Heat pump waste heat recovery cooperative heat supply method and system

Abstract

The invention discloses a heat pump waste heat recovery cooperative heat supply method and system, and relates to the technical field of intelligent heat supply control, and the method comprises the steps: collecting environment data and heat pump operation data through a sensor, and carrying out the data preprocessing; modeling the operation state of the heat pump system by using a gradient lifting tree algorithm, constructing a heat pump load prediction model, and outputting load prediction; in combination with the load prediction result and the heat pump operation data, a collaborative heat supply strategy is obtained through an optimization algorithm; and waste heat recovery scheduling is carried out according to the load prediction result and the current waste heat recovery condition. According to the method, the accuracy and the real-time performance of load prediction can be remarkably improved, heat source configuration and heat supply requirements are highly matched, the waste heat utilization rate is improved to the maximum extent, power consumption is reduced, meanwhile, the indoor comfort degree and the system operation stability are guaranteed, finally, the operation cost is remarkably reduced, and the service life of equipment is prolonged.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent heating control, in particular to a heat pump waste heat recovery collaborative heating method and system. BACKGROUND

[0002] With the global energy shortage and the continuous improvement of carbon emission reduction targets, heat pump technology and waste heat recovery technology have attracted widespread attention in the field of building heating and industrial waste heat utilization. Traditional heat pump systems mainly use mechanical compression refrigeration cycle, which has a high coefficient of performance (COP) under medium and high temperature difference conditions, but its energy utilization efficiency is still limited by fixed compression ratio and simple control strategy. In recent years, multi-stage compression, adsorption heat pump and variable frequency speed regulation technology have been applied to heating systems, effectively improving the operation stability and energy efficiency level; at the same time, waste heat recovery technology has also developed from simple waste heat capture to waste heat driven combined heat and power, waste heat refrigeration / heat integration and other composite systems. With the rise of Internet of Things (IoT), big data and cloud computing, real-time data acquisition and monitoring platforms based on sensors have gradually been established, and intelligent technologies such as PI control, fuzzy control and model predictive control (MPC) have been applied to heat pump and waste heat recovery systems, realizing online monitoring and control of temperature, pressure and flow and other operating parameters, but there are still obvious deficiencies in load prediction accuracy, collaborative optimization and fault warning. In addition, significant progress has been made in heating operation state visualization and fault diagnosis technology based on big data and cloud platform, which realizes remote diagnosis and warning of system operation state through real-time monitoring and trend analysis of heating load and pipe network parameters.

[0003] Existing heat pump heating and waste heat recovery systems generally rely on experience rules or single PID control, and their load prediction mostly uses linear regression or statistical model-based methods, which are difficult to capture the nonlinear coupling characteristics of environmental conditions and heat demand, resulting in large prediction errors; the waste heat recovery scheduling and heat pump operation strategy are disconnected, lacking a collaborative scheduling mechanism, and unable to allocate heat sources according to load fluctuations and waste heat resources, resulting in energy waste and rising operating costs. At the same time, multi-objective optimization scheduling methods often rely on mixed integer programming (MIP) or linear programming models, which can achieve certain results in small-scale scenarios, but have high computational complexity and poor real-time performance when facing large-scale multi-heat sources, time-varying loads and nonlinear constraints, which is not conducive to online scheduling; traditional model objective functions mostly focus on single economic or energy-saving indicators, lacking multi-dimensional collaborative consideration of system reliability, equipment life and user comfort, and being difficult to meet the comprehensive optimization demand of energy consumption, emission and heating quality. In addition, in the aspect of waste heat recovery, existing technologies mostly stay at the static scheduling level, lacking a mechanism for dynamic matching and priority allocation based on heat pump load prediction results, resulting in insufficient utilization of waste heat resources. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a heat pump waste heat recovery and heating method to solve the technical problems of inaccurate load prediction, poor real-time online scheduling and lack of dynamic optimization of waste heat scheduling in the existing heat pump waste heat recovery and heating technology.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In the first aspect, the present application provides a heat pump waste heat recovery and heating method, which comprises collecting environmental data and heat pump operation data through a sensor, and performing data preprocessing;

[0008] The gradient boosting tree algorithm is used to model the operation state of the heat pump system, a heat pump load prediction model is constructed, and the load prediction is output.

