TW200839204A - Energy-saving verification method for air-conditioning system - Google Patents

Abstract

An energy-saving verification method for air-conditioning system, which can verify various energy-saving effects. The method begins by defining independent variables and dependent variables, establishing various modes and performing mode selection to confirm the mode most suitable for the system, then proceeds data screening to prevent mode accuracy from interference, and uses the technique of drawing scattered figures to ensure that the computed power consumption is not obtained by extrapolation, thereby assuring the accuracy. In computation, the actual power consumptions before energy-saving are summed and identical independent variables are inputted into the power consumption mode after energy-saving, which are of the same condition and base and which are not of transient data, thereby having accuracy and representativeness.

Description

200839204 IX. Description of the Invention: [Technical Field of the Invention] The present invention relates to a method for verifying the power saving effect of an air conditioning system. ‘In particular, it refers to a verification method for calculating energy saving performance by establishing an accurate power consumption mode. [Prior Art] With industrial development and economic growth, electricity consumption in various regions has not only remained high but has even reached 4 per year. The use of electricity is not only a problem of the cost of the earth's resources. At present, more than 75% of Taiwan's power generation methods are thermal power generation. Each time it is produced, it produces 0.67 kilograms of carbon dioxide. The carbon dioxide that is continuously emitted is being converted to & The ball pocket is warm and the climate is abnormal. Therefore, saving electricity is a priority. In the case of Taiwan's overall summer electricity use, it is nearly air-conditioning, and if it can be energy-saving for the air-conditioning system, it can directly and effectively suppress power peak load and carbon dioxide emissions. t factors affecting the power consumption of the air-conditioning system, in addition to the insulation and ring of the building, the system itself is the rated power consumption including the design of the ice water mainframe, and the maintenance of the cold water department of the township. Products or methods that improve the performance of air-conditioning systems have a certain degree of help in energy conservation, but there is no accurate and effective way to verify the energy-saving effect of these products or methods. In terms of the impact of maintenance work on power consumption, for example, if the quality of the frozen oil in the mechanical-freezing system deteriorates and the lubrication effect is poor, the internal friction of the compressor becomes large and the power consumption is increased; for example, the evaporator Conductive heat-related components such as condensers, etc., once impurities or cold oil are deposited in the internal pipeline, the heat transfer efficiency is reduced and the power consumption is increased. 5 200839204 For the maintenance and maintenance of power consumption problems, there is currently a solution 疋, using polarized cold oil additives, which can improve the lubrication effect of cold bed oil, and the heat exchange rate between copper official and cold coal. As for the verification of the energy-saving effect of the additive, it is generally recorded before and after the use of the additive data, and the power consumption of the same cooling water temperature, ice water outlet temperature and ~ East, force conditions are taken before and after the addition. Compare. This verification method can only prove that the product is effective. However, because the ice water host has different cooling water temperature, ice water temperature and freezing capacity, the power consumption is different, so it is impossible to accurately calculate the continuous hit, the real section. Energy, even energy saving potential, recycling years. In terms of the impact of cooling water management on power consumption, the impact of cooling water on the power consumption of ice water mains is often the fastest and most influential. This is because cooling water plays the role of indoor space in air conditioning systems. The role of heat in the atmosphere for heat exchange, cooling water is equivalent to being in an open environment and most affected by the environment; on the other hand, the wall of the condenser (to receive cooling water and exchange heat with ice water) The power consumption of dirt and ice water host increases linearly. For example, when there is 0 3mm thick dirt on the wall of the condenser, the power consumption of the whole ice water main unit increases by about 12%. The thicker the dirt, the more power consumption. It can be seen that the dirt of the cooling water is enough to control the power consumption, and improving the quality of the cooling water is an important part of the air conditioning energy saving. In response to this problem, chemicals that can clearly see the effects of water quality improvement are commonly used in the industry. However, after running for a period of time, the method will face the problem of concentration of cooling water chemicals, and it must be discharged and replenished with new water. However, the cooling water to be discharged contains various chemical components and needs to be treated with wastewater to meet environmental