Energy-saving control method and system for solar combined heat pump

Abstract

The invention relates to the technical field of heat pumps, and provides an energy-saving control method and system for a solar combined heat pump, and the method comprises the steps: obtaining a heat energy feature vector according to a heat absorption power feature, a heat release power feature, an environment temperature feature, an environment humidity feature and a solar radiation intensity feature; obtaining a uniform coefficient according to the heat absorption uniformity, the heat release uniformity, the pipeline resistance coefficient and the equipment aging coefficient; a self-adaptive control target is obtained according to the heat energy feature vector, the uniformity coefficient and historical data; and operation control over the solar heat pump set is achieved according to the self-adaptive control target. According to the method, factors such as environment temperature and humidity which have remarkable influence on the performance of the heat pump are considered, the accuracy of heat pump unit uniformity identification can be improved, and the finally obtained uniformity coefficient better reflects the actual uniformity of the heat pump unit; the problems that according to an energy-saving control method of a solar combined heat pump in the prior art, heat energy characteristic analysis is limited, and a uniformity recognition model is single are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of heat pumps, and in particular to an energy-saving control method and system for a solar-energy combined heat pump. Background Art

[0002] Solar energy and heat pump technology are two of the most promising ways to conserve conventional fossil energy. Their combined strengths complement each other. A combined solar energy and heat pump system is called a solar heat pump system, and there are different ways to combine these technologies. Depending on the combination of the solar collector and the heat pump evaporator, solar heat pump technology can be categorized as either indirect expansion or direct expansion.

[0003] The solar heat pump thermal system mainly consists of three parts: energy collection area, energy conversion area and energy utilization area. The energy collection area is composed of trough collectors and thermal auxiliary equipment. The energy conversion area is composed of hot water heat pump main unit and water tank. The energy utilization area is the indoor part of the hot water use place, including hot water pipes, faucets, etc.

[0004] The Chinese invention patent application number CN202310805653.6 discloses an energy-saving control method for a solar-powered combined heat pump. This method solves the technical problem that the existing technology lacks energy-saving control over solar-powered combined heat pumps, resulting in the solar-powered combined heat pumps being very energy-consuming. It achieves precise energy-saving control over the solar-powered combined heat pumps and reduces the energy consumption of the solar-powered combined heat pumps.

[0005] However, the aforementioned energy-saving control method for solar-powered heat pumps suffers from the following drawbacks: 1. It only considers the heat absorption and release power characteristics for thermal energy signature analysis, which may not fully reflect the actual operating status of the heat pump system; 2. The first uniformity coefficient is calculated based solely on the heat absorption and release uniformity, resulting in a relatively simple model. In other words, existing energy-saving control methods for solar-powered heat pumps suffer from limitations in thermal energy signature analysis and a single uniformity identification model. Summary of the Invention

[0006] Based on this, in order to solve the problems of thermal energy characteristic analysis limitations and uniformity identification model simplicity in the energy-saving control methods of solar-powered combined heat pumps in the prior art, the present invention provides an energy-saving control method and system for solar-powered combined heat pumps. The specific technical solutions are as follows:

[0007] A method for controlling energy conservation of a solar-heat-pump system, comprising:

[0008] Obtaining heat absorption power characteristics and heat release power characteristics of each heat pump in the solar heat pump group;

[0009] Acquire an ambient temperature characteristic, an ambient humidity characteristic, and a solar radiation intensity characteristic, and acquire a thermal energy characteristic vector according to the heat absorption power characteristic, the heat release power characteristic, the ambient temperature characteristic, the ambient humidity characteristic, and the solar radiation intensity characteristic;

[0010] Obtaining heat absorption uniformity, heat release uniformity, pipeline resistance factor, and equipment aging factor, and obtaining a uniformity coefficient based on the heat absorption uniformity, heat release uniformity, pipeline resistance coefficient, and equipment aging factor;

[0011] Obtaining an adaptive control target according to the thermal energy characteristic vector, uniformity coefficient and historical data;

[0012] The operation of the solar heat pump group is controlled according to the adaptive control target.

[0013] The energy-saving control method of the solar energy combined heat pump obtains a thermal energy characteristic vector through the heat absorption power characteristic, the heat release power characteristic, the ambient temperature characteristic, the ambient humidity characteristic and the solar radiation intensity characteristic, obtains a uniformity coefficient based on the heat absorption uniformity, the heat release uniformity, the pipeline resistance coefficient and the equipment aging coefficient, and obtains an adaptive control target according to the thermal energy characteristic vector, the uniformity coefficient and the historical data. The method takes into account factors such as ambient temperature and humidity that have a significant impact on the performance of the heat pump, and can improve the accuracy of uniformity identification of the heat pump group. The uniformity coefficient finally obtained better reflects the actual uniformity of the heat pump group, and solves the problems of thermal energy characteristic analysis limitations and the singleness of the uniformity identification model in the energy-saving control method of the solar energy combined heat pump in the prior art.

[0014] Preferably, the specific method for obtaining the uniformity coefficient includes:

[0015] Obtaining a degree of thermal imbalance according to the heat absorption uniformity and the heat release uniformity;

[0016] The pipeline resistance factor and the equipment aging factor are processed by a fuzzy logic method to obtain a uniformity compensation value;

[0017] Obtaining an environmental disturbance factor and a timing attenuation factor, and obtaining a uniformity offset according to the environmental disturbance factor and the timing attenuation factor;

[0018] A uniformity coefficient is obtained according to the thermal imbalance, the uniformity compensation value, and the uniformity offset.

[0019] Preferably, the specific method for obtaining the adaptive control target includes:

[0020] Acquire a state vector according to the thermal energy characteristic vector, uniformity coefficient and historical data;

[0021] Obtaining system efficiency, and constructing a reward function based on the uniformity coefficient and the system efficiency;

[0022] A dynamic adaptive target generation function is constructed according to the state vector and the reward function, and an adaptive control target is obtained based on the dynamic adaptive target generation function.

[0023] Preferably, the specific method for obtaining system efficiency includes the following steps:

[0024] Obtain the useful heat output by the system, the total energy consumption of the system, and the equipment aging correction factor;

[0025] Obtaining an energy allocation weight coefficient based on the ambient temperature characteristics, ambient humidity characteristics, and solar radiation intensity characteristics;

[0026] The system efficiency is obtained according to the useful heat, the total system energy consumption, the equipment aging correction factor and the energy allocation weight coefficient.

