1. A heating two-network balance method is characterized by comprising the following steps:

establishing a heat supply model of heat supply two-network balance, wherein the heat supply two-network balance comprises heat supply balance in a building and heat supply balance among building units;

dividing the building into a high area, a middle area and a low area according to the height of the building, and arranging a variable frequency pump at the starting floor of each high area;

constructing a heat supply balance neural network of each height area to obtain the frequency of the variable frequency pump, and activating a circulation control instruction according to the adjustable range of the variable frequency pump to achieve heat supply balance in the building;

and regularly acquiring parameters of each building unit in the heat supply range of each heat exchange station, constructing a heat supply two-network balance neural network, and adjusting the water supply flow to achieve heat supply two-network balance.

2. A heating network balancing method according to claim 1, wherein the heating model is G_r(E, V), wherein E represents directed graph G_rE ═ E₁,e₂,...,e_n+1N is the number of building units, and V represents a directed graph G_rSet of edges V ═ V₁,v₂,...,v_n}。

3. A heating distribution network balancing method according to claim 2, characterized in that said set of nodes E comprises a main node E_mAnd child node E_sMaster node E_mFor heat exchange stations, each sub-node E_sThe properties of the heat exchange station comprise water supply temperature, water return temperature and total valve opening degree; the properties of the building units comprise water supply temperature, water return temperature, heat supply area, frequency of a variable frequency pump, distance between the variable frequency pump and a heat exchange station and valve opening degree of a high region, a middle region and a low region of each building unit.

4. A heating supply two-network balancing method according to claim 1, characterized in that the input data of the zones of different heights are fitted, the optimum frequency of the variable frequency pump is output, the variable frequency pumps of the zones of different heights are set with a threshold value, and if the output optimum frequency is within an adjustable range, the variable frequency pumps are adjusted to the optimum frequency; otherwise, the output result is 0, the neural network is automatically activated, the valve opening with a certain span is increased, and new flow data f are obtained_dFitting the cyclic calculation again until the frequency R of the optimal variable frequency pump is output^j。

5. The heat supply two-network balancing method according to claim 4, wherein the height district heat supply balancing neural network comprises an input layer, a fitting layer, a selection layer and an output layer, wherein the input layer sends the processed data to the fitting layer, and the fitting layer sends the calculation result to the selection layerThe selection layer comprises two neurons connected in series, each neuron screens input data according to a filtering standard, and if the input data R are all met, the input data R are input^jSending to an output layer; otherwise, sending the empty set to an output layer; the output of the output layer isWherein R is^jIs the frequency of the variable frequency pump.

6. A heat supply two-network balancing method according to claim 5, characterized in that the calculation process of the fitting layer isWherein rho is fluid medium density, xi is resistance coefficient of the regulating valve, S is cross-sectional area of the pipeline, f is flow, xi' is resistance coefficient of the pipeline section, c₀、c₁、c₂As a parameter of the circulation pump head, K_rIs the ratio of the actual operating frequency to the rated operating frequency of the circulating pump, f₀Is the actual flow of the circulating pump.

7. The heat supply two-network balancing method according to claim 1, wherein the heat supply two-network balancing comprises calculating the total temperature of the heat exchange station two-network circulation, setting the water supply temperature based on the steady-state heat balance according to the outdoor temperature change and the indoor temperature requirement through a climate compensator, and calculating the water supply temperature and the water return temperature of the heat exchange station in a steady state according to the operation regulation of a heat supply system;

calculating the weight of each building unit for any heat exchange station and the coverage area thereof;

setting a minimum threshold Td of temperature difference between supply water and return water_εAnd a maximum flow threshold f_εWhen | Tg_d-Th_d|＜Td_εThen, according to the result of the neural network in the high region, a heat supply two-network balance neural network is constructed, and the temperature differences Td between the water supply and the water return of the n building units_dUsing time series as network input to obtain optimum flow value, wherein Tg_dFor the temperature of the water supply, Th_dFor returning waterAnd (3) temperature.

8. The heating two-network balancing method according to claim 7, wherein the heating two-network balancing neural network comprises an input layer, a mode layer, a summation layer and an output layer, wherein the input layer is connected with the mode layer and transfers input variables to the mode layer, and the neuron transfer function of the mode layer is a radial basis function:wherein the content of the first and second substances,for neuron-corresponding learning samples, σ²Is the variance of the input variable with its corresponding sample.

9. The method of claim 8, wherein the summation layer comprises a first type of neurons and a second type of neurons, the first type of neurons being the sum of the outputs of each mode layer neuron, the mode layer connection weight with each neuron being 1; the second class of neurons is a weighted sum of the expected result and each mode layer neuron.

