CN113158309B - Heating and ventilation equipment operation strategy identification method - Google Patents

Abstract

The invention provides a heating and ventilation equipment operation strategy identification method, which comprises the following steps: s110, generating probability density distribution of single equipment power; s120, generating probability density distribution of the power of the equipment group; s130, identifying a heating and ventilation equipment strategy of the power line; s140, self-correcting the strategy identification parameters. The invention solves the problem of operation strategy identification of heating and ventilation equipment under the condition that the existing electric power monitoring platform is provided with a plurality of pieces of equipment in a single line.

Description

A method for identifying operation strategies of HVAC equipment

technical field

The invention relates to the field of building energy management, and more particularly, to a method for identifying an operation strategy of HVAC equipment.

Background technique

The application and innovation of the algorithm integrating building data and intelligent data mainly focus on improving the comfort of the building environment and reducing the energy consumption of building operation. point. The research on BAS system data focuses on the use of internal data of building equipment, the research and discussion on the correlation of each data in the BAS data, and the prediction of building load or energy consumption. These research and applications can improve equipment efficiency and diagnose equipment failures. and retrofit system forms to indirectly improve the comfort of the building environment or reduce building energy consumption.

However, the optimization method of diagnostic equipment requires building data of multiple dimensions. Since the equipment parameters of different buildings are different or missing, at present, related methods may produce results with low robustness and low accuracy, and the optimization in the form of transformation system The method leads to higher retrofit costs.

Therefore, there is an urgent need in the prior art to use the existing building data to mathematically study the identification of HVAC operation equipment and equipment groups and identify the current HVAC operation strategy, in order to integrate intelligent data algorithms. Guide technical solutions to improve the comfort of the building environment and improve the performance of the building.

SUMMARY OF THE INVENTION

In view of the above problems, the present invention proposes a method for identifying the operation strategy of HVAC equipment.

The technical scheme adopted by the present invention for solving the above-mentioned technical problems is as follows:

A method for identifying operation strategies of HVAC equipment, comprising the following steps:

S110. Generate a probability density distribution of the power of a single device;

S120. Generate a probability density distribution of the power of the equipment group;

S130. Identify HVAC equipment strategies for power lines;

S140, the strategy identification parameters are self-calibrated.

In step S110, the power of the single device includes the power of the fixed-frequency device and the power of the variable-frequency device;

The probability density distribution function of fixed frequency equipment power adopts Gaussian distribution;

Frequency conversion equipment includes fixed frequency equipment and non-fixed frequency equipment;

Among them, the probability density distribution function of the power of quasi-fixed-frequency equipment also adopts Gaussian distribution; the probability density distribution of the power of non-quasi-fixed-frequency equipment adopts a statistical method. The probability density distribution function of power.

Step S120 includes the following sub-steps:

S121. Computing equipment group:

Calculate all possible device startup combinations, where, for a device group containing n different devices, there are theoretically 2 ⁿ different device startup combinations;

S122. Calculate the probability density distribution of each equipment group:

For a device group containing n different devices, the power-on power probability density distribution function of device n is denoted as p _n (x), and the probability density distribution function of the device group is denoted as p _1,2,3… (x), where, 1, 2, 3, ..., n represents the device 1, device 2, device 3, ..., device n in the ON state in the device group; p represents the probability density function, and x represents the given power value;

The power of the device group is accumulated by the power of the devices that are currently turned on. The probability density of the power of the device group is calculated as follows:

where x ₁ +x ₂ +x ₃ +...+x _n =x;

S123, convolution calculation:

The device group is decomposed into a device-on combination containing 2 devices and a device-on combination containing n-2 devices; among them, the probability density of the device-on combination containing 2 devices is calculated as follows:

The device startup combination containing 2 devices is regarded as a new device. Therefore, the device startup combination containing n devices is converted into a device startup combination containing n-2+1 devices. Through the convolution operation, the device startup combination containing n The probability density calculation of the power of the device-on combination of each device is converted into the probability density calculation of the power of the device-on combination of n-1 devices; the convolution operation is repeated to realize the probability density calculation of the power of n device combinations; for A device startup combination containing n devices requires n-1 convolution operations;

S124. Calculate the probability density function of the power of the equipment group:

Perform multiple convolution calculations for 2 ⁿ different device turn-on combinations to obtain the probability density function of the power of the device group.