[0027] Preferably, according to the formula U=sigmoid(α1·f(P in ,P out )+α2·g(R,A)+α3·h(E d ,T d ))·C scale Get the uniformity coefficient U;

[0028] Among them, f(P in ,P out )、g(R,A)、h(E d ,T d ) represent the thermal imbalance, uniformity compensation value and uniformity offset respectively, α1, α2, α3 represent the weight coefficients of the thermal imbalance, uniformity compensation value and uniformity offset respectively, P in 、P out They represent heat absorption uniformity and heat release uniformity respectively, R and A represent pipeline resistance factor and equipment aging factor respectively, E d 、T d Represent the environmental disturbance factor and the time series attenuation factor, C scale Indicates the scaling factor.

[0029] Preferably, according to formula G t =argmax a∈A [β1·R imm (S t ,a)+β2·E[V θ (S t+1 )]]+β3·Η(π(·|S t )) Obtain a dynamic adaptation target generation function;

[0030] Among them, S t represents the state vector, Rimm represents the reward function, a represents the control action, V θ represents the long-term value network, H(π) represents the policy entropy, β1, β2, β3 represent the weight coefficients of the state vector, reward function and policy entropy respectively, A represents the action space, E and θ represent the expected value and policy network parameters respectively.

[0031] An energy-saving control system for a solar-energy combined heat pump, used to implement the energy-saving control method for the solar-energy combined heat pump, comprising:

[0032] The characteristic parameter acquisition module is used to obtain the heat absorption power characteristics and heat release power characteristics of each heat pump in the solar heat pump group, as well as the ambient temperature characteristics, ambient humidity characteristics and solar radiation intensity characteristics;

[0033] a feature vector acquisition module, configured to acquire a thermal energy feature vector based on the heat absorption power feature, the heat release power feature, the ambient temperature feature, the ambient humidity feature, and the solar radiation intensity feature;

[0034] A uniformity coefficient acquisition module is used to obtain heat absorption uniformity, heat release uniformity, pipeline resistance factor and equipment aging factor, and obtain the uniformity coefficient based on the heat absorption uniformity, heat release uniformity, pipeline resistance coefficient and equipment aging factor;

[0035] A control target acquisition module, configured to acquire an adaptive control target based on the thermal energy characteristic vector, uniformity coefficient, and historical data;

[0036] An operation control module is used to realize operation control of the solar heat pump group according to the adaptive control target.

[0037] Preferably, the uniformity coefficient acquisition module includes:

[0038] a thermal imbalance degree obtaining unit, configured to obtain the thermal imbalance degree according to the heat absorption uniformity and the heat release uniformity;

[0039] a compensation value obtaining unit, configured to process the pipeline resistance factor and the equipment aging factor by a fuzzy logic method to obtain a uniformity compensation value;

[0040] an offset acquiring unit, configured to acquire an environmental disturbance factor and a time series attenuation factor, and acquire a uniformity offset according to the environmental disturbance factor and the time series attenuation factor;

[0041] The uniformity coefficient acquisition unit is used to acquire the uniformity coefficient according to the thermal imbalance degree, the uniformity compensation value and the uniformity offset.

[0042] Preferably, the control target acquisition module includes:

[0043] A state vector acquisition unit, configured to acquire a state vector according to the thermal energy characteristic vector, the uniformity coefficient, and historical data;

[0044] A reward function construction unit, configured to obtain system efficiency and construct a reward function based on the uniformity coefficient and the system efficiency;

[0045] A control target acquisition unit is used to construct a dynamic adaptive target generation function according to the state vector and the reward function, and acquire an adaptive control target based on the dynamic adaptive target generation function.

[0046] Preferably, the reward function construction unit includes:

[0047] System parameter acquisition subunit, used to obtain the useful heat output by the system, the total energy consumption of the system and the equipment aging correction factor;

[0048] A weight coefficient acquisition subunit is used to obtain an energy allocation weight coefficient according to the ambient temperature characteristics, ambient humidity characteristics and solar radiation intensity characteristics;

[0049] The system efficiency acquisition subunit is used to obtain the system efficiency according to the useful heat, the total system energy consumption, the equipment aging correction coefficient and the energy allocation weight coefficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The present invention can be further understood from the following description in conjunction with the accompanying drawings. The components in the figures are not necessarily drawn to scale, but rather the emphasis is placed on illustrating the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0051] Figure 1 This is a schematic diagram of the overall process of an energy-saving control method for a solar combined heat pump in one embodiment of the present invention;

[0052] Figure 2 is a flow chart of a specific method for obtaining a uniformity coefficient in one embodiment of the present invention;

[0053] Figure 3 is a flowchart of a specific method for obtaining an adaptive control target in one embodiment of the present invention;

[0054] Figure 4 is a flowchart of a specific method for obtaining system efficiency in one embodiment of the present invention;

[0055] Figure 5 This is a schematic diagram of the overall structure of an energy-saving control system for a solar combined heat pump according to an embodiment of the present invention;

[0056] Figure 6 2 is a functional structure diagram of a uniformity coefficient acquisition module according to an embodiment of the present invention;

[0057] Figure 7 Schematic diagram of the training iteration process of the Q-Learning algorithm in one embodiment of the present invention;

[0058] Figure 8 1 is a schematic diagram of the simulation experiment effect of the uniformity coefficient acquisition module in one embodiment of the present invention. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with its embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0060] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly attached to the other element or there may be an intermediate element. When an element is referred to as being "connected to" another element, it may be directly connected to the other element or there may be an intermediate element. The terms "vertical," "horizontal," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only implementation methods.

[0061] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used herein in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0062] The "first" and "second" in the present invention do not represent specific quantities and orders, but are only used to distinguish names.

[0063] Before describing the embodiments of the present invention, a brief introduction to the prior art is given first.

[0064] A solar combined heat pump is a composite energy system that combines solar energy utilization technology with heat pump technology. It aims to achieve more efficient and energy-saving heat acquisition, transfer and utilization through the synergy of the two technologies. Among them, solar technology absorbs solar radiation energy through solar collectors (such as flat-plate collectors and vacuum tube collectors) and converts it into thermal energy for heating water or air. Solar energy is a clean and renewable heat source, but it is greatly affected by weather, day and night, and seasons (such as rainy days and reduced efficiency in winter). Heat pumps extract heat from low-temperature heat sources (such as air, water, soil) by consuming a small amount of electricity or other energy, and "pump" it into a high-temperature environment for use, such as air source heat pumps and ground source / water source heat pumps.

[0065] Solar energy combined with heat pumps achieves complementary advantages and a circular energy cycle. When solar energy is insufficient (e.g., at night or on rainy days), the heat pump automatically activates to provide additional heat. When solar energy is in excess, excess heat can be stored or used to assist the heat pump, improving overall efficiency. Both systems can share heat storage devices (e.g., hot water tanks), forming an intelligent control system that prioritizes solar energy and supplements heat pumps, reducing reliance on traditional energy sources.

