodiment of the invention compared with prior art, beneficial effect is: obtain the parameter amount that affects the air-conditioning system load, and with the Air-conditioning Load Prediction neural network model of described parameter input training in advance, thereby obtain the predicted value of air conditioner load, realize loading when the Accurate Prediction air-conditioning system is pursued, guarantee that refrigerating capacity is consistent with the variation of air conditioner load, avoid energy waste, and regulating for handpiece Water Chilling Units operation provides foundation, guarantees that handpiece Water Chilling Units is under the best energy efficiency state to move.

Description of drawings

In order to be illustrated more clearly in the technical scheme of the embodiment of the invention, the accompanying drawing of required use was done to introduce simply during the below will describe embodiment, apparently, accompanying drawing in the following describes only is some embodiments of the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain according to these accompanying drawings other accompanying drawing.

Fig. 1 is the flow chart of the realization of the Air-conditioning Load Prediction method that provides of the embodiment of the invention one;

Fig. 2 is the schematic diagram of the Air-conditioning Load Prediction neural network model that provides of the embodiment of the invention one;

Fig. 3 is the flow chart of the realization of the described Air-conditioning Load Prediction neural net model method of training in advance that provides of the embodiment of the invention one;

Fig. 4 is the structure chart of the Air-conditioning Load Prediction system that provides of the embodiment of the invention two;

Fig. 5 is the structure chart of the training unit that provides of the embodiment of the invention two.

The specific embodiment

In order to make purpose of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, is not intended to limit the present invention.

The embodiment of the invention provides a kind of Air-conditioning Load Prediction method, and described method comprises:

Obtain the parameter amount that affects the air-conditioning system load;

With the Air-conditioning Load Prediction neural network model of described parameter amount input training in advance, wherein, described Air-conditioning Load Prediction neural network model comprises: input layer, intermediate layer, feedback layer and output layer;

Obtain the predicted value of the air conditioner load of described Air-conditioning Load Prediction neural network model output.

The embodiment of the invention also provides a kind of Air-conditioning Load Prediction system, and described system comprises:

Parameter amount acquiring unit is used for obtaining the parameter amount that affects the air-conditioning system load;

Parameter amount input block is used for the Air-conditioning Load Prediction neural network model with described parameter amount input training in advance, and wherein, described Air-conditioning Load Prediction neural network model comprises: input layer, intermediate layer, feedback layer and output layer;

The predicted value acquiring unit is for the predicted value of the air conditioner load that obtains described Air-conditioning Load Prediction neural network model output.

Below in conjunction with specific embodiment realization of the present invention is described in detail:

Embodiment one

Fig. 1 shows the flow chart of the realization of the Air-conditioning Load Prediction method that the embodiment of the invention one provides, and details are as follows:

In S101, initialize corresponding weights and the threshold value of Air-conditioning Load Prediction neural network model;

In the present embodiment, initialized weights and threshold value can be arbitrary value.

In S102, obtain the parameter amount that affects the air-conditioning system load;

In the present embodiment, the parameter amount can comprise: the parameters such as indoor temperature, outdoor temperature, indoor humidity, outside humidity, air conditioning water system supply and return water temperature and flow, specifically can adjust parameter according to actual conditions.

In S103, with the Air-conditioning Load Prediction neural network model of described parameter amount input training in advance, wherein, described Air-conditioning Load Prediction neural network model comprises: input layer, intermediate layer, feedback layer and output layer;

In S104, obtain the predicted value of the air conditioner load of described Air-conditioning Load Prediction neural network model output.

For the ease of understanding, below provide the schematic diagram of an Air-conditioning Load Prediction neural network model, as shown in Figure 2, but the implementation with this Air-conditioning Load Prediction neural network model is not limited: in this realization example, described Air-conditioning Load Prediction neural network model comprises: input layer, intermediate layer, feedback layer and output layer are specially:

Described input layer: receive input parameter x _i(n);

Described intermediate layer: input parameter is processed $x_j\left(n\right)=f_2(\underset{i=1}{\overset{n_1}{\mathrm{\Σ}}}{w}_{\mathrm{ij}}\left(n\right)x_i\left(n\right)+\underset{k=1}{\overset{n_2}{\mathrm{\Σ}}}{w}_{\mathrm{kj}}{x}_j(n-1)-{\mathrm{\θ}}_j\left(n\right)),$ f ₂＝1/(1+exp(-x))；