Step S130 adopts the Bayesian discriminant method:

Suppose that each device opening combination is denoted by C, and C _i represents the i-th device combination mode. For a device group containing n different devices, there are 2 ⁿ device opening combinations, C ₁ , C ₂ , C ₃ , ..., C ₂ ⁿ represents the 1, 2, 3, … 2 ^nth device turn-on combinations, given X represents the specific power value of the device group, predict that X belongs to the class with the largest posterior probability under the condition X, using Bayesian classification method predicts that X belongs to class C _i if and only if:

P(C _i |X)＞P(C _j |X)1≤i,j≤2 ⁿ ,i≠j

Maximizing P(C _i |X), the class C _i that maximizes P(C _i |X) is called the maximum a posteriori hypothesis, according to Bayes' theorem:

P(X) is constant for all classes, only need to maximize P(X|C _i )P(C _i ), if the prior probability of all classes is unknown, it is assumed that these classes are equal probability, namely P( C ₁ )=P(C ₂ )=...=P(C ₂ ⁿ ); otherwise maximize P(X|C _i )P(C _i ).

Step S140 includes the following sub-steps:

S141: Given the rated power of a certain equipment group;

S142: Obtain the device startup combination with the highest probability according to the probability density distribution of the power of the device group;

S143: Recalculate the power of the device to turn on the combination;

S144: Determine whether the power of the device startup combination calculated in step S143 is equal to the rated power of the device group given in step S141, if not, then go back to step S141, re-specify the rated power of the device group, and correct it at the same time The rated power of the equipment group; if it is equal, go to step S145;

S145: Correct the Gaussian distribution parameters of the equipment group.

Step S141 includes: assuming that the expected μ of the equipment group power is the rated power of the equipment, and assuming that the standard deviation δ is 10% of the rated power.

Step S144 includes: using multiple convolutions to calculate the probability distribution of each device combination, and finally obtaining a start-stop matrix B _m×n of the devices under a given power X _m×1 , where X _m×1 represents m actual line power values , which can be regarded as the line power at different times, and B _m×n represents the start and stop conditions of n line equipment corresponding to m actual line power values; the rated power of the calibration equipment group is calculated by the following formula:

P'=B ^-1 X

Among them, B ^-1 is the pseudo-inverse of B, that is, the least squares method is used to calculate the equipment power parameter P that satisfies the power sum of X; when the difference between the corrected power parameter P' and the given rated power parameter P is less than the iterative calculation requirement When the minimum value, the currently given rated power parameter P is approximately equal to the actual power, that is, it is determined that the power of the device startup combination calculated in step S143 is the rated power of the device group; when the difference between the two is relatively large, then P ' As the rated power of the equipment group initially given in step S141, continue the strategy identification self-correction calculation process.

The beneficial effects of the present invention compared with the prior art are:

1. Design the typical probability density distribution model of common equipment, use convolution to calculate the probability density distribution of the power of the equipment group composed of multiple equipment, and solve the problem of HVAC equipment when the existing power monitoring platform has multiple equipment under a single line The operational strategy identification problem.

2. The strategy identification parameter self-calibration method can find the equipment power parameter group that meets the actual power consumption performance through an iterative update method given the initial equipment power parameter group, thereby improving the equipment operation strategy identification effect.

Description of drawings

Fig. 1 is the principle frame diagram of the operation strategy identification method of HVAC equipment.

FIG. 2 is a schematic flowchart of step S120 of the HVAC equipment operation strategy identification method.

FIG. 3 is a schematic flowchart of step S140 of the HVAC equipment operation strategy identification method.

Detailed ways

The technical solutions of the present invention will be clearly and completely described and explained below with reference to specific embodiments, but this is not intended to limit the protection scope of the present application.

As shown in Figure 1-3, the HVAC equipment operation strategy identification method includes the following steps:

S110. Generate a probability density distribution of the power of a single device;

S120. Generate a probability density distribution of the power of the equipment group;

S130. Identify HVAC equipment strategies for power lines;

S140, the strategy identification parameters are self-calibrated.