[0066] The main components of a solar heat pump system include: 1. A solar thermal collector module, used to collect solar thermal energy; 2. A heat pump unit, including a compressor, condenser, evaporator, etc., responsible for heat transport; 3. A heat storage device, such as an insulated water tank, which stores excess heat for use during off-peak hours; and 4. A control system, which intelligently switches between solar and heat pump operating modes to optimize energy efficiency.

[0067] The working modes of solar heat pump system generally include

[0068] 1. Solar energy priority mode: On sunny days, the solar collector directly heats the medium (water or air). After the demand is met, the remaining heat is stored in the heat storage device.

[0069] 2. Heat pump auxiliary mode: When solar energy is insufficient, the heat pump starts and uses the waste heat in the heat storage device or absorbs heat from the outside to supplement heating.

[0070] 3. Composite mode: In extreme weather conditions, solar energy and heat pumps operate simultaneously to quickly reach the required temperature (such as winter heating in extremely cold regions).

[0071] The Chinese invention patent with application number CN202310805653.6 discloses an energy-saving control method for a solar-powered combined heat pump, which has the problems of limitations in thermal energy feature analysis and the singleness of the uniformity identification model. Specifically, the energy-saving control method only considers the heat absorption power characteristics and heat release power characteristics to perform thermal energy feature analysis, which may not fully reflect the actual operating status of the heat pump system. For example, factors such as ambient temperature and humidity that have a significant impact on the performance of the heat pump are not considered, which may lead to inaccurate uniformity identification of the heat pump group. In addition, the energy-saving control method only calculates the first uniformity coefficient based on the heat absorption uniformity and heat release uniformity, and the model is relatively simple. In actual operation, the performance of the heat pump is also affected by factors such as differences in pipeline resistance and the degree of equipment aging. These factors are not reflected in the uniformity identification, which may make the uniformity coefficient unable to truly reflect the overall uniformity of the heat pump group.

[0072] In order to solve the problems existing in the existing energy-saving control methods of solar combined heat pumps, such as Figure 1 As shown, a solar energy combined heat pump energy-saving control method includes the following steps:

[0073] S1, obtaining heat absorption power characteristics and heat release power characteristics of each heat pump in the solar heat pump group.

[0074] The heat absorption power characteristic represents the amount of heat absorbed by the heat pump per unit time, and the heat release power characteristic represents the amount of heat released by the heat pump per unit time.

[0075] S2, obtaining ambient temperature characteristics, ambient humidity characteristics, and solar radiation intensity characteristics, and obtaining a thermal energy characteristic vector according to the heat absorption power characteristics, heat release power characteristics, ambient temperature characteristics, ambient humidity characteristics, and solar radiation intensity characteristics.

[0076] Ambient temperature characteristics can be collected by temperature sensors, ambient humidity characteristics can be collected by humidity sensors, and solar radiation intensity characteristics can be obtained by photodiodes or thermopile sensors.

[0077] The heat absorption power feature P1, heat release power feature P2, and ambient temperature feature T can be analyzed by multivariate data analysis methods, such as principal component analysis (PCA). a , environmental humidity characteristics H and solar radiation intensity characteristics I are subjected to dimensionality reduction and comprehensive analysis to construct a more comprehensive thermal energy feature vector to more accurately reflect the operating status of the heat pump system.

[0078] Specifically, first construct the basic feature vector Perform dimensionality reduction through principal component analysis and obtain the eigenvector after dimensionality reduction Where W is the eigenvector matrix of the covariance matrix, which can be obtained by singular value decomposition.

[0079] Then, the continuous time series data is analyzed at multiple scales through the time series convolution network to obtain the time series convolution feature vector Among them, * represents convolution operation, W dil is the dilated convolution kernel, ReLU is the activation function, and b1 is the bias term.

[0080] Redesign the fuzzy rule base, map discrete parameters into continuous features, and obtain fuzzy logic membership Finally, the neural network is combined to output the final thermal energy feature vector Among them, b2 is the bias term, and the softmax function is used to convert the input vector into a probability distribution, ensuring that the value of each element is in the range [0,1] and the sum is 1. The weight matrix W nnIt is defined as a trainable parameter matrix connecting the input feature vectors to the output layer of a neural network. It is optimized by the backpropagation algorithm to adjust the weights of different features. Fuzzy logic membership is defined as the output of the membership function of the elements in the fuzzy set, mapping continuous parameters (such as temperature) to fuzzy sets (such as high temperature and medium temperature).

[0081] function The complete process is as follows: first, the eigenvector after dimensionality reduction, the time series convolution eigenvector and the fuzzy logic membership are spliced ​​into a unified vector, and then a linear transformation is performed to obtain the vector through the weight matrix W. nn The weighted summation is performed with the bias term b2 to generate an intermediate vector. Finally, the intermediate variable is converted into a probability distribution through the softmax function to output the final thermal energy feature vector.

[0082] S3, obtaining heat absorption uniformity, heat release uniformity, pipeline resistance factor and equipment aging factor, and obtaining a uniformity coefficient based on the heat absorption uniformity, heat release uniformity, pipeline resistance coefficient and equipment aging factor.

[0083] Pipeline resistance factor R = pipeline pressure difference / pipeline flow, which can be corrected by combining Bernoulli equation. Equipment aging factor A = e -KT When the equipment aging factor is 1, it means that the equipment is in a brand new state and has no aging. When the equipment aging factor is 0, it means that the equipment is close to being scrapped and its performance is seriously degraded.

[0084] e represents a natural constant, K represents the aging rate constant, and T represents the equipment operating time. The larger the aging rate constant, the faster the equipment ages, and vice versa. The aging rate constant can be obtained by fitting historical operating data, often using the least squares method or maximum likelihood estimation. The aging rate constant is affected by environmental conditions (temperature, humidity), equipment material, load intensity, and so on. For example, high temperature and high humidity environments increase K and accelerate equipment aging. The relationship between the aging rate constant and equipment operating time can be established through analysis and fitting of large amounts of historical data (such as equipment performance monitoring and fault records).

[0085] As a preferred technical solution, Figure 2 As shown in Figure 2, the specific methods for obtaining the uniformity coefficient include:

[0086] S31 , obtaining a thermal imbalance degree according to the heat absorption uniformity and the heat release uniformity.

[0087] S32, processing the pipeline resistance factor and the equipment aging factor by a fuzzy logic method to obtain a uniformity compensation value.

[0088] S33: Obtain an environmental disturbance factor and a timing attenuation factor, and obtain a uniformity offset according to the environmental disturbance factor and the timing attenuation factor.