Described feedback layer: the output x that feeds back the intermediate layer of last training process _j(n-1);

Described output layer: the predicted value y of output air conditioner load _l(n)=f ₁(u _l(n)),

Wherein, y _l(n) be the output of Air-conditioning Load Prediction neural network model, l=1, f ₁Be linear output function, f ₁=x;

Wherein, n is frequency of training, the number of i input layer input parameter, and j is the neuronic number in intermediate layer, k is that the neuronic number l of feedback layer is the neuronic number of output layer, w _IjInput layer is to the weights in intermediate layer, w _KjThe intermediate layer is to the weights of feedback layer, w _JlFeedback layer is to the weights of output layer, θ _lBe the threshold value in intermediate layer, θ _jBe the threshold value of output layer, preferred j=k.

Wherein, n ₁, n ₂Value can arrange according to actual conditions, for example, can get n ₁=5, n ₂=9.

In the present embodiment, obtain the parameter amount that affects the air-conditioning system load, and with the Air-conditioning Load Prediction neural network model of described parameter input training in advance, thereby obtain the predicted value of air conditioner load, realize loading when the Accurate Prediction air-conditioning system is pursued, guarantee that refrigerating capacity is consistent with the variation of air conditioner load, avoid energy waste, and regulating for handpiece Water Chilling Units operation provides foundation, guarantees that handpiece Water Chilling Units is under the best energy efficiency state to move.

Fig. 3 shows the flow chart of the realization of the described Air-conditioning Load Prediction neural net model method of training in advance that the embodiment of the invention one provides, and details are as follows:

In S301, training sample is inputted the Air-conditioning Load Prediction neural network model of setting up in advance, and obtain output valve corresponding to described training sample;

In S302, whether judge the error of described output valve and standard output value less than preset error value, if, then carry out S303, if not, then carry out S304.

In S303, the weights that current Air-conditioning Load Prediction neural network model is corresponding and threshold value are corresponding weights and the threshold value of neutral net after training;

In S304, recomputate current Air-conditioning Load Prediction neural network model corresponding weights and threshold value, bring the weights and the threshold value that recomputate into set up in advance Air-conditioning Load Prediction neural network model, and S301 is carried out in redirect.

Still describe take the situation of above-mentioned realization example as the process of example to the described Air-conditioning Load Prediction neural network model of training in advance, but be not limited with the training method of this process:

(1) input trained values: x ¹, x ²X ^p, by the output valve of described Air-conditioning Load Prediction neural network model be: y ¹, y ²Y ^p, teacher's value of described output valve is t ¹, t ²T ^p

(2) judge | t ^p-y ^p| whether＜ε sets up, and ε is preset error value;

(3) if establishment, the then w of current training _Jl, w _Ij, w _Kj, θ _l, θ _jBe described Air-conditioning Load Prediction neural network model corresponding weights and threshold value;

(4) if be false, then recomputate weights corresponding to neural network model and threshold value, and with the w that recalculates _Jl, w _Ij, w _Kj, θ _l, θ _jBring described Air-conditioning Load Prediction neural network model into, and execution (1)

Wherein, η is step-length;

${\mathrm{<figure-callout\; id="1"\; label="E\; p"\; filenames="HDA00002690454600021.png"\; state="\{\{state\}\}">E</figure-callout>}}_p=\frac{1}{2}\underset{l=0}{\overset{m-1}{\mathrm{\&Sigma;}}}{(t_l^{p_1}-y_l^{p_1})}^2,$ $E_{\mathrm{pl}}=\frac{1}{2}\underset{l=0}{\overset{m-1}{\mathrm{\&Sigma;}}}{(t_l^{p_1}-y_l^{p_1})}^2,$

Wherein, E _PlThe error of the generation of pl trained values, the value of pl are 1 ... p, E _AlwaysOverall error for all trained values generations

By the actual output y to the Air-conditioning Load Prediction neural network model ¹, y ²Y ^pWith corresponding teacher's value (standard value) t ¹, t ²T ^pError revise weights and threshold value in the model, thereby obtain weights and the threshold value of optimum corresponding to described Air-conditioning Load Prediction neural network model, thereby make the output valve of real output value and expectation approaching as far as possible, realize loading when the Accurate Prediction air-conditioning system is pursued, guarantee that refrigerating capacity is consistent with the variation of air conditioner load, avoid energy waste, and regulate for handpiece Water Chilling Units operation foundation is provided, guarantee that handpiece Water Chilling Units is under the best energy efficiency state to move.