[0009] The load prediction result and the heat pump operation data are combined, and a collaborative heating strategy is obtained through an optimization algorithm.

[0010] According to the load prediction result and the current waste heat recovery condition, waste heat recovery scheduling is performed.

[0011] As a preferred scheme of the heat pump waste heat recovery and heating method of the present application, wherein: the data preprocessing comprises: the data collected by the sensor is transmitted to the data processing unit through the Internet of Things device; the data collected by the sensor is denoised and missing value is interpolated, and is standardized;

[0012] The periodic time characteristics of the heat pump operation are extracted from the sensor data according to the time stamp, and the heat pump operation characteristics and load influencing factors are analyzed.

[0013] As a preferred scheme of the heat pump waste heat recovery and heating method of the present application, wherein: the gradient boosting tree algorithm comprises: constructing input features according to the heat pump operation characteristics and the load influencing factors to form an input feature data set, which specifically comprises: environmental features, historical load features and time series features;

[0014] The input feature data set is divided according to the time sequence to construct a GBT algorithm training data set; the XGBoost gradient boosting tree algorithm is selected, and the initialization parameters are set;

[0015] The gradient boosting tree algorithm is used to iteratively learn the training data set; a regression tree is constructed on the training set to predict the initial value of the heat pump load and obtain the initial prediction result;

[0016] The residual error between the predicted value after each iteration and the true value is calculated.

[0017] The new tree is trained to fit the residual error, and in the next iteration, the current residual error is used as the target value to train the next regression tree to reduce the prediction error of the last iteration, forming a model ensemble learning;

[0018] The model prediction result is updated, the prediction result is updated using the new regression tree, and the prediction accuracy of the model as a whole is improved by continuously reducing the prediction error;

[0019] Repeat the iteration training until the model reaches the set stop condition;

[0020] After training is completed, the model performance is preliminarily evaluated through the validation set to determine whether hyperparameter adjustment is needed; the model accuracy is calculated using the validation set; the cross-validation and grid search methods are used to adjust the key hyperparameters; and according to the evaluation result, the optimal model hyperparameter combination is selected.

[0021] As a preferred scheme of the heat pump waste heat recovery and heat supply method, the heat pump load prediction model is constructed by continuously collecting the latest data from the environment sensor and the heat pump controller, and inputting the preprocessed real-time data into the trained gradient boosting tree model; and a use heat load prediction value sequence in a time period is output.

[0022] The load prediction sequence is exponentially weighted and smoothed, and is corrected in combination with the latest actual load measurement value.

[0023] The optimization target is to minimize the total energy consumption under the conditions of indoor comfort and safe operation of the system.

[0024] The constraint conditions include that the room temperature must be kept within the user set range.

[0025] The minimum interval of device protection must be met between each start and stop.

[0026] The total power must not exceed the capacity of the standby power supply.

[0027] According to the smoothed and corrected load prediction result, the entire prediction period is divided into a plurality of control periods; in each control period, the heat pump optimal start-stop period and the target outlet water temperature are calculated to match the predicted load and the heat supply capacity; and an optimization algorithm is used to traverse a limited start-stop and temperature setting combination to select a set of schemes that meet the constraints and have the minimum energy consumption.

[0028] As a preferred scheme of the heat pump waste heat recovery and heat supply method, the optimization algorithm includes generating a batch of initial scheme sets randomly within the range that meets the constraints; calculating the operation cost index of each scheme as the fitness value; and if the scheme temperature is not up to standard, excessive switching occurs, or the energy balance is exceeded, a penalty score is added to the fitness.

[0029] Each scheme records the parameter combination at the lowest cost; in each iteration, each scheme is adjusted from two aspects: one is to approach the best parameter combination in the history of the scheme itself, and the other is to approach the best parameter combination in the current global, continuously generating new scheme candidates;

[0030] After generating a new scheme each time, if the heat pump proportion of certain time period exceeds the reasonable interval, it is immediately trimmed to the boundary value; if the scheme still violates the continuous operation or energy balance coupling constraint, the scheme is adjusted back to the nearest feasible solution region through projection, and the iteration process is continued until the preset maximum number of iterations is reached.