10. A heating distribution network balancing method according to claim 9, wherein the first neuron is S₁，The second type of neuron is S₂,The flow value output by the heat supply two-network balance neural network is Y,and Y is more than or equal to f_εOtherwise, take f_ε。

Background

At present, a secondary pipe network hydraulic imbalance phenomenon generally exists in heat supply enterprises, the pipe network is high in power consumption, low in heat efficiency and high in operation cost, and the hydraulic imbalance easily causes uneven room temperature cold and hot and large complaint amount.

The traditional heat supply adjusting method generally needs repeated temperature measurement and repeated adjustment, and has the disadvantages of large workload, time and labor consumption and poor effect. The heat exchange station and the two-network equipment are independently adjusted, so that integral linkage is not formed, and the control idea is too single; meanwhile, the two-network balance adjustment adopts a flow adjustment mode, the adjustment is usually performed according to a set target value, the target value is mainly set by manual experience, so that the control of a heat supply system is not fine and fixed, more accurate control cannot be performed aiming at a specific heat supply area, the flow of each unit or heat user cannot be reasonably distributed, and the waste of resources is caused.

Therefore, it is desirable to provide a balancing method for heating networks to solve the above problems.

Disclosure of Invention

The invention provides a two-network heat supply balancing method aiming at the problems that in the prior art, heat supply enterprises generally have the phenomenon of hydraulic imbalance of a secondary pipe network, the pipe network is high in power consumption, low in heat efficiency, high in operation cost and unbalanced in heat supply.

The technical scheme for solving the technical problems is as follows: a heating two-network balancing method, comprising: establishing a heat supply model of heat supply two-network balance, wherein the heat supply two-network balance comprises heat supply balance in a building and heat supply balance among building units;

dividing the building into a high area, a middle area and a low area according to the height of the building, and arranging a variable frequency pump at the starting floor of each high area;

constructing a heat supply balance neural network of each height area to obtain the frequency of the variable frequency pump, and activating a circulation control instruction according to the adjustable range of the variable frequency pump to achieve heat supply balance in the building;

and regularly acquiring parameters of each building unit in the heat supply range of each heat exchange station, constructing a heat supply two-network balance neural network, and adjusting the water supply flow to achieve heat supply two-network balance.

The invention has the beneficial effects that: by changing the traditional operation mode of large flow and small temperature difference, the electric consumption of a water pump is reduced, the phenomenon of uneven cooling and heating of a user at room temperature is changed, the heat supply efficiency is improved, and heat energy is saved;

by setting the variable range of the variable frequency pump, the heat supply temperature of the high region is ensured to reach the standard, and the temperature in the low region is ensured not to generate an overheating phenomenon, so that the room temperature difference of the high, middle and low regions is within the standard range, and the heat supply balance is achieved;

on the basis of ensuring the heat supply balance among buildings, the heat supply balance in the buildings is met to the maximum extent, heat energy is saved according to the principle of large temperature difference and small flow, and the electricity consumption of the water pump is reduced.

On the basis of the technical scheme, in order to achieve the convenience of use and the stability of equipment, the invention can also make the following improvements on the technical scheme:

further, the heating model is Gr ═ (E, V), where E denotes a set of nodes E ═ { E ] of the directed graph Gr₁,e₂,...,e_n+1N is the number of building units, and V represents the set of edges V ═ V of the directed graph Gr₁,v₂,...,v_n}。

The beneficial effect of adopting the further technical scheme is that: a heat supply model is established based on a directed graph, various heat supply entities and the relation among the entities are represented by a directed graph data structure, each node represents an entity type, and the relation among the entities is represented by edges, so that a real heat supply relation network can be described more directly.

Further, the node set E comprises a master node E_mAnd child node E_sMaster node E_mFor heat exchange stations, each sub-node E_sThe properties of the heat exchange station comprise water supply temperature, water return temperature and total valve opening degree; the properties of the building units comprise water supply temperature, water return temperature, heat supply area, frequency of a variable frequency pump, distance between the variable frequency pump and a heat exchange station and valve opening degree of a high region, a middle region and a low region of each building unit.

Further, fitting the input data of the different height areas, outputting the optimal frequency of the variable frequency pump, setting a threshold value for the variable frequency pumps of the different height areas, and adjusting the variable frequency pumps to the optimal frequency if the output optimal frequency is within an adjustable range; otherwise, the output result is 0, the neural network is automatically activated, the valve opening with a certain span is increased, and new flow data f are obtained_dFitting againCircularly calculating until outputting the frequency R of the optimal variable frequency pump^j。

The beneficial effect of adopting the further technical scheme is that: the output frequency of the variable frequency pump is divided by setting a variable frequency pump frequency threshold value, and the variable frequency pump frequency is adjusted to the optimal frequency by different methods, so that the heat supply temperature of a high area reaches the standard, and the temperature in a low area is prevented from overheating, so that the temperature difference of the room temperature of the high, medium and low areas is within the standard range, and the heat supply balance is achieved.