[0089] S34: Obtain a uniformity coefficient according to the thermal imbalance, the uniformity compensation value, and the uniformity offset.

[0090] According to the formula U=sigmoid(α1·f(P in ,P out )+α2·g(R,A)+α3·h(E d ,T d ))·C scale Get the uniformity coefficient U; where f(P in ,P out )、g(R,A)、h(E d ,T d ) represent the thermal imbalance, uniformity compensation value and uniformity offset respectively, α1, α2, α3 represent the weight coefficients of the thermal imbalance, uniformity compensation value and uniformity offset respectively, P in 、P out They represent heat absorption uniformity and heat release uniformity respectively, R and A represent pipeline resistance factor and equipment aging factor respectively,

[0091] E d 、T d Represent the environmental disturbance factor and the time series attenuation factor, C scale Indicates the scaling factor.

[0092] Specifically, f(·), g(·), and h(·) are mapping functions corresponding to the thermal imbalance, uniformity compensation value, and uniformity offset, respectively. α1, α2, and α3 can be set based on experience or adjusted in real time through reinforcement learning.

[0093] in, Here, the thermal imbalance factor replaces the traditional uniformity calculation, solving the problem of static mean ignoring local fluctuations.

[0094] g(R, A) is used to quantify the coupled effects of pipeline resistance and equipment aging using fuzzy logic. The mapping function g(·) uses the pipeline resistance factor as its input variable to the fuzzy set {low, medium, high} and the equipment aging factor as its input variable to the fuzzy set {new, medium, old}. The output g(R, A) is the uniformity compensation value {negative large, negative small, zero, positive small, positive large}. The following table shows an example of the fuzzy rule base:

[0095] R A g(R,A) Low new Zhengda high old Big middle middle zero

[0096] Uniformity offset h(E d ,Td ) can be obtained by constructing an LSTM neural network and using the LSTM neural network to predict the impact of environmental disturbances. Environmental disturbance factor E d =[T a ,H,I], the input variables of the LSTM neural network include the environmental disturbance factor and the time series attenuation factor, and the output uniformity offset h(E d ,T d ).

[0097] Scaling factor C scale =1+λ·CV, which is used to adjust the output range of the uniformity coefficient U to adapt it to different operating conditions. The base value of the scaling factor is 1, indicating no scaling. As CV increases, the scaling factor increases linearly, expanding the adjustment range of U. As CV decreases, the scaling factor approaches 1, narrowing the adjustment range.

[0098] Wherein, λ represents the preset adjustment coefficient, which is used to control the influence of the coefficient of variation CV on the scaling factor. CV represents the coefficient of variation, and It measures the degree of data dispersion relative to the mean. A larger CV indicates greater data fluctuations and a more unstable system. A smaller CV indicates less data fluctuations and a more stable system.

[0099] σ' and μ' respectively represent the standard deviation and variance of system operating parameters, which include but are not limited to heat absorption power, heat release power, ambient temperature and humidity, and equipment efficiency.

[0100] S4, obtaining an adaptive control target according to the thermal energy characteristic vector, uniformity coefficient and historical data.

[0101] As a preferred technical solution, Figure 3 As shown in Figure 2, the specific method for obtaining the adaptive control target includes:

[0102] S41, obtaining a state vector based on the thermal energy characteristic vector, uniformity coefficient and historical data. D represents historical data.

[0103] S42, obtaining system efficiency, and constructing a reward function according to the uniformity coefficient and the system efficiency.

[0104] Specifically, the system efficiency It quantifies the efficiency of a solar-heat pump system in converting input energy into useful heat. Its core goal is to maximize energy utilization while balancing the operating ratio of solar energy and heat pumps. This system efficiency function dynamically adjusts system parameters to achieve an optimal solution for energy-saving control and coordinated energy utilization.

[0105] Among them, Quseful Indicates useful heat, that is, the heat actually output by the system to meet user needs (such as heating or cooling needs), taking into account heat loss and equipment efficiency. total Indicates the total energy consumption of the system, including: heat pump power consumption, such as compressor power consumption, circulation pump power consumption; other losses, such as pipeline heat conduction loss, etc.

[0106] S43: construct a dynamic adaptive target generation function according to the state vector and the reward function, and obtain an adaptive control target based on the dynamic adaptive target generation function.

[0107] S5: Implementing operational control of the solar heat pump system based on the adaptive control objective. Specifically, the strategy network may output a probabilistic action distribution and select an action that maximizes the adaptive control objective to implement operational control of the solar heat pump system, such as adjusting the solar energy ratio and heat pump power.

[0108] In summary, the energy-saving control method of the solar energy combined heat pump obtains the thermal energy characteristic vector through the heat absorption power characteristic, heat release power characteristic, ambient temperature characteristic, ambient humidity characteristic and solar radiation intensity characteristic, obtains the uniformity coefficient based on the heat absorption uniformity, heat release uniformity, pipeline resistance coefficient and equipment aging coefficient, and obtains the adaptive control target according to the thermal energy characteristic vector, uniformity coefficient and historical data. It takes into account factors such as ambient temperature and humidity that have a significant impact on the performance of the heat pump, and can improve the accuracy of uniformity identification of the heat pump group. The uniformity coefficient finally obtained better reflects the actual uniformity of the heat pump group, and solves the problems of thermal energy characteristic analysis limitations and the singleness of the uniformity identification model in the energy-saving control method of the solar energy combined heat pump in the prior art.

[0109] As a preferred technical solution, Figure 4 As shown, the specific method for obtaining system efficiency includes the following steps:

[0110] S421, obtain the useful heat Q output by the system useful , total system energy consumption E total and the equipment aging correction factor β.

[0111] S422, according to the ambient temperature characteristic T a , environmental humidity characteristics H and solar radiation intensity characteristics I to obtain the energy allocation weight coefficient.

[0112] The energy allocation weight coefficient can be calculated by fuzzy logic based on the ambient temperature characteristics, ambient humidity characteristics and solar radiation intensity characteristics. Specifically, according to the formula ω t =f(T a ,H,I) obtain the energy allocation weight coefficient ω t, that is, based on fuzzy logic, according to the ambient temperature characteristics T a , environmental humidity characteristics H and solar radiation intensity characteristics I set rules to obtain energy allocation weight coefficient ω t .

[0113] First, define the input variables and fuzzify them, and transform the ambient temperature feature T a , environmental humidity characteristics H and solar radiation intensity characteristics I are set as input parameters, and fuzzy set partitioning is performed to convert T a The temperature is divided into low temperature, medium temperature, and high temperature; H is divided into low humidity, medium humidity, and high humidity; I is divided into weak light, medium light, and strong light. The precise value is mapped to the fuzzy linguistic variable through the membership function.