In step S110, the power of the single device includes the power of the fixed-frequency device and the power of the variable-frequency device;

The probability density distribution function of fixed frequency equipment power adopts Gaussian distribution;

Frequency conversion equipment includes fixed frequency equipment and non-fixed frequency equipment;

Among them, the probability density distribution function of the power of quasi-fixed-frequency equipment also adopts Gaussian distribution; the probability density distribution of the power of non-quasi-fixed-frequency equipment adopts a statistical method. The probability density distribution function of power.

The expression for the Gaussian distribution is as follows:

Among them, μ is the average power value of the device, and σ is the standard deviation.

In the frequency conversion equipment, the frequency conversion pump frequency is artificially set to 45~47Hz frequency conversion equipment, which has the properties similar to the fixed frequency equipment.

Other frequency conversion equipment is non-type fixed-frequency equipment. The probability density distribution of non-type fixed-frequency equipment power adopts statistical method, and the statistical power probability density distribution function of equipment is obtained by counting the actual distribution of power within a certain period of time. However, in the case that the non-classical fixed-frequency equipment operates exactly according to the power meter when the device is turned on, the probability density of the non-classical fixed-frequency device power is fitted to the actual distribution curve through Gaussian distribution.

Step S120 includes the following sub-steps:

S121. Computing equipment group:

Calculate all possible device startup combinations, where, for a device group containing n different devices, there are theoretically 2 ⁿ different device startup combinations;

S122. Calculate the probability density distribution of each equipment group:

For a device group containing n different devices, the power-on power probability density distribution function of device n is denoted as p _n (x), and the probability density distribution function of the device group is denoted as p _1,2,3… (x), where, 1, 2, 3, ..., n represents the device 1, device 2, device 3, ..., device n in the ON state in the device group; p represents the probability density function, and x represents the given power value;

The power of the device group is accumulated by the power of the devices that are currently turned on. The probability density of the power of the device group is calculated as follows:

where x ₁ +x ₂ +x ₃ +...+x _n =x;

S123, convolution calculation:

The device group is decomposed into a device-on combination containing 2 devices and a device-on combination containing n-2 devices; among them, the probability density of the device-on combination containing 2 devices is calculated as follows:

The device startup combination containing 2 devices is regarded as a new device. Therefore, the device startup combination containing n devices is converted into a device startup combination containing n-2+1 devices. Through the convolution operation, the device startup combination containing n The probability density calculation of the power of the device-on combination of each device is converted into the probability density calculation of the power of the device-on combination of n-1 devices; the convolution operation is repeated to realize the probability density calculation of the power of n device combinations; for A device startup combination containing n devices requires n-1 convolution operations;

S124. Calculate the probability density function of the power of the equipment group:

Perform multiple convolution calculations for 2 ⁿ different device turn-on combinations to obtain the probability density function of the power of the device group.

Among them, when the equipment group is empty, the probability density distribution of the empty combination is obviously as follows:

However, there may be some interference sources in the line, resulting in a small amount of power consumption even when the known equipment is fully turned off. In order to improve the stability of the calculation process, the probability density distribution of the null combination can be set to an average probability density distribution near 0.

Step S130 adopts the Bayesian discriminant method:

Suppose that each device opening combination is denoted by C, and C _i represents the i-th device combination mode. For a device group containing n different devices, there are 2 ⁿ device opening combinations, C ₁ , C ₂ , C ₃ , ..., C ₂ ⁿ represents the 1, 2, 3, … 2 ^nth device turn-on combinations, given X represents the specific power value of the device group, predict that X belongs to the class with the largest posterior probability under the condition X, using Bayesian classification method predicts that X belongs to class C _i if and only if:

P(C _i |X)＞P(C _j |X)1≤i,j≤2 ⁿ ,i≠j

Maximizing P(C _i |X), the class C _i that maximizes P(C _i |X) is called the maximum a posteriori hypothesis, according to Bayes' theorem:

P(X) is constant for all classes, only need to maximize P(X|C _i )P(C _i ), if the prior probability of all classes is unknown, then these classes are assumed to be equal probability, namely P( C ₁ )=P(C ₂ )=...=P(C ₂ ⁿ ); otherwise maximize P(X|C _i )P(C _i ).