Further, height district heat supply balance neural network includes input layer, fit layer, optional layer and output layer, the input layer is with data transmission after handling for the fit layer, the fit layer is with the result of calculation for the optional layer, the optional layer includes the neuron of two series connections, and every neuron screens the data of input according to filtering criteria, if all accord with, then with input data R^jSending to an output layer; otherwise, sending the empty set to an output layer; the output of the output layer isWherein R is^jIs the frequency of the variable frequency pump.

Further, the calculation process of the fitting layer isWherein rho is fluid medium density, xi is resistance coefficient of the regulating valve, S is cross-sectional area of the pipeline, f is flow, xi' is resistance coefficient of the pipeline section, c₀、c₁、c₂As a parameter of the circulation pump head, K_rIs the ratio of the actual operating frequency to the rated operating frequency of the circulating pump, f₀Is the actual flow of the circulating pump.

The beneficial effect of adopting the further technical scheme is that: redundant data and noise data are removed through an input layer of the heat supply balance neural network in the height area, the frequency of the variable frequency pump is calculated through a fitting layer according to factors such as resistance coefficient, medium density and flow, real-time data of the frequency of the variable frequency pump can be obtained, calculation accuracy of the neural network is guaranteed, and accuracy of output frequency division is improved.

Further, the heat supply two-network balance comprises the steps of calculating the total circulating temperature of the heat exchange station two-network, setting the water supply temperature based on the steady-state heat balance through a climate compensator according to the outdoor temperature change and the indoor temperature requirement, and adjusting the steady-state according to the operation of a heat supply system to calculate the water supply temperature and the water return temperature of the heat exchange station;

calculating the weight of each building unit for any heat exchange station and the coverage area thereof;

setting a minimum threshold Td of temperature difference between supply water and return water_εAnd a maximum flow threshold f_εWhen | Tg_d-Th_d|＜Td_εThen, according to the result of the neural network in the high region, a heat supply two-network balance neural network is constructed, and the temperature differences Td between the water supply and the water return of the n building units_dUsing time series as network input to obtain optimum flow value, wherein Tg_dFor the temperature of the water supply, Th_dThe temperature of the return water is shown.

The beneficial effect of adopting the further technical scheme is that: when the temperature difference between the supply water and the return water is smaller than the minimum threshold value, the balance adjustment of the heat supply secondary network needs to be carried out through the heat supply secondary network balance neural network, the waste of a large amount of system resources is avoided, the function of frequency conversion and speed regulation operation is effectively exerted, and the heat supply index is reached.

Further, the heat supply two-network balance neural network comprises an input layer, a mode layer, a summation layer and an output layer, wherein the input layer is connected with the mode layer and transmits an input variable to the mode layer, and a neuron transfer function of the mode layer is a radial basis function:wherein the content of the first and second substances,for neuron-corresponding learning samples, σ²Is the variance of the input variable with its corresponding sample.

Further, the summation layer comprises a first type of neurons and a second type of neurons, the first type of neurons is the output sum of each mode layer neuron, and the connection weight of the mode layer and each neuron is 1; the second class of neurons is a weighted sum of the expected result and each mode layer neuron.

The beneficial effect of adopting the further technical scheme is that: the method is based on the radial basis network, has good nonlinear approximation performance, calculates the matching degree of the input feature vector and each pattern in the training set, greatly accelerates the convergence speed of the learning process, and avoids being limited to local minimum values.

Further, the first type of neuron isThe second type of neuron isThe flow value output by the heat supply two-network balance neural network is Y,and Y is more than or equal to f_εOtherwise, take f_ε。

The beneficial effect of adopting the further technical scheme is that: the output of the summation layer is divided into two parts, the output of the first node is the arithmetic sum of the output of the mode layer, the output of the other nodes is the weighted sum of the output of the mode layer, no model parameter needs to be trained, and the convergence speed is high.

Drawings

FIG. 1 is a schematic diagram of a heat supply two-network balance method;

fig. 2 is a structural diagram of a heat supply balance neural network and a heat supply two-network balance neural network in the altitude area.

Detailed Description

The principles and features of this invention are described below in conjunction with the following drawings, which are set forth by way of illustration only and are not intended to limit the scope of the invention.