standards. Wastewater treatment costs are extremely high. In proving the effectiveness of the water treatment party 200839204, it is mainly to observe changes in the trend temperature, coefficient of performance (COP), water quality analysis, etc., but it is still impossible to calculate the actual energy saving effect. In other words, the improvement of the quality of the cooling water can actually improve the performance of the operation. However, the industry must understand the energy saving potential to consider the cost and benefits and find the most suitable energy-saving method. It can be seen that there is still a lack of a scientific and effective air conditioning system section (4) method in the industry. The quasi-disc is used to find the power saved and the potential for energy saving is different, so that the best energy-saving methods can be widely adopted. SUMMARY OF THE INVENTION [The object of the present invention is to provide an energy-saving verification method for an air-conditioning system capable of accurately determining the saved power: calculating the energy-saving potential. * Another object of the present invention is to provide a U that can accurately determine the savings. Air conditioning (4) energy-saving verification module β. Then 'the air conditioning system energy-saving verification method of the present invention, including the following steps to improve the month' collection of energy-saving pre-variables and strain numbers. Equation, #1) according to step (A) The data of the set establishes a data of the first power consumption mode over-screening 1=—: the electric mode screening error is too large, and then the total (power consumption) is increased, and the power consumption before energy saving is obtained as the respective variables. For the coordinate axis, the edge system - the distribution of the data before the energy saving = Lang can improve the 'self-variable and the number of strains after the energy saving. Fill the energy-saving self-variables collected in step (D) into the distribution 200839204, and The data distribution range after energy saving must cover the data before energy saving. The above steps (D) and (E) can also be changed to active drive independent variables, and the distribution range of data from the data collection to the energy saving in the scatter diagram covers the pre-energy saving. According to the energy-saving data, retraining to obtain a second consumption mode, and pre-energy-saving data (inclusive data) into the second power consumption mode to calculate the total power consumption The quantity is used to obtain the energy consumption after energy saving. / (G) Calculate the energy saving effect based on the power consumption before energy saving and the power consumption after energy saving. The /it method can be applied according to the convenience of actually collecting data, for example,珂 When the collected data can cover the post-energy-saving data, that is, when the condition obtained in step (E) is reversed, step (F) is to substitute the energy-saving data (in the data included) into the first power consumption mode ( The power consumption before the energy saving calculation calculates the power consumption before the energy saving; the step (G) calculates the energy saving effect according to the power consumption before the energy saving and the actual power consumption collected in the step D, and the energy consumption after the energy saving. The foregoing and other technical contents, features, and advantages of the present invention will be apparent from the following detailed description of the preferred embodiments of the referenced drawings. Before the present invention is described in detail, note In the following description, like elements are denoted by the same reference numerals. As shown in Fig. 1, in an air conditioning system 100 having an ice water host, the ice water main unit 11 sends out ice water of a preset outlet temperature. Passed to the air-conditioning box, ice water 200839204 "to the inside of the T hot father after the water temperature increase" and the heat back to the ice water host π then the ice water host u will increase the water temperature to the cooling tower η for heat Exchange, get P preparation, La Xi, people owing nt j + / orphan cold water and then sent back to the ice water host 11. Air conditioning system 10 0 energy-saving, readable private through the ice water host 11, or cooling tower 13, cooling water pump 131, ice skin, go! 0! ^ 尺果浦121 for replacement or repair to improve operational efficiency b, and then energy saving, bite use too much before using this wood to ask people to ask The Republic of China No. 095136001 Shen Shi Dongan ", Yu Shenming wood ice water host minimum power consumption ice water outlet temperature design method" to save energy. The present invention can be used for any energy saving improvement method, and is not limited to the empty (four) system (10) type method described in the embodiment. The first preferred embodiment of the present invention is directed to the improvement of the efficiency of the air conditioning system 〇〇 〇〇 〇〇 主机 主机 ( ( ( 勤 勤 勤 勤 勤 勤 勤 勤 勤 勤 勤 勤 勤 勤 勤 勤 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 The main power consumption of the H water main unit 11 is tested; the fine skin, the four τ 〇 水 water Tcwr, the ice water effluent temperature Tchws, and the freezing capacity (with a negative interception rate pT R ^ ^ 贞戟 sheep PLR representative , pLR = actual load / rated capacity