Step S140 includes the following sub-steps:

S141: The rated power of a certain equipment group is given. Step S141 further includes: assuming that the expected μ of the power of the equipment group is the rated power of the equipment, and the standard deviation σ is assumed to be 10% of the rated power.

S142: Obtain a device startup combination with the highest probability according to the probability density distribution of the power of the device group.

S143: Recalculate the power of the device to turn on the combination.

S144: Determine whether the power of the device startup combination calculated in step S143 is equal to the rated power of the device group given in step S141, if not, then go back to step S141, re-specify the rated power of the device group, and correct it at the same time The rated power of the equipment group; if it is equal, go to step S145.

Step S144 includes: using multiple convolutions to calculate the probability distribution of each device combination, and finally obtaining a start-stop matrix B _m×n of the devices under a given power X _m×1 , where X _m×1 represents m actual line power values , which can be regarded as the line power at different times, and B _m×n represents the start and stop conditions of n line equipment corresponding to m actual line power values; the rated power of the calibration equipment group is calculated by the following formula:

P'=B ^-1 X

Among them, B ^-1 is the pseudo-inverse of B, that is, the least squares method is used to calculate the equipment power parameter P that satisfies the power sum of X; when the difference between the corrected power parameter P' and the given rated power parameter P is less than the iterative calculation requirement When the minimum value, the currently given rated power parameter P is approximately equal to the actual power, that is, it is determined that the power of the device startup combination calculated in step S143 is the rated power of the device group; when the difference between the two is relatively large, then P ' As the rated power of the equipment group initially given in step S141, continue the strategy identification self-correction calculation process.

S145: Correct the Gaussian distribution parameters of the equipment group, where the Gaussian distribution parameters include the expected μ and the standard deviation σ of the equipment group power.

Among them, the two parameters of expectation μ and standard deviation σ are used as the parameters of Gaussian distribution in step S110 to obtain the probability density distribution function of the power of a single device; and then the convolution calculation in step S120 is performed to obtain the probability density distribution function of the power of the device group . In this way, through the selection of the probability distribution model of the equipment and the setting of the parameters, the preliminary strategy identification result of the HVAC equipment group can be obtained.

The purpose of the calibration calculation process is to make the probability distribution of the calibration power close to the distribution of the actual power, so as to provide more accurate basic information for the strategy identification and make the identification equipment start and stop strategy more accurate. The idea of the square method to find the optimal parameters.

Like numerals and letters refer to like items in the following figures, so once an item is defined in one figure, it does not require further definition and explanation in subsequent figures. Meanwhile, in the description of the present application, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

The above are only the preferred embodiments of the present invention, but the present invention is not limited to the above-mentioned specific embodiments. For those of ordinary skill in the art, without departing from the inventive concept of the present invention, the Several modifications and improvements are made, which all belong to the protection scope of the present invention.

Claims (5)

1. A heating and ventilation equipment operation strategy identification method is characterized by comprising the following steps:

s110, generating probability density distribution of single equipment power;

s120, generating probability density distribution of the power of the equipment group;

s130, identifying a heating and ventilation equipment strategy of the power line;

s140, self-correcting the strategy identification parameters;

wherein, step S120 includes the following substeps:

s121, computing equipment group:

calculate all possible device on combinations, where a device group containing n different devices theoretically has 2 ⁿ Different equipment opening combinations are planted;

s122, calculating the probability density distribution of each equipment group:

for a group of devices comprising n different devices, the probability density distribution function for the on-power of device n is denoted as p _n (x) The probability density distribution function of the group is denoted as p _1,2,3… (x) Wherein 1,2,3, …, n represents device 1, device 2, device 3, …, device n in the on state in the device group; p represents a probability density function, x represents a given power value;

the power of the equipment group is accumulated by the power of the equipment in the current on state, and the probability density of the power of the equipment group is calculated in the following mode:

wherein x is ₁ +x ₂ +x ₃ +...+x _n ＝x；

S123, convolution calculation:

decomposing the device group into a device on combination containing 2 devices and a device on combination containing n-2 devices; the probability density calculation mode of the device opening combination comprising 2 devices is as follows:

regarding the device on combination containing 2 devices as 1 new device, then converting the device on combination containing n devices into a device on combination containing n-2+1 devices, and converting the probability density calculation of the power of the device on combination containing n devices into the probability density calculation of the power of the device on combination containing n-1 devices through convolution operation; repeatedly performing convolution operation to realize probability density calculation of power of n equipment combinations; for a device on combination comprising n devices, n-1 convolution operations are required;

s124, calculating a probability density function of the power of the equipment group:

for 2 ⁿ Performing multiple convolution calculation on different equipment starting combinations respectively to obtain a probability density function of the power of the equipment group; and is