As shown in fig. 1 to 2, a heating two-network balance method includes establishing a heating model of heating two-network balance, which includes in-building heating balance and heating balance between building units;

the temperature difference of water supply and return of each building is consistent to serve as a two-network balance target, the temperature difference of water supply and return is consistent by controlling the water supply flow of the building according to the temperature difference of water supply and return of each building inlet, the balance adjustment of the two-network is completed, and the energy-saving and stable heat supply effect is achieved.

Establishing two-network heat supply model G_rE denotes a directed graph G_rThe node set of (1) comprises a master node E_mAnd child node E_sThe master node is E_mHeat exchange station, each sub-node E_sE is a building unit within the range of the heat exchange station, E ═ E₁,e₂,...,e_n+1N is the number of the building units. The attributes of the heat exchange station comprise water supply temperature, water return temperature and total valve opening; the properties of the building unit include: the water supply temperature, the water return temperature, the distance from the heat exchange station, the heat supply area and the valve opening are divided into a high area, a middle area and a low area according to the height of the building, a variable frequency pump is arranged at the starting floor of each high area to increase the flow lift of the pipe network liquid, and the attributes of the building unit also comprise the water supply temperature, the water return temperature, the heat supply area and the frequency of the variable frequency pump of the high area, the middle area and the low area of each building unit.

V denotes a directed graph G_rSet of edges of (V) { V ═ V₁,v₂,...,v_nAnd each side is divided into a water supply flow direction and a water return flow direction, and the attribute of each side comprises the cross section area of the pipeline and the flow rate of the pipeline.

Constructing a heat supply balance neural network of each height area to obtain the frequency of the variable frequency pump, and activating a circulation control instruction according to the adjustable range of the variable frequency pump to achieve heat supply balance in the building;

for in-building heat supply balance, a height zone heat supply balance neural network is constructed, as shown in fig. 2. The water supply temperature Tsh of different height areas_iReturn water temperature Trh_iPipeline cross-sectional area S, pipeline flow f, liquid density rho and heat supply area Mh of each zone_iAnd data such as supply and return water temperature difference standard Dhr are input into the neural network in the form of multidimensional matrix and are expressed asi represents a high, middle, and low area, j represents a jth group of data,each set of input sets consists of m-dimensional data.

And outputting the optimal frequency of the variable frequency pump by fitting the input data. According to the invention, a group of frequency threshold values of the variable frequency pump is set according to the actual situation, for the variable frequency pumps in different height areas, the upper threshold value and the lower threshold value of the adjustable range of the frequency are set, and if the output optimal frequency is within the adjustable range, the variable frequency pump is adjusted to the optimal frequency; otherwise, the output result is 0, and the slave neural network is automatically activated.

The height area heat supply balance neural network comprises an input layer, a fitting layer, a selection layer and an output layer. The input layer preprocesses the data to remove redundant data and noise data, and the method adopts the prior method, and the invention is not explained in more detail here. And the input layer sends the processed data to the fitting layer, and the connection weight between the input layer and the fitting layer is 1.

The calculation process for fitting layer neurons was:

wherein R is^jThe frequency of the variable frequency pump is shown as rho, the density of a fluid medium, xi, the resistance coefficient of an adjusting valve, S, the cross-sectional area of a pipeline, f, the resistance coefficient of a pipe section, and c₀、c₁、c₂As a parameter of the circulation pump head, K_rIs the ratio of the actual operating frequency to the rated operating frequency of the circulating pump, f₀Is the actual flow of the circulating pump. The fitting layer sends the calculation result to the selection layer.

Two series-connected neurons are arranged in the selection layer, each neuron is equivalent to a filter, after data is input, each neuron screens the data according to the own filtering standard, and if the data are all in accordance with the filtering standard, the data R is input^jSending to an output layer; otherwise, the empty set is sent to the output layer. The filtering threshold value of the first neuron is the upper limit delta of the adjustable range of the variable frequency pump_uThe filtering threshold value of the second neuron is the lower limit delta of the adjustable range of the variable frequency pump_dAnd the upper limit and the lower limit of the adjustable range are determined according to actual conditions.

The output of the output layer neurons is:

the output layer determines output according to the data sent by the selection layer, if the output layer is an empty set and outputs 0, the output layer activates a cycle control instruction and increases the valve opening of one span so as to obtain new flow data f_dAnd inputting new data into the neural network, and performing fitting calculation again to obtain an output result. Calculating circularly until outputting the optimal frequency R^j. Wherein, the span of the valve opening can be determined according to actual conditions.