Embodiment two

Fig. 4 shows the structure chart of the Air-conditioning Load Prediction system that the embodiment of the invention two provides, for convenience of explanation, only show the part relevant with the embodiment of the invention, this system can be software unit, hardware cell or the soft or hard combining unit that is built in the air-conditioning.

Described system comprises: parameter amount acquiring unit 41, parameter amount input block 42 and predicted value acquiring unit 43.

Parameter amount acquiring unit 41 is used for obtaining the parameter amount that affects the air-conditioning system load;

Parameter amount input block 42 is used for the Air-conditioning Load Prediction neural network model with described parameter amount input training in advance, and wherein, described Air-conditioning Load Prediction neural network model comprises: input layer, intermediate layer, feedback layer and output layer;

Predicted value acquiring unit 43 is for the predicted value of the air conditioner load that obtains described Air-conditioning Load Prediction neural network model output.

Optionally, described system also comprises initialization unit, is used for initializing corresponding weights and the threshold value of Air-conditioning Load Prediction neural network model.

Optionally, described system also comprises training unit: be used for the described Air-conditioning Load Prediction neural network model of training in advance, described training unit specifically comprises sample input block 51 and judging unit 52, sees also Fig. 5:

Sample input block 51 is used for training sample is inputted the Air-conditioning Load Prediction neural network model of setting up in advance, and obtains output valve corresponding to described training sample;

Judging unit 52, be used for judging that whether the error of described output valve and standard output value is less than preset error value, if, the weights that then current Air-conditioning Load Prediction neural network model is corresponding and threshold value are corresponding weights and the threshold value of neutral net after training, if not, then recomputate current Air-conditioning Load Prediction neural network model corresponding weights and threshold value, and bring the weights and the threshold value that recomputate into set up in advance Air-conditioning Load Prediction neural network model, and start sample input block 51.

Optionally, the input layer of described Air-conditioning Load Prediction neural network model, intermediate layer, feedback layer and output layer are specially:

Input layer: receive input parameter x _i(n);

Described intermediate layer: input parameter is processed $x_j\left(n\right)=f_2(\underset{i=1}{\overset{n_1}{\mathrm{\&Sigma;}}}{w}_{\mathrm{ij}}\left(n\right)x_i\left(n\right)+\underset{k=1}{\overset{n_2}{\mathrm{\&Sigma;}}}{w}_{\mathrm{kj}}{x}_j(n-1)-{\mathrm{\&theta;}}_j\left(n\right)),$ f ₂＝1/(1+exp(-x))；

Described feedback layer: the output x that feeds back the intermediate layer of last training process _j(n-1);

Described output layer: the predicted value y of output air conditioner load _l(n)=f ₁(u _l(n)),

Wherein, y _l(n) be the output of Air-conditioning Load Prediction neural network model, l=1, f ₁Be linear output function, f ₁=x;

Wherein, n is frequency of training, the number of i input layer parameter, and j is the number of intermediate variable, l is the number of output layer parameter, w _Jl, w _Kj, w _IjBe weights corresponding to Air-conditioning Load Prediction neural network model, θ _l, θ _jBe threshold value corresponding to Air-conditioning Load Prediction neural network model.

Optionally, described sample input block 51, concrete input trained values: the x that is used for ¹, x ²X ^p, by the output valve of described Air-conditioning Load Prediction neural network model be: y ¹, y ²Y ^p, teacher's value of described output valve is t ¹, t ²T ^p

Described judging unit 52, concrete being used for judged | t ^p-y ^p| whether＜ε sets up, and ε is preset error value, if establishment, the then w of current training _Jl, w _Ij, w _Kj, θ _l, θ _jBe described Air-conditioning Load Prediction neural network model corresponding weights and threshold value, then recomputate weights corresponding to neural network model and threshold value if be false:

Wherein, η is step-length;

${\mathrm{<figure-callout\; id="1"\; label="E\; p"\; filenames="HDA00002690454600021.png"\; state="\{\{state\}\}">E</figure-callout>}}_p=\frac{1}{2}\underset{l=0}{\overset{m-1}{\mathrm{\&Sigma;}}}{(t_l^{p_1}-y_l^{p_1})}^2,$

And with the w that recalculates _Jl, w _Ij, w _Kj, θ _l, θ _jBring model into, and start the sample input block.

The Air-conditioning Load Prediction system that the embodiment of the invention provides can use in the embodiment of the method one of aforementioned correspondence, and details do not repeat them here referring to the description of above-described embodiment one.

It should be noted that among the said system embodiment that included unit is just divided according to function logic, but is not limited to above-mentioned division, as long as can realize corresponding function; In addition, the concrete title of each functional unit also just for the ease of mutual differentiation, is not limited to protection scope of the present invention.

In addition, one of ordinary skill in the art will appreciate that all or part of step that realizes in the various embodiments described above method is to come the relevant hardware of instruction to finish by program, corresponding program can be stored in the computer read/write memory medium, described storage medium is such as ROM/RAM, disk or CD etc.

The above only is preferred embodiment of the present invention, not in order to limiting the present invention, all any modifications of doing within the spirit and principles in the present invention, is equal to and replaces and improvement etc., all should be included within protection scope of the present invention.

Claims (10)

1. an Air-conditioning Load Prediction method is characterized in that, described method comprises:

Obtain the parameter amount that affects the air-conditioning system load;

With the Air-conditioning Load Prediction neural network model of described parameter amount input training in advance, wherein, described Air-conditioning Load Prediction neural network model comprises: input layer, intermediate layer, feedback layer and output layer;

Obtain the predicted value of the air conditioner load of described Air-conditioning Load Prediction neural network model output.

2. the method for claim 1 is characterized in that, described obtaining before the parameter amount that affects the air-conditioning system load, and described method comprises:

Initialize corresponding weights and the threshold value of Air-conditioning Load Prediction neural network model.

3. the method for claim 1 is characterized in that, described method also comprises the described Air-conditioning Load Prediction neural network model of training in advance, is specially:

A1, the Air-conditioning Load Prediction neural network model that training sample input is set up in advance, and obtain output valve corresponding to described training sample;

A2, judge that whether the error of described output valve and standard output value is less than preset error value;

If a3, the weights that then current Air-conditioning Load Prediction neural network model is corresponding and threshold value are corresponding weights and the threshold value of neutral net after training;

A4, if not then recomputates current Air-conditioning Load Prediction neural network model corresponding weights and threshold value, and brings the weights and the threshold value that recomputate into the described Air-conditioning Load Prediction neural network model of setting up in advance, and carries out a1.

4. method as claimed in claim 3 is characterized in that,

Described input layer: receive input parameter x _i(n);

Described intermediate layer: input parameter is processed $x_j\left(n\right)=f_2(\underset{i=1}{\overset{n_1}{\mathrm{\&Sigma;}}}{w}_{\mathrm{ij}}\left(n\right)x_i\left(n\right)+\underset{k=1}{\overset{n_2}{\mathrm{\&Sigma;}}}{w}_{\mathrm{kj}}{x}_j(n-1)-{\mathrm{\&theta;}}_j\left(n\right)),$ f ₂＝1/(1+exp(-x))；

Described feedback layer: the output x that feeds back the intermediate layer of last training process _j(n-1);

Described output layer: the predicted value y of output air conditioner load _l(n)=f ₁(u _l(n)), Wherein, y _l(n) be the output of Air-conditioning Load Prediction neural network model, l=1, f ₁Be linear output function, f ₁=x;

Wherein, n is frequency of training, the number of i input layer input parameter, and j is the neuronic number in intermediate layer, k is that the neuronic number l of feedback layer is the neuronic number of output layer,, w _IjInput layer is to the weights in intermediate layer, w _KjThe intermediate layer is to the weights of feedback layer, w _JlFeedback layer is to the weights of output layer, θ _lBe the threshold value in intermediate layer, θ _jThreshold value for output layer.

5. method as claimed in claim 4 is characterized in that, the described Air-conditioning Load Prediction neural network model of described training in advance is:

(1) input trained values: x ¹, x ²X ^p, by the output valve of described Air-conditioning Load Prediction neural network model be: y ¹, y ²Y ^p, teacher's value of described output valve is t ¹, t ²T ^p

(2) judge | t ^p-y ^p| whether＜ε sets up, and ε is preset error value;

(3) if establishment, the then w of current training _Jl, w _Ij, w _Kj, θ _l, θ _jBe described Air-conditioning Load Prediction neural network model corresponding weights and threshold value;

(4) if be false, then recomputate weights corresponding to neural network model and threshold value, and with the w that recalculates _Jl, w _Ij, w _Kj, θ _l, θ _jBring described Air-conditioning Load Prediction neural network model into, and execution (1)

Wherein, η is step-length;

Wherein, the value of pl is 1,2 ... p, E _AlwaysOverall error for all trained values generations.

6. an Air-conditioning Load Prediction system is characterized in that, described system comprises:

Parameter amount acquiring unit is used for obtaining the parameter amount that affects the air-conditioning system load;

Parameter amount input block is used for the Air-conditioning Load Prediction neural network model with described parameter amount input training in advance, and wherein, described Air-conditioning Load Prediction neural network model comprises: input layer, intermediate layer, feedback layer and output layer;

The predicted value acquiring unit is for the predicted value of the air conditioner load that obtains described Air-conditioning Load Prediction neural network model output.

7. system as claimed in claim 6 is characterized in that, described system also comprises training unit: be used for the described Air-conditioning Load Prediction neural network model of training in advance, described training unit specifically comprises:

The sample input block is used for training sample is inputted the Air-conditioning Load Prediction neural network model of setting up in advance, and obtains output valve corresponding to described training sample;

Judging unit, be used for judging that whether the error of described output valve and standard output value is less than preset error value, if, the weights that then current Air-conditioning Load Prediction neural network model is corresponding and threshold value are corresponding weights and the threshold value of neutral net after training, if not, then recomputate current Air-conditioning Load Prediction neural network model corresponding weights and threshold value, and bring the weights and the threshold value that recomputate into set up in advance Air-conditioning Load Prediction neural network model, and open the sample input block.

8. system as claimed in claim 7 is characterized in that,

Described input layer: receive input parameter x _i(n);

Described intermediate layer: input parameter is processed $x_j\left(n\right)=f_2(\underset{i=1}{\overset{n_1}{\mathrm{\&Sigma;}}}{w}_{\mathrm{ij}}\left(n\right)x_i\left(n\right)+\underset{k=1}{\overset{n_2}{\mathrm{\&Sigma;}}}{w}_{\mathrm{kj}}{x}_j(n-1)-{\mathrm{\&theta;}}_j\left(n\right)),$ f ₂＝1/(1+exp(-x))；

Described feedback layer: the output x that feeds back the intermediate layer of last training process _j(n-1);

Described output layer: the predicted value y of output air conditioner load _l(n)=f ₁(u _l(n)),

Wherein, y _l(n) be the output of Air-conditioning Load Prediction neural network model, l=1, f ₁Be linear output function, f ₁=x;

Wherein, n is frequency of training, the number of i input layer input parameter, and j is the neuronic number in middle intermediate layer, k is that the neuronic number l of feedback layer is the neuronic number of output layer,, w _IjInput layer is to the weights in intermediate layer, w _KjThe intermediate layer is to the weights of feedback layer, w _JlFeedback layer is to the weights of output layer, θ _lBe the threshold value in intermediate layer, θ _jThreshold value for output layer.

9. system as claimed in claim 8 is characterized in that,

Described sample input block, concrete input trained values: the x that is used for ¹, x ²X ^p, by the output valve of described Air-conditioning Load Prediction neural network model be: y ¹, y ²Y ^p, teacher's value of described output valve is t ¹, t ²T ^p

Described judging unit, concrete being used for judged | t ^p-y ^p| whether＜ε sets up, and ε is preset error value, if establishment, the then w of current training _Jl, w _Ij, w _Kj, θ _l, θ _jBe described Air-conditioning Load Prediction neural network model corresponding weights and threshold value, then recomputate weights corresponding to neural network model and threshold value if be false:

Wherein, η is step-length;

$E_{p1}=\frac{1}{2}\underset{l=0}{\overset{m-1}{\mathrm{\&Sigma;}}}{(t_l^{p_1}-y_l^{p_1})}^2,$

And with the w that recalculates _Jl, w _Ij, w _Kj, θ _l, θ _jBring model into, and start the sample input block.

10. an air-conditioning is characterized in that, described air-conditioning comprises the described Air-conditioning Load Prediction of the arbitrary claim of claim 6 to 9 system.

Images