Wherein, step S140 includes the following substeps:

s141: given the rated power of a certain equipment group;

s142: obtaining the equipment opening combination with the maximum probability according to the probability density distribution of the equipment group power;

s143: recalculating the power of the device turn-on combination;

s144: judging whether the power of the device on combination calculated in step S143 is equal to the rated power of the device group given in step S141, if not, returning to step S141, re-giving the rated power of the device group, and correcting the rated power of the device group; if so, go to step S145;

s145: the gaussian distribution parameters of the set of devices are corrected.

2. The method for identifying an operation strategy of heating and ventilation equipment as claimed in claim 1, wherein in step S110, the single-equipment power comprises a fixed frequency equipment power and a variable frequency equipment power;

the probability density distribution function of the power of the fixed frequency equipment adopts Gaussian distribution;

the frequency conversion equipment comprises class frequency fixing equipment and non-class frequency fixing equipment;

wherein, the probability density distribution function of the class fixed frequency equipment power also adopts Gaussian distribution; the probability density distribution of the power of the non-class fixed frequency equipment adopts a statistical method, and the statistical probability density distribution function of the power of the equipment is obtained by counting the actual distribution condition of the power within a certain time.

3. The method for identifying the operation strategy of the heating and ventilation equipment as claimed in claim 1, wherein the step S130 adopts a bayesian discrimination method:

let each equipment open combination be denoted by C, C _i 2 in total for a group of devices including n different devices ⁿ Seed equipment opening combination, C ₁ ,C ₂ ,C ₃ ,…,C ₂ ⁿ Denotes the 1 st, 2 nd, 3 rd, … 2 nd ⁿ The equipment starting combination is set, given X represents the specific power value of the equipment group, the forecast X belongs to the class with the maximum posterior probability under the condition X, and the Bayesian classification method is adopted to forecast that X belongs to the class C _i And if and only if:

P(C _i |X)＞P(C _j |X)1≤i,j≤2 ⁿ ,i≠j

maximizing P (C) _i |X)，P(C _i | X) largest class C _i Called maximum a posteriori hypothesis, according to bayes' theorem:

p (X) is constant for all classes, only P (X | C) _i )P(C _i ) Maximization, i.e. if the prior probabilities of all classes are unknown, it is assumed that these classes are equiprobable, i.e. P (C) ₁ )＝P(C ₂ )＝…＝P(C ₂ ⁿ )；Otherwise, maximize P (X | C) _i )P(C _i )。

4. The method for identifying the operation strategy of the heating and ventilation equipment as claimed in claim 1, wherein the step S141 comprises: the expected μ for the device group power is assumed to be the device rated power and the standard deviation δ is assumed to be 10% of the rated power.

5. The method for identifying an operation strategy of heating and ventilation equipment as claimed in claim 1, wherein the step S144 comprises: calculating probability distribution of each equipment combination by utilizing multiple convolutions to finally obtain given power X _m×1 Starting and stopping matrix B of lower equipment _m×n Wherein X is _m×1 Representing m actual line power values, which can be regarded as line power at different times, B _m×n Representing the starting and stopping conditions of n line devices corresponding to the m actual line power values; the rated power of the correction device group is calculated by the following formula:

P'＝B ^-1 X

wherein B is ^-1 The pseudo inverse of B, namely, the least square method is used for calculating the power parameter P of the equipment which meets the power sum X; when the difference between the power parameter P' calculated by correction and the given rated power parameter P is smaller than the minimum value required by iterative calculation, the current given rated power parameter P is approximately equal to the actual power, that is, the power of the device opening combination calculated in step S143 is determined to be the rated power of the device group; if the difference between the two values is large, P' is used as the rated power of the device group initially specified in step S141, and the policy identification self-calibration calculation process is continued.

Images