[0114] Then, a fuzzy rule base is established. Based on expert experience or historical data, the defined rules include but are not limited to the following:

[0115] Rule 1: If T a For low temperature AND H If the humidity is high, then ω t Preference towards heat pumps;

[0116] Rule 2: If I is strong light, then ω t Prefer solar energy.

[0117] The number of rules is the Cartesian number of the fuzzy set of the input variables (e.g. 3×3×3=27)

[0118] Then, fuzzy reasoning and synthesis are performed, and the output ω is obtained by calculating the matching degree of each rule (taking the minimum value of the input variable membership degree). t The fuzzy sets (such as low, medium, and high weights) are weighted summed to generate intermediate results.

[0119] Finally, defuzzification is performed by converting the fuzzy result into an exact value ω through the centroid method or the maximum membership method. t ∈[0,1]. For example, if the output is "high weight", then ω t =0.8; if it is "medium weight", then ω t =0.5.

[0120] S423: Obtain system efficiency according to the useful heat, the total system energy consumption, the equipment aging correction coefficient, and the energy allocation weight coefficient.

[0121] The system efficiency Here, the system efficiency function is explained in detail. Environmental factors have a certain impact on the useful heat. When the temperature is low and the humidity is high, the heat loss increases and the useful heat may decrease. At this time, the energy allocation weight coefficient ω is required. tImprove the heat pump weight compensation. β represents the equipment aging correction coefficient, which reflects the impact of equipment efficiency attenuation on energy consumption. It is obtained through training of historical operation data and is used to correct E caused by equipment aging. total to avoid underestimation of energy efficiency.

[0122] Specifically, we first collect historical operating data including the equipment's cumulative operating time, number of starts, load changes, photovoltaic panel irradiation time, ambient temperature and humidity, and vibration values, and then clean and normalize the historical data to eliminate outliers, and perform time alignment and unit unification on multi-source data.

[0123] Next, the model is selected and its parameters calibrated. Either a mechanical fatigue model or a data-driven model can be used. Mechanical fatigue models are applicable to mechanical components such as motors and pumps. They calculate the remaining life based on the Paris law or Miner's damage accumulation theory, combined with the number of stress cycles. Data-driven models include machine learning models, such as LSTM neural networks or random forest models, which output β values ​​based on input parameters such as operating time and temperature. Parameter calibration involves experimental calibration and least squares calibration. Experimental calibration can obtain key parameters through accelerated aging tests (such as high-temperature cycling tests). The least squares method optimizes model parameters by fitting the error between historical data and model predictions.

[0124] Here, the β value can be adjusted according to the real-time operating conditions (such as temperature and load). For example, a high temperature environment may accelerate aging, in which case a temperature correction factor can be introduced into the calculation.

[0125] Finally, verification and update are performed. Model parameters are regularly recalibrated using the latest three months of data to prevent model drift. Alternatively, Kalman filtering is used to correct the β value in real time to eliminate noise interference and update the β value to reflect the current status of the device.

[0126] α represents the efficiency degradation coefficient, which is defined as the degree to which the efficiency of a quantified device decreases over time or with the number of uses (e.g., the power degradation rate of a photovoltaic panel or the reduction in the COP of a heat pump).

[0127] Specifically, α=f(T 设备 , frequency of use). T 设备 Refers to the equipment operation time, in hours or years, and the frequency of use refers to the equipment's operating cycle or load intensity (such as the number of daily starts, load rate). The specific methods for obtaining the efficiency reduction coefficient include:

[0128] The first step is to collect efficiency data of the equipment at different time points or times of use through experiments or field monitoring. For example, the output power of a photovoltaic panel changes over time, or the COP of a heat pump changes over operating hours.

[0129] The second step is to select an appropriate decay model based on the device type. Common models include linear, exponential, and polynomial decay models. For example, photovoltaic panels typically use an exponential decay model, while certain mechanical components, such as motors, may be more appropriately modeled using a linear model.

[0130] The third step is to use the collected data to fit the model parameters, such as fitting by the least squares method.

[0131] The fourth step is to verify the accuracy of the model through cross-validation or independent data sets, adjust the parameters to improve the prediction accuracy, and finally obtain an acceptable efficiency decay coefficient.

[0132] Here, the system efficiency function By dynamically adjusting the energy allocation ratio between solar energy and heat pumps based on real-time environmental parameters through dynamic weighting coefficients, the system prioritizes renewable energy and reduces electricity consumption, demonstrating adaptive optimization. By introducing a correction term (1-β·α) for equipment aging, it dynamically corrects for increased energy consumption due to equipment efficiency degradation, compensating for equipment status and ensuring energy efficiency stability during long-term operation. Furthermore, the system's efficiency function balances energy utilization and system reliability while ensuring useful heat output, making it suitable for intelligent control under complex operating conditions and achieving multi-objective collaborative optimization.

[0133] As a preferred technical solution, according to formula G t =argmax a∈A [β1·R imm (S t ,a)+β2·E[Vθ(S t+1 )]]+β3·Η(π(·|S t )) Get the dynamic adaptation target generation function G t ; Among them, S t represents the state vector, R imm Represents the reward function, a represents the control action, Vθ represents the long-term value network, H(π) represents the policy entropy, β1, β2, β3 represent the weight coefficients of the state vector, reward function and policy entropy respectively, A represents the action space, E and θ represent the expected value and policy network parameters respectively.

[0134] G t This adaptive control objective can be understood as the optimal control objective (e.g., heat pump power allocation, uniformity coefficient threshold) generated at time step t, balancing short-term and long-term objectives by maximizing the weighted overall benefit. The state vector, which contains the thermal energy eigenvector, uniformity coefficient, and historical data, is used to capture the dynamic characteristics of the system. a∈A can be understood as the control action a belonging to a feasible operation in the action space A (e.g., solar / heat pump power ratio).

[0135] in, Indicates the uniformity improvement rate, which is used to promote the balanced operation of the heat pump group. |η total -η opt | represents the system efficiency deviation, η total represents the real-time efficiency of the system, η opt Represents the theoretical optimal value of system efficiency, ΔP grid Indicates the grid power P grid The change in || ||2 represents the L2 norm, ||ΔP grid ||2 represents the grid power fluctuation penalty term, used to reduce grid impact. w1, w2, and w3 represent corresponding weight coefficients, which can be adjusted based on the priority of different scenarios, such as w1:w2:w3 = 0.6:0.3:0.1 or w1:w2:w3 = 0.5:0.4:0.1.

[0136] Long-term value network V θ =LSTM θ (S t )+β Φ ·Φ thermal (T res ,m fluid ); Among them, LSTM θ represents a temporal feature extraction network for learning environmental dynamics, such as illumination periodicity. thermal Represents the thermal inertia compensation term, quantifying the fluid heat capacity m fluid and the water storage tank temperature T res The influence of β Φ It represents the attenuation coefficient of the thermal inertia compensation term and can be set based on experience.

[0137] The reward function quantifies the benefits of single-step control, including uniformity improvement rate, system efficiency deviation, and grid fluctuation penalty. The long-term value network addresses thermal system delay issues by integrating LSTM time series feature extraction with thermal inertia compensation.

[0138] Dynamically adapt the target generation function G t The calculation process includes the following steps:

[0139] Step 1: State encoding. Input thermal energy feature vector Uniformity coefficient U t And the historical data series D t-k:t , generate hidden state h through LSTM t : Among them, a t-1 represents the selected action at the previous moment t-1, h t-1 Indicates the hidden state at the previous moment. ht-1 is the internal state of STM, implying the historical data sequence D t-k:t long-term dependency information.

[0140] LSTM dynamically integrates current input and historical information through forget gate, input gate, and output gate. The forget gate determines which information in the previous hidden state needs to be retained or discarded. For example, if the ambient temperature fluctuates greatly in the historical sequence, the forget gate may reduce the sensitivity to short-term temperature changes and retain the long-term trend. The input gate converts the current input With h t-1 Combined, a new memory unit is generated. For example, if the current solar radiation drops sharply, the input gate will enhance the heat pump power regulation action a t-1 The output gate generates the current hidden state based on the updated memory unit for subsequent action decisions.

[0141] The historical data series is implicitly encoded, and its role is reflected in the following ways:

[0142] Long-term dependency capture: LSTM hidden state h t-1 The features of the past k time steps (such as the trend of ambient temperature changes and the cumulative effect of equipment aging) are implicitly included, without the need to explicitly input the complete historical sequence.

[0143] Dynamic weight allocation: Through the gating mechanism, LSTM automatically assigns different weights to features at different historical moments. For example, if the system has triggered a protection shutdown due to high temperature at time t-3, this event may have an impact on the current hidden state h t Have a lasting impact.

[0144] Time series feature extraction of environmental parameters: LSTM accumulates historical fluctuations in ambient temperature and humidity to identify seasonal changes (such as increased heating demand in winter) or equipment aging. Solar radiation intensity features in historical data help the model learn daily cycles in sunlight intensity (such as peaks at noon) and sudden weather changes (such as cloud cover).

[0145] Action continuity modeling: If action a t-1 To "reduce heat pump power", LSTM will combine the historical action sequence to determine whether to continue the strategy (e.g., gradual cooling is required under continuous high temperature). t-2 Fault protection has been triggered, the current h t Will enhance the U t sensitivity to avoid repeated anomalies.

[0146] Learning of system dynamics: by accumulating history R and A (not explicitly input but reflected in U t LSTM can learn nonlinear responses under different working conditions.

[0147] Step 2, through the policy network π(a|S t ) Output the action probability distribution, adopt the ε-greedy strategy, and select the maximum value action with probability 1-ε: a t =argmax a (β1·R imm +β2·V θ (S t+1 |a)).

[0148] Step 3, rolling optimization, re-evaluates the action sequence and corrects the current action by predicting the next N steps of state.

[0149] Step 4: Dynamically update the target. Update the target network parameters of the long-term value network every cycle and adaptively adjust the weight coefficients β1, β2, and β3.

[0150] Assume that control action a is defined as a = [solar power percentage, heat pump power percentage, valve opening adjustment], and the constraints include: solar power percentage + heat pump power percentage cannot exceed 100%, heat pump power percentage is greater than or equal to 20%, and valve opening adjustment cannot exceed 5%. Valve opening adjustment is mainly used to adjust the circulation pump flow or the amount of water added to the heat storage tank to balance the temperature difference between the collector and the water tank (or maintain a stable return water temperature).

[0151] According to the policy network π(a|S t ) The generated probability action distribution is: a1 = [70%, 30%, +3%], probability 0.6; a2 = [60%, 40%, +2%], probability 0.3; a3 = [50%, 50%, +1%], probability 0.1. The final output control action a t =【70%,30%,+3%】. If the maximum solar input power is 5KW and the maximum heat pump input power is 4KW, the final solar power is adjusted to 35KW and the heat pump power is adjusted to 12KW.

[0152] In the dynamic adaptation target generation function G t In [1], the reward function and the long-term value network correspond to short-term efficiency and long-term stability (such as thermal inertia compensation), respectively. The combination of the two breaks through the limitations of traditional single-objective optimization. This dynamically adaptive target generation function integrates real-time optimization (immediate reward function), long-term prediction (long-term value network), exploration mechanism (strategy entropy), and physical property compensation (Φ) through a reinforcement learning framework. thermal ), and dynamically adjust each weight in combination with weight coefficients β1, β2, and β3, which can achieve efficient adaptive control in complex environments and solve the problem of rigid preset targets in the energy-saving control method of the existing solar combined heat pump.

[0153] like Figure 5As shown, an energy-saving control system of a solar energy combined heat pump in one embodiment of the present invention is used to implement the energy-saving control method of the solar energy combined heat pump, including a characteristic parameter acquisition module, a characteristic vector acquisition module, a uniformity coefficient acquisition module, a control target acquisition module and an operation control module.

[0154] The characteristic parameter acquisition module is used to obtain the heat absorption power characteristics and heat release power characteristics of each heat pump in the solar heat pump group, as well as the ambient temperature characteristics, ambient humidity characteristics and solar radiation intensity characteristics; the characteristic vector acquisition module is used to obtain the thermal energy characteristic vector based on the heat absorption power characteristics, heat release power characteristics, ambient temperature characteristics, ambient humidity characteristics and solar radiation intensity characteristics.

[0155] The uniformity coefficient acquisition module is used to obtain heat absorption uniformity, heat release uniformity, pipeline resistance factor and equipment aging factor, and obtains the uniformity coefficient based on the heat absorption uniformity, heat release uniformity, pipeline resistance coefficient and equipment aging factor; the control target acquisition module is used to obtain the adaptive control target based on the thermal energy characteristic vector, uniformity coefficient and historical data; the operation control module is used to realize the operation control of the solar heat pump group according to the adaptive control target.

[0156] As a preferred technical solution, Figure 6 As shown, the uniformity coefficient acquisition module includes a thermal imbalance acquisition unit, a compensation value acquisition unit, an offset acquisition unit and a uniformity coefficient acquisition unit.

[0157] The thermal imbalance degree acquisition unit is used to obtain the thermal imbalance degree according to the heat absorption uniformity and the heat release uniformity; the compensation value acquisition unit is used to process the pipeline resistance factor and the equipment aging factor through a fuzzy logic method to obtain a uniformity compensation value.

[0158] The offset acquisition unit is used to acquire the environmental disturbance factor and the timing attenuation factor, and acquire the uniformity offset according to the environmental disturbance factor and the timing attenuation factor; the uniformity coefficient acquisition unit is used to acquire the uniformity coefficient according to the thermal imbalance, the uniformity compensation value and the uniformity offset.

[0159] like Figure 8 As shown, through experimental simulation of the uniform coefficient acquisition module, it can be seen that the uniform coefficient acquisition module in this embodiment is more outstanding in index performance than the existing technology, and the RMSE is reduced from 8.694 to 7.903.

[0160] The control target acquisition module includes a state vector acquisition unit, a reward function construction unit and a control target acquisition unit.

[0161] The state vector acquisition unit is used to acquire the state vector based on the thermal energy characteristic vector, the uniformity coefficient and the historical data; the reward function construction unit is used to acquire the system efficiency and construct the reward function based on the uniformity coefficient and the system efficiency; the control target acquisition unit is used to construct a dynamic adaptive target generation function based on the state vector and the reward function, and acquire the adaptive control target based on the dynamic adaptive target generation function.

[0162] The dynamic adaptation target generation function is expressed as:

[0163] G t =argmax a∈A [β1·R imm (S t ,a)+β2·E[V θ (S t+1 )]]+β3·Η(π(·|S t ))

[0164] Among them, S t represents the state vector, R imm Represents the reward function, a represents the control action, Vθ represents the long-term value network, H(π) represents the policy entropy, β1, β2, β3 represent the weight coefficients of the state vector, reward function and policy entropy respectively, A represents the action space, E and θ represent the expected value and policy network parameters respectively.

[0165] Reward Function in, Indicates the uniformity improvement rate, which is used to promote the balanced operation of the heat pump group. |η total -η opt | represents the system efficiency deviation, η total represents the real-time efficiency of the system, η opt Represents the theoretical optimal value of system efficiency, ΔP grid Indicates the grid power P grid The change in || ||2 represents the L2 norm, ||ΔP grid ||2 represents the grid power fluctuation penalty term, used to reduce grid impact. w1, w2, and w3 represent corresponding weight coefficients, which can be adjusted based on the priority of different scenarios, such as w1:w2:w3 = 0.6:0.3:0.1 or w1:w2:w3 = 0.5:0.4:0.1.

[0166] Long-term value network function V θ =LSTM θ (S t )+β Φ ·Φ thermal (T res ,m fluid ); Among them, LSTMθ represents a temporal feature extraction network for learning environmental dynamics, such as illumination periodicity. thermal Represents the thermal inertia compensation term, quantifying the fluid heat capacity m fluid and the water storage tank temperature T res The influence of β Φ It represents the attenuation coefficient of the thermal inertia compensation term and can be set based on experience.

[0167] The reward function construction unit includes a system parameter acquisition subunit, a weight coefficient acquisition subunit and a system efficiency acquisition subunit.

[0168] The system parameter acquisition subunit is used to obtain the useful heat output by the system, the total energy consumption of the system and the equipment aging correction coefficient; the weight coefficient acquisition subunit is used to obtain the energy allocation weight coefficient based on the ambient temperature characteristics, ambient humidity characteristics and solar radiation intensity characteristics; the system efficiency acquisition subunit is used to obtain the system efficiency based on the useful heat, the total energy consumption of the system, the equipment aging correction coefficient and the energy allocation weight coefficient.

[0169] The system efficiency Here, the system efficiency function is explained in detail. Environmental factors have a certain impact on the useful heat. When the temperature is low and the humidity is high, the heat loss increases and the useful heat may decrease. At this time, the energy allocation weight coefficient ω is required. t Improve the heat pump weight compensation. β represents the equipment aging correction coefficient, which reflects the impact of equipment efficiency attenuation on energy consumption. It is obtained through training of historical operation data and is used to correct E caused by equipment aging. total α represents the efficiency decay coefficient, which is defined as the degree to which the efficiency of a device decreases over time or with the number of times it is used (such as the power decay rate of a photovoltaic panel or the reduction in the COP of a heat pump).

[0170] The following further describes the algorithm selection and training process for obtaining the adaptive control target:

[0171] In the first step, for discrete action spaces, the Q-Learning algorithm can be used to train the target network through experience replay and stable training. The specific training iteration process of the Q-Learning algorithm is as follows: Figure 7 As shown in the figure, for continuous action space, PPO (Proximal Policy Optimization) or SAC (Soft Actor-Critic) algorithm is used to optimize continuous parameter adjustment using policy gradient.

[0172] In the second step, we first sample the initial state s0 from the historical data, execute action a0, obtain the reward R0 and the new state s1, and then use the experience replay pool to store (st ,a t ,R t ,s t+1 ).

[0173] The third step is to update the network parameters to maximize the cumulative reward.

[0174] In actual operation, you can enter the current s every 5 minutes t , output action a through the trained strategy network t , dynamically adjust the adaptive control target G t It should be noted that the adaptive control target can be understood as the most effective control target generated in real time, which is essentially the state vector S t Dynamically balance multi-dimensional indicators such as efficiency, stability, and response speed to avoid the rigidity of preset goals and simultaneously optimize indicators such as uniformity, system efficiency, and grid fluctuations.

[0175] G t As a comprehensive optimization objective, by maximizing β1·R imm (S t ,a)+β2·E[V θ (S t+1 )] directly guides action selection. Specifically, when the reward function R imm When the short-term uniformity demand is high, you can choose radical actions (such as increasing the heat pump power); when the long-term value network V θ Lead (such as long-term efficiency optimization), choose smooth actions (such as prioritizing the use of solar energy).

[0176] G t Embedding thermal inertia compensation items can prevent the action selection from exceeding the physical limit of the device and ensure that the selected action is dynamically adjusted within a safe range. By rolling optimization to correct the action sequence, G t Updating the target network parameters every cycle allows action selection to continuously adapt to environmental changes such as real-time fluctuations in solar radiation intensity.

[0177] In summary, the energy-saving control system maximizes the overall efficiency of the system by dynamically adjusting the working ratio of solar energy and heat pumps. During the control process, it combines the overall efficiency optimization goal of the system with the uniformity control goal of the heat pump group, thereby realizing multi-objective optimization control and improving the energy-saving control performance and overall operating efficiency of the system.

[0178] The various technical features of the embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0179] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the patent for this invention shall be determined by the appended claims.

Claims

1. A solar energy combined heat pump energy saving control method, characterized in that: The energy-saving control method includes: Obtaining heat absorption power characteristics and heat release power characteristics of each heat pump in the solar heat pump group; Acquire an ambient temperature characteristic, an ambient humidity characteristic, and a solar radiation intensity characteristic, and acquire a thermal energy characteristic vector according to the heat absorption power characteristic, the heat release power characteristic, the ambient temperature characteristic, the ambient humidity characteristic, and the solar radiation intensity characteristic; Obtaining heat absorption uniformity, heat release uniformity, pipeline resistance factor, and equipment aging factor, and obtaining a uniformity coefficient based on the heat absorption uniformity, heat release uniformity, pipeline resistance coefficient, and equipment aging factor; Obtaining an adaptive control target according to the thermal energy characteristic vector, uniformity coefficient and historical data; The operation of the solar heat pump group is controlled according to the adaptive control target.

2. The energy-saving control method of a solar combined heat pump according to claim 1, characterized in that: The specific methods for obtaining the uniformity coefficient include: Obtaining a degree of thermal imbalance according to the heat absorption uniformity and the heat release uniformity; The pipeline resistance factor and the equipment aging factor are processed by a fuzzy logic method to obtain a uniformity compensation value; Obtaining an environmental disturbance factor and a timing attenuation factor, and obtaining a uniformity offset according to the environmental disturbance factor and the timing attenuation factor; A uniformity coefficient is obtained according to the thermal imbalance, the uniformity compensation value, and the uniformity offset.

3. The energy-saving control method of a solar combined heat pump according to claim 2, characterized in that: The specific methods for obtaining the adaptive control target include: Acquire a state vector according to the thermal energy characteristic vector, uniformity coefficient and historical data; Obtaining system efficiency, and constructing a reward function based on the uniformity coefficient and the system efficiency; A dynamic adaptive target generation function is constructed according to the state vector and the reward function, and an adaptive control target is obtained based on the dynamic adaptive target generation function.

4. The energy-saving control method of a solar-heat-pump combined with heat pump according to claim 3, characterized in that: The specific method for obtaining system efficiency includes the following steps: Obtain the useful heat output by the system, the total energy consumption of the system, and the equipment aging correction factor; Obtaining an energy allocation weight coefficient based on the ambient temperature characteristics, ambient humidity characteristics, and solar radiation intensity characteristics; The system efficiency is obtained according to the useful heat, the total system energy consumption, the equipment aging correction factor and the energy allocation weight coefficient.

5. The energy-saving control method of a solar combined heat pump according to claim 4, characterized in that: According to the formula U=sigmoid(α1·f(P in ,P out )+α2·g(R,A)+α3·h(E d ,T d ))·C scale Get the uniformity coefficient U; Among them, f(P in ,P out )、g(R,A)、h(E d ,T d ) represent the thermal imbalance, uniformity compensation value and uniformity offset respectively, α1, α2, α3 represent the weight coefficients of the thermal imbalance, uniformity compensation value and uniformity offset respectively, P in 、P out They represent heat absorption uniformity and heat release uniformity respectively, R and A represent pipeline resistance factor and equipment aging factor respectively, E d 、T d Represent the environmental disturbance factor and the time series attenuation factor, C scale Indicates the scaling factor.

6. The energy-saving control method of a solar-heat-pump combined with heat pump according to claim 5, characterized in that: According to formula G t =argmax a∈A [β1·R imm (S t ,a)+β2·E[V θ (S t+1 )]]+β3·Η(π(·|S t )) Obtain a dynamic adaptation target generation function; Among them, S t represents the state vector, R imm represents the reward function, a represents the control action, V θ represents the long-term value network, H(π) represents the policy entropy, β1, β2, β3 represent the weight coefficients of the state vector, reward function and policy entropy respectively, A represents the action space, E and θ represent the expected value and policy network parameters respectively.

7. An energy-saving control system for a solar-energy combined heat pump, used to implement the energy-saving control method for a solar-energy combined heat pump according to any one of claims 1 to 6, characterized in that: The energy-saving control system includes: The characteristic parameter acquisition module is used to obtain the heat absorption power characteristics and heat release power characteristics of each heat pump in the solar heat pump group, as well as the ambient temperature characteristics, ambient humidity characteristics and solar radiation intensity characteristics; a feature vector acquisition module, configured to acquire a thermal energy feature vector based on the heat absorption power feature, the heat release power feature, the ambient temperature feature, the ambient humidity feature, and the solar radiation intensity feature; A uniformity coefficient acquisition module is used to obtain heat absorption uniformity, heat release uniformity, pipeline resistance factor and equipment aging factor, and obtain the uniformity coefficient based on the heat absorption uniformity, heat release uniformity, pipeline resistance coefficient and equipment aging factor; A control target acquisition module, configured to acquire an adaptive control target based on the thermal energy characteristic vector, uniformity coefficient, and historical data; An operation control module is used to realize operation control of the solar heat pump group according to the adaptive control target.

8. The energy-saving control system of a solar energy combined heat pump according to claim 7, characterized in that: The uniformity coefficient acquisition module includes: a thermal imbalance degree obtaining unit, configured to obtain the thermal imbalance degree according to the heat absorption uniformity and the heat release uniformity; a compensation value obtaining unit, configured to process the pipeline resistance factor and the equipment aging factor by a fuzzy logic method to obtain a uniformity compensation value; an offset acquiring unit, configured to acquire an environmental disturbance factor and a time series attenuation factor, and acquire a uniformity offset according to the environmental disturbance factor and the time series attenuation factor; The uniformity coefficient acquisition unit is used to acquire the uniformity coefficient according to the thermal imbalance degree, the uniformity compensation value and the uniformity offset.

9. The energy-saving control system of a solar energy combined heat pump according to claim 8, characterized in that: The control target acquisition module includes: A state vector acquisition unit, configured to acquire a state vector according to the thermal energy characteristic vector, the uniformity coefficient, and historical data; A reward function construction unit, configured to obtain system efficiency and construct a reward function based on the uniformity coefficient and the system efficiency; A control target acquisition unit is used to construct a dynamic adaptive target generation function according to the state vector and the reward function, and acquire an adaptive control target based on the dynamic adaptive target generation function.

10. The energy-saving control system of a solar energy combined heat pump according to claim 9, characterized in that: The reward function construction unit includes: System parameter acquisition subunit, used to obtain the useful heat output by the system, the total energy consumption of the system and the equipment aging correction factor; A weight coefficient acquisition subunit is used to obtain an energy allocation weight coefficient according to the ambient temperature characteristics, ambient humidity characteristics and solar radiation intensity characteristics; The system efficiency acquisition subunit is used to obtain the system efficiency according to the useful heat, the total system energy consumption, the equipment aging correction coefficient and the energy allocation weight coefficient.

Images