[0031] As a preferred scheme of the heat pump waste heat recovery and heat supply method, wherein: the obtained heat supply strategy includes obtaining the heat pump load prediction value and the current heat capacity value of the waste heat recovery system based on the gradient boosting tree model;

[0032] For each prediction period, the maximum heat supply that the waste heat recovery system can meet in the waste heat priority mode is calculated, and the initial allocation ratio of the heat pump and the waste heat recovery is determined; the initial allocation ratio is determined according to the comprehensive degree of satisfaction and economy;

[0033] According to the optimization target, the power consumption of the heat pump system and the waste heat utilization cost are taken as optimization items; the objective function considers the electricity price, system conversion efficiency and waste heat utilization rate parameters of each period to quantitatively evaluate the allocation ratio;

[0034] Under the premise of meeting the constraint condition, an optimization algorithm is used to repeatedly adjust the allocation ratio of the heat pump and the waste heat recovery, the objective function value is recalculated each time, and whether to accept the current solution is judged according to the descending amplitude of the objective function; when the objective function no longer significantly decreases, the optimal allocation scheme is determined.

[0035] As a preferred scheme of the heat pump waste heat recovery and heat supply method, wherein: the waste heat recovery scheduling includes simulating and checking the preliminary generated scheduling scheme, and if it is found that the scheme will cause the room temperature to exceed the set range, the outlet water temperature of the scheme cycle is fine-tuned and the running time is extended;

[0036] The finally determined start-stop timing and temperature setting are issued to the on-site heat pump controller through the control bus; the on-off operation and temperature adjustment are automatically performed at the corresponding time to ensure that the load and the prediction scheduling are synchronized;

[0037] According to the finally determined allocation ratio, generate scheduling instructions, including the start-stop time of the heat pump in each period, the output temperature setting, and the flow regulation instruction of the waste heat recovery system, in the form of time sequence, and issue to the on-site controller for adjustment.

[0038] In a second aspect, the present application provides a heat pump waste heat recovery and collaborative heating system, comprising a data acquisition and preprocessing module, a load prediction module, a collaborative heating strategy optimization module, and a waste heat recovery scheduling module.

[0039] The data acquisition and preprocessing module acquires environmental and equipment data in real time through a running state acquisition unit of the heat pump and temperature, humidity, pressure, and flow sensors arranged indoors and outdoors.

[0040] The load prediction module comprises a data processing unit that predicts the heating load in each control period based on the preprocessed environmental characteristics, historical load sequence, and time characteristics.

[0041] The collaborative heating strategy optimization module automatically searches for the optimal combination of heat pump start and stop periods and outlet water temperature through a modified swarm intelligence optimization algorithm.

[0042] The waste heat recovery scheduling module schedules the waste heat recovery system according to the heat pump scheduling results and the available heat of the waste heat recovery system.

[0043] In a third aspect, the present application provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and wherein the computer program is executed by the processor to implement any step of the heat pump waste heat recovery and collaborative heating method according to the first aspect of the present application.

[0044] In a fourth aspect, the present application provides a computer readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement any step of the heat pump waste heat recovery and collaborative heating method according to the first aspect of the present application.

[0045] The present application has the following advantages: through real-time acquisition and preprocessing of environmental and heat pump running data, accurate load prediction is performed using a gradient boosting tree algorithm, and then optimization scheduling is performed based on the prediction results and actual running state, finally realizing collaborative heating of the heat pump and waste heat recovery system. This method can significantly improve the accuracy and real-time performance of load prediction, highly match heat source configuration and heating demand, maximize waste heat utilization rate and reduce power consumption, while ensuring indoor comfort and system running stability, ultimately significantly reducing operating costs and prolonging equipment life. BRIEF DESCRIPTION OF DRAWINGS

[0046] To more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0047] Figure 1 Flow chart of heat pump waste heat recovery and heat supply method. DETAILED DESCRIPTION

[0048] In order to make the above objectives, features and advantages of the present application more apparent, specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0049] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. However, it will be apparent to one skilled in the art that the present application can be practiced without the specific details given herein, that the present application can be practiced with other than the described implementations, and that variations from the particular examples given can be made and practiced within the scope of the present application. Accordingly, the particular implementation given above is illustrative only as the present application can be practiced with a wide variety of specific implementations that are apparent to those skilled in the art.