) The two parameters are related, the relationship is nonlinear; if you want tb_ A right! Compared with the seasons or verifying the energy saving, the power consumption must be built: Litai-will go to the left-_ 乂 shoulder is built on the same basis, but the parameters

Ik changes the weather pattern and load size at any time, „ L j入夂I finds the parameters after the improvement. “The same is not easy, and even if you find a waiter, it may be transient data.” It is not suitable for long time. Power consumption status. - Thus, the present invention provides a method of verifying by establishing a quasi-rotation power consumption mode. As shown in FIG. 2, the steps of this embodiment are as follows: Step 21 - Collect the self-variable and the number of strains before energy saving. In this embodiment, the independent variables include "cooling water return water temperature" and "ice water outlet temperature", of course, "200839204 two-load", which is also one of the variables; the strain number is "host power consumption". The number of methods, for example, the frequency of the _ minute record _ group data, the energy saving " days data, if there is a frequency conversion equipment, the record is recorded in a set of 5 seconds of data, and then into a _ minute _ pen average value. Of course, the force rate and time of the data are not limited to this, and can be adjusted according to the actual situation, or calculated by “average”. The next steps 22 to 24 are to find the most accurate power consumption mode that is most suitable for the air conditioning system. Step 22 - Establish various power consumption modes according to the data collected in step 21. In this embodiment, a cubic function (such as Equation 1), a triple function (such as Equation 2), an ASHRAE standard pattern (such as Equation 3), and a back-propagation artificial neurai are established. The power consumption mode (hereinafter referred to as ANN mode) is established on the basis of the network. However, the present invention can be applied to any power consumption mode, and is not limited to the above mode. Kw = k^k, X^- k2Y + k, XY + k4X2 + k5Y2 + keX2Y + knXY2 + k, X3 + k9Y3 ............ [Formula 1]

Y = PLR γ _Tcwr - Tch — kw^k.+kX + kJ + kZ + k.XY + k.XZ^kJZ + h.X1 +k,Y2+k9Z2 + kl0X2Y^kuX2Z + KnXY2-^knY2Z^kuXZ2 + ^16x3 +^17f3 +knz3

+ kl9XYZ [Formula 2]

Z = PLR r = 10 200839204 [Formula 3] where Χ = r = uz = (freezing tons) As for the ANN mode, it is equivalent to a black box, and for the layer 1 network, it includes an input layer (/=〇) , multiple hidden layers (/= 1~L4), and one output layer (/=L), a total of L+1 layers; the number of neurons in each layer is different, expressed as (/= 0~L), in this implementation For example, the input layer includes three neurons: cooling water temperature Tcwr, ice water outlet temperature Tchws, and freezing capacity (plr); and the output layer contains a neuron··power consumption kW. The ANN mode is adapted to the input data after the training and the weighting value learning training, by adjusting the weighting values of the layers, so that the error between the corresponding output value and the target value is less than the preset value; if the error is too large, the adjustment and training are performed. Therefore, the 忒杈 is trained to predict the correspondence between the self-variable and the strain number. 0 The parameters required to construct the ANN mode are as follows: The data vector of the k-th input data} (9) = Bu (5),,, (4), also It is composed of several independent variables. The target vector of the first input data is ice) = μ (five) eight (10), which is the number of strains. The output value of the ρth neuron of the first layer is #(9). The total input value of entering the /1/th layer*q neurons is ten (five). (4) μ $ The weight value to enter the qth neuron of the /1/th layer is the threshold value of the qth neuron of the Kebei丨疋δhai/+1st layer 〇η neurons, corresponding The target value of the kth input data output at the output layer 200839204 - is dn(k). The target value is the actual output value corresponding to the kth input data. The step of establishing the reverse pass ANN module is as follows (not shown): Step 221 - setting the initial value of the weighted value and the threshold. This embodiment is randomly set to a minimum value. Step 222 - Enter the data vector and the target vector. Step 223 - Perform forward signal transmission to obtain output values of each layer: 4/+1) (k) = sgm{w^l)(t) + ^{k))........... ............[Formula 4] where 1/1+,; q=1.....H1+1; / = 〇, 1,···, L - l〇 Step 224 - Calculate the error signal occupancy of the hidden layer and the output layer by using Equation 5.

Hl+1 ^p\k) = ...........................[Formula 5] q=\ i Step 225 - After t times After the iterative operation, the t+1th weighted value is Equation 6 〇 = <,,) + On C) (,)..................... ............[Formula 6] where α is the momentum coefficient (〇< ^ < 1), and the threshold becomes: + = + must < +1) (7); where Δ&lt ;+1)(,) = <+1)(8) , and = public), P=1,···, Hi,q=l,···, Hi+i, / = L_1, • · · , Q 〇Step 226~ Repeat steps 222~225 until each pen wheel is invested 12 200839204 - the error of the output data after passing and learning is less than a preset positive real value ε, and the formula is as shown in Equation 7; The transmission-like neural network power consumption mode is established. ZKW-^W]2<^.......................................[Form 7 】 n=l Step 23 — Calculate the determination coefficient (R2 ) of each power consumption mode and the error value of each group of data; the determination coefficient can be used to judge the accuracy of the air conditioning system or the measurement equipment, if the determination coefficient is too low , indicating that there is a problem with the collected data, and must proceed to the system and equipment calibration. The formula for determining the coefficient is as follows:

R2= SSR/SST

Among them, SST= SSR+SSE 2 2 view? = ; £ [estimate 値 - sample mean 値]; view: mystery [sample 値 an estimate 値] z = l / = 1 and the error value =

dM Step 24—Select the power consumption mode. In this step, the determination coefficient or the error value is used as the basis for selecting the power consumption mode; generally, the power consumption mode in which the judgment coefficient is the southernmost, or the error value is the smallest, and the basic example is to select the neural network. The neural network power consumption mode is taken as an example. The power consumption mode determined in this step is defined as the first power consumption mode, that is, the power consumption mode before energy saving. Step 2 5 - Filter the data. Since there are times when the variable suddenly changes drastically 13 200839204 夂' The data for the first consumption must be removed, so the data with higher error value in the electrical mode after the power consumption mode is selected is removed. vStep 26 - Calculate the total power consumption. The total power consumption of the filtered data is added to the total power consumption, which is the “power consumption before energy saving”. Step 27 - Draw a scatter plot with the respective variables as the coordinate axes. Since the local application has a self-variable of two (because the ice water outlet temperature is usually set to a fixed value), the sample material is shown in the scatter diagram shown in Figure 3 (hereinafter referred to as the map), and the right second independent variable is not fixed. , this step should be based on χ ζ ζ scatter plot, or draw ΧΥ and ΖΥ scatter plot. Here, 代表 represents Ding, PLR, and

Tchws. Continued steps 28~33 I force Shishiduo μ » J疋 is carried out after the energy saving improvement of the workers, that is, the ice water host 11 (Figure η performance is improved. Step 28 - Collect energy-saving self-variables and strains The data collection method is similar to the step ^, but it takes a long time, for example, this embodiment is 1 day. Step 29 - Fill in the data collected in step 28 into the figure of step 27. v Step 30 - View the data after energy saving Whether the distribution range in the χ γ graph is the data before the 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 The distribution of post-data in the χγ map should be able to cover the data before energy saving. If the data before energy saving cannot be covered, return to the step and continue to collect until the range can cover the data before energy saving. Yes, the time for collecting data in the present invention is not necessarily #14 200839204: After the energy saving of the gem~fm, the long-term cage is concentrated, and the focus is on the distribution of the data in the χγ map must cover the knife of another period of time. Cloth, in this way, can avoid the use of extrapolation to make the model 1 accurate enough. In addition, the time before and after the energy saving, the season can not be too different. Retraining is defined as the second power consumption... 3i—According to the above-mentioned energy-saving data neural network power consumption mode (using steps such as ~ coffee), 杈, that is, power-saving mode after energy saving. 2nd step 2 - check whether data needs to be filtered. After the energy saving is collected, the input side is the power consumption mode, and the error between the output value of each group of data and the actual power consumption is calculated. Check whether there is, :: Big data. If there is data with excessive error, proceed to step ~, step 2, step 〇 Φ, then step 33 - screen out the data of the error society, and return to step 3q to pay attention to whether the data distribution range covers the data before energy saving. Step 34 - Before the energy saving screen filtered by step 25, the "self-variable input second power consumption mode, and the corresponding strain number - power consumption" is obtained according to the total power consumption, which is the "power consumption before the energy saving" Add the same self Under the condition of "energy-saving power consumption" for comparison, in the phase step 35 - calculate the energy-saving effect, the formula is as follows. [(Step 26: Energy consumption before energy saving) 1 (Step 34 & v Power )]+ (Step 26: Energy consumption before energy saving) It can be used later ~ ............···[8]

In detail, in the embodiment, the data before the energy saving in the scatter diagram is in the "inside" of the data range after the energy saving, so the self-variable data 15 200839204 can be brought into the second power consumption mode (step M, find the energy consumption and quantity after energy saving, and (the same self-variable number) the actual energy-saving pre-consumption power consumption is the same as the energy-saving New Zealand; otherwise, if the energy-saving data is in the scope of the scatter map, the section covers the section; Then, the energy-saving self-variable data (data covered) should be brought into the first-power consumption mode to calculate the power consumption before energy saving, and calculate the energy saving together with the energy consumption after energy saving (corresponding to the same self-variable) The result is that the principle is to use the "inside" #self-variable data to enter another mode, and then compare it with the actual power consumption. 'The same is the same self-variable (representing the same weather "and load conditions and - segment Time (four) material) input power consumption mode, the condition is the same and non-transitory data, so it is representative. As shown in Figure 4, 'the second preferred embodiment of the present invention is applicable to the f-controlable f month; brother' Conduct an active verification method. The second preferred embodiment differs from the first preferred embodiment only in steps 28 and 29, and is replaced by the following step 28', and the remaining steps are the same. Step 28' drives the self-variable and fills in the χγ scatter plot, and then proceeds. Step 30, until the data is collected until the distribution range of the data in the χγ map covers the data range before the energy saving. The second preferred embodiment of the present invention is directed to the cooling tower 丨3 or the cooling water pump 131. The energy saving effect is verified, and the difference from the first preferred embodiment is that the self-variable is sequentially changed into the cooling water outlet temperature τ^, the external gas wet bulb temperature WBT, and the cooling tower load pLR, and the strain number is the cooling water tower. Power consumption or cooling water pump power consumption. The fourth preferred embodiment of the present invention verifies the energy saving effect of the air conditioner 12 or the ice water pump 121 to improve the efficiency, and the difference from the first preferred embodiment is 16 200839204 The difference is only that the self-variable is sequentially changed into the air wet bulb temperature WBT, the ice water outlet temperature Tchws, and the ice water load PLR, and the strain number is the air conditioner box power consumption or the ice water pump power consumption. The fifth preferred embodiment of the present invention verifies the energy saving effect of the optimization operation of the ice water host, and the difference from the first preferred embodiment is that the self-variable is sequentially changed to the total return water temperature of the cooling water, and the ice. The total water discharge temperature and the total system load, the power consumption is the sum of the power consumption of each host. The sixth preferred embodiment of the present invention verifies the effect of the cooling water temperature optimization operation, and the fifth comparison The difference between the preferred embodiments is that the self-variable is sequentially changed to the total return water temperature of the cooling water, the external wet bulb temperature and the total system load, and the power consumption is the sum of the power consumption of the main unit, the cooling water pump and the cooling tower. In this embodiment, the glaciers that are turned on before and after the optimization operation must be identical. The seventh preferred embodiment of the present invention verifies the energy-saving effect of the ice water temperature optimization operation, compared with the fifth The difference between the preferred embodiments is that the self-variables are sequentially changed into the air-conditioning box cooling plate camp. #,目. 祁7丨 sparse & empty milk wet bulb temperature, ice water total water temperature and total system load, and consumption Power 丨 Ribe] for the host, The sum of the power consumption of the ice pump and the air conditioner. This embodiment # J must be consistent in the operation of the ice water host before and after the operation. In summary, the present invention can be applied to the ice machine main unit or cooling tower, cooling m ice pump performance improvement, or ice water host, cooling water temperature, ice water temperature optimization operation, etc. To verify the effectiveness, first set the self-change and strain parameters, establish various types of n+n & ethics, and make a "mode selection" to confirm the mode that is most suitable for the system. Avoid 17 200839204 mode accuracy is disturbed, and the first use of the "edge scatter map" technology to ensure that the calculated power consumption is not extrapolated, to ensure accuracy, in terms of computing, because the energy calculation is calculated under the same independent variables The power consumption before and after energy saving is the same as the conditional basis and is not (4)#, so it is representative and can achieve the object of the present invention. σ 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 Variations and modifications are still within the scope of the invention. [Simple description of the diagram] The architecture of the air-conditioning system with the ice water host is shown in the first section. The flowchart of the air-conditioning system of the present invention illustrates the effectiveness of the method for the ice-water host single-machine bait early machine. The step of performing energy saving verification; FIG. 3 is an independent variable scatter diagram; and FIG. 4 is a flowchart step. The implementation of the second preferred embodiment of the present invention 18 200839204 [Explanation of main component symbols] 100... Air conditioning system 13 ... Cooling water tower 11 ... • Ice water host 21 to 35 · ... Step 12... ...air conditioner 28,··...step 19

Claims (1)

200839204 Patent application scope: The seeding system can verify the square r λ, ^ knife to go, including the following steps · (A) Lang can improve before, W "Gang Ji Lang can pre-self-variable and strain number ^ ^ (Α) The data establishment mode of the wasteland; after the power consumption (c) can be improved, Chu (9) will step (A), = variables and strain numbers, and teach the energy-saving before and after the self-change map, and ^1 The number of texts is the distribution of the green axis of the coordinate axis.... Talk about the data distribution range 刖 data, formula and r # ,, 丄 皿 即 即 b b b energy saving number: ..., the data distribution range must be covered (8) according to the energy saving Data, re-power consumption mode; U Shiyidi (F) will add power in the covered power, frost furnace A A Ah,, 丨 丨 丨 X X 侍 侍 耵 耵 耵 耵 耵 耵And the self-contained helmets are also in the bag, and the power consumption of the stand is based on the self-variables; and the power consumption (G) is based on the energy consumption before the energy consumption (G) In the electric bag and the energy consumption after the energy saving effect. 4 different 2 · according to the scope of the patent application item i / /, is 彡 + # , &quo t; industrial system energy-saving verification method 疋 for the ice water host single-machine performance, including the cooling water return water temperature production, and the energy-saving verification, and become the host power consumption. Ruler, dish and load 'strain number 20 200839204 =:! The air conditioning system energy saving verification method described in item 1 is energy saving verification of the white water tower or cooling water pump efficiency improvement, and the independent variables include the cooling water outlet temperature, the external air wet bulb temperature, and the α = tower load. 'The number of strains is the power consumption of the cooling tower or the cooling method of the fruit system according to the item i of the scope of patent application'. The energy saving verification method for the air conditioner or the ice pump is improved. The variables include the air wet bulb temperature into the air conditioning box cooling coil, the ice water output = the amount and the ice water load, and the strain number is the air conditioner box power consumption or the ice water chestnut 5. (4) The air conditioner described in the third paragraph of the patent scope The system energy-saving verification method is to verify the energy-saving of the ice water host optimization operation, and the self-variable is the total return water temperature of the cold water, the total water output temperature of the ice water, the total load of m, and the total number of strains is the total power consumption of each host. 6. According to the air conditioning system energy-saving verification method described in the third paragraph of the patent scope, the energy-saving effect of the cooling water temperature optimization operation is verified, and the k-number is the total return water temperature of the cooling water, and the external air is wet. The ball temperature and the total system load, the number of strains is the sum of the power consumption of the main engine, the cooling water pump and the cooling water tower; and the ice water main unit that is turned on before and after the optimization operation is consistent. The energy-saving verification method for the air-conditioning system described in the first item is to verify the energy-saving effect of the ice water temperature optimization operation, and the self-intersection and the number of intersections are changed to the air wet bulb temperature of the air-conditioning box cooling coil, and the total water outlet temperature of the ice water. And the total system load, the number of strain is the sum of the power consumption of the main engine, the ice water pump and the air conditioning box; and the operation is started before/after the optimization operation. 200839204 Ice water host's. 8. According to the scope of claim 1 of the patent application, the second method of energy saving verification method i (B) includes: step (B1) establishing various power consumption modes according to the data set in step (A); (B2) Calculate the (four) coefficient of each power consumption mode and the error value of each group of data; and the step (called the judgment coefficient or the error value according to 'select power consumption mode, one power consumption mode. 9. According to the patent application scope 8 The air conditioning system energy-saving verification method described in the item, wherein the step (B includes establishing a χ γ cubic function, a χ γ ζ cubic function, an ASHRAE standard 桓 formula, Fen 丨, and 丨 & & 杈 及 及 及 及 及Based on the neural network, a power consumption mode (hereinafter referred to as "ΑΝΝ mode") is established. Π). According to the invention, the air conditioning system energy-saving verification method described in item 9 of the patent scope, wherein the step (5) establishes the ΑΝΝ mode includes: steps ( Β11) set the initial value and threshold of the weighting value; step (βι2) input data vector and target vector; step (Β13) perform forward signal transmission to obtain output values of each layer, step (Β14) calculate hidden And the error signal of the output layer, step (Β15) after t times of iterations to obtain the t+1th weighting value; step (BIO repeats the steps (B12)~(B15) until each input data is passed And the error of the output data after learning is less than the positive real value of the pre-discrimination ^, then the power transmission mode of the inverted transmission type neural network is established. 11. According to the scope of the patent application, wherein the step (D is filtered) In the air conditioning system energy saving verification method described in the item, the self-variables of the scattergram are filled in, and the filtering method is to remove the data with higher error value for the first power consumption module 22 200839204 12. The method for verifying the air conditioner according to claim 11 of the patent application scope, wherein the data screening method is calculated by calculating the data of the parent group by the first power consumption mode. The error between the output value and the actual power consumption, and the data with large error is removed. 13. The method according to the scope of the patent application, the method for verifying the energy, wherein the step (C) Collecting energy-saving Demu 〗 〖Personal data time is long Tang is different from the step (A) to collect data before energy saving. 14) According to the scope of application patent scope 丨 . . 糸 能 能 能 能 能 能 能 能 能 能 能 能 能 能 能 能 能 能 能After collecting the data, the same season.] 疋 is similar or phase 15. According to the scope of the patent application scope, the system can verify the method, /, in the special variable has three, respectively , γ, Z This step (D) is to draw χ γ ζ 3D 4, the table is therefore a scatter diagram. 'The layout of the layout, or the edge of the ΧΥ and 16. According to the scope of the patent application, Α 4 work with the energy saving verification The method, wherein the self-variables have -, 闵μ /2r double another 因此 ΧΥ 因此 该 该 该 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。.疋,,,曰1 2 17·A kind of energy-saving system energy-saving verification method, with α ^ person, with ice water host ★ withered system, including the following steps · · J two (1) energy-saving improvement before '1# energy-saving before Variable ❹ Add the actual power consumption to the total power consumption before energy saving; u and (b) according to the step () ; mode; establish a first power consumption 23 200839204 (c) draw a pre-energy data with their respective variables as the coordinate axis (d) After the energy saving is improved, the self-variable is driven, and the data collected to the energy-saving data in the scatter map covers the data range before the energy saving; (e) retraining based on the energy-saving data Obtaining a second power consumption mode, and calculating the total power consumption by substituting the pre-energy consumption data into the second power consumption mode, and obtaining the power consumption after the energy saving; (f) according to the power consumption before the energy saving and the power consumption after the energy saving, Calculate energy savings. twenty four