For the heat supply balance among the buildings, the parameters of each building unit in the heat supply range of each heat exchange station are regularly acquired according to the principle of large temperature difference and small flow, a heat supply two-network balance neural network is constructed, the water supply flow is adjusted, and the heat supply two-network balance is achieved.

And setting a sampling period for heat supply balance among the buildings, and regularly acquiring the two-network water supply temperature, the two-network water return temperature and the opening degree of the regulating valve, namely the flow, of each building unit heating power inlet and outlet in the heat supply range of each heat exchange station device.

Firstly, calculating the total circulating temperature of a second network of the heat exchange station, setting the water supply temperature based on the steady-state heat balance according to the change of the outdoor temperature and the requirement of the indoor temperature through a climate compensator, and calculating the water supply temperature and the water return temperature of the heat exchange station in a steady-state mode according to the operation regulation of a heat supply system:

wherein Tg is the water supply temperature of the heat exchange station, Th is the water return temperature of the heat exchange station, T_nAnd the temperature is indoor temperature, Tg 'and Th' are respectively the standards of water supply temperature and water return temperature of the heat exchange station, Q is heat supply amount of the heat exchange station, and b is heat performance coefficient of the radiator.

For any heat exchange station and the coverage area thereof, calculating the weight of each building unit:

Mz＝M₁+M₂+...+M_D

wherein Mz is the sum of the heat supply areas of the heat exchange stations,d belongs to D and M for the heat supply area of any building unit_DThe heat supply area of the D-th building unit, D is the number of buildings, W_dIs the weight of any one building unit, l_dThe distance between any one building unit and the heat exchange station, l is the sum of the distances between all the building units and the heat exchange station, Ts_dSupply water temperature, Tr, for any one building unit_dFor the return water temperature of any building unit, M_dAccording to the actual situation as the set value, the design data can be obtained only by inquiring the relevant design data.

The water supply temperature Tg of each building unit_dWater return temperature Th_dComprises the following steps:

Tg_d＝W_d*Tg

Th_d＝W_d*Th

setting a minimum threshold Td of the temperature difference between the supplied water and the returned water according to the principle of large temperature difference and small flow_εAnd a maximum flow threshold f_ε. If | Tg_d-Th_d|＜Td_εIf the principle is not satisfied, a heat supply two-network balance neural network is constructed according to the result obtained by the height area heat supply balance neural network, and the temperature difference Td between the water supply and return of the n building units is used_dAnd (4) taking the time sequence as network input to obtain an optimal flow value.

The heat supply two-network balance neural network comprises an input layer, a mode layer, a summation layer and an output layer. The network input is denoted as Td ═ Td₁,Td₂,...,Td_nAnd determining an acquisition time interval according to actual requirements.

The number of neurons in the input layer is N, each neuron is a simple distribution unit, an input variable is directly transmitted to the mode layer, the input component is a vector, and the number of samples is N.

The mode layer is fully connected with the input layer, no connection exists in the layer, the number of neurons in the mode layer is equal to that of samples, each neuron corresponds to a different sample, and the transfer function of the neurons in the mode layer is a radial basis function:

wherein, the output of the neuron is in an exponential form of the squared Euclidean distance between the input variable and the corresponding sample,for neuron-corresponding learning samples, σ²Is the variance of the input variable with its corresponding sample.

Two types of neurons are arranged in the summation layer, the first type of neuron is the output sum of each neuron of the mode layer, and the connection weight value of the mode layer and each neuron is 1; the second type of neurons is a weighted sum of the expected outcome and each mode layer neuron.

The formula for the first class of neurons is:

the formula for the second class of neurons is:

the output of the output layer being the second node of the summation layer divided by the first node, i.e.

Y is the flow value output by the heat supply two-network balance neural network, and Y is more than or equal to f_εOtherwise, take f_ε。

According to the water supply and return temperature of the heat exchange station and the water supply and return temperature of each building unit, the opening degree of the regulating valve is controlled on the premise that the water supply pressure, the water return pressure, the water supply temperature and the water return temperature of the heat source do not exceed the protection set values. And after the heat supply balance neural network in the height area adjusts the opening degree of the valve, checking whether the principle of large temperature difference and small flow is met, if not, recalculating the flow through the heat supply two-network balance neural network, adjusting the opening degree of the valve, and readjusting the heat supply balance in the height area. The heat supply balance among the buildings is larger than the heat supply balance in the buildings.

In summary, the heat supply balance in the buildings is met to the maximum extent on the basis of ensuring the heat supply balance among the buildings, the heat energy is saved according to the principle of large temperature difference and small flow, and the electricity consumption of the water pump is reduced.

The above description is only a preferred embodiment of the present invention and should not be taken as limiting the scope of the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the present invention.