[0050] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an embodiment that is independent or alternative to other embodiments.

[0051] Referring to the drawings, one embodiment of the present application provides a heat pump waste heat recovery and heat supply method, which comprises the following steps:

[0052] S1: Collecting environmental data and heat pump operation data through sensors and performing data preprocessing.

[0053] At each control time, five types of original data, i.e. indoor temperature, outdoor temperature, indoor humidity, heat pump compressor power and actual load in the previous period, are collected in real time through temperature and humidity sensors arranged indoors and outdoors, and current, voltage and flow sensors on the heat pump controller.

[0054] The collected original data are processed in the following order:

[0055] Median filtering denoising: for each sensor signal, the median value of the data in a fixed-width sliding window (such as a 3-point or 5-point window) is calculated, and the median value is used to replace the observation value at the center point of the window to eliminate occasional impulse noise.

[0056] Missing value interpolation: when the sensor data at a certain time is missing, linear interpolation is performed using the arithmetic mean of the valid data at adjacent times to ensure the continuity of the data sequence.

[0057] Standardization processing: each feature value after denoising and interpolation is subtracted from its historical mean and divided by the historical standard deviation, so that all feature values are distributed in the same dimension, eliminating the influence of dimension difference on model training.

[0058] Periodic feature extraction: daily and weekly period features are extracted according to the timestamp, and the time information is converted into numerical features that can be directly used for model training through sine and cosine encoding forms, revealing the operation law of the heat pump system at different time periods.

[0059] Let t be the time, and the vector collected by the sensor is:

[0060] x t = [T in (t), T out (t), H(t), P(t), Q hist (t-1)]

[0061] Wherein, T in (t) represents indoor temperature, T out (t) represents outdoor temperature, H(t) represents indoor humidity, P(t) represents heat pump compressor power, and Q hist (t-1) represents the actual load of the previous period.

[0062] De-noising and missing value interpolation, median filter de-noising:

[0063]

[0064] Wherein, (window width 2k+1, such as k=1).

[0065] Missing value interpolation, if Missing, then:

[0066]

[0067] Further standardization processing:

[0068]

[0069] Wherein, mu i , sigma i represent the mean and standard deviation of feature i on historical data.

[0070] Further, the data preprocessing includes that the data collected by the sensor is transmitted to the data processing unit through the Internet of Things device; the data collected by the sensor is de-noised and missing value interpolated, and is standardized.

[0071] Periodic time features of heat pump operation are extracted from sensor data according to the timestamp, and heat pump operation characteristics and load influencing factors are analyzed.

[0072] It should be noted that through multi-level preprocessing, including denoising, interpolation, standardization and periodic feature extraction, the system obtains high-quality, scale-consistent and operation rule-containing input data, providing a reliable foundation for subsequent load prediction and dispatching optimization.

[0073] S2: using gradient boosting tree algorithm to model the operation state of the heat pump system, constructing a heat pump load prediction model, and outputting the load prediction.

[0074] Further, the standardized environmental features, historical load features and periodic time features obtained in step S1 are combined into a model input feature vector to form a training data set.

[0075] The data set is divided into training set and validation set according to time sequence, ensuring the time sequence consistency of training process and validation process.

[0076] The XGBoost algorithm is used to initialize the gradient boosting tree model, the loss function is set to mean square error, and the best tree depth, learning rate, leaf node number and other hyperparameters are determined through grid search and cross-validation.

[0077] In each iteration, according to the residual error between the prediction results of the current model on the training set and the true load, a new weak regression tree is automatically generated to fit the residual error, and the new tree output is weighted and accumulated with a learning rate, so as to gradually reduce the overall prediction error.

[0078] After training, the model performance is evaluated on the validation set, and according to the mean square error and determination coefficient on the validation set, it is determined whether to adjust the hyperparameters or stop training.

[0079] The gradient boosting tree algorithm includes constructing input features according to heat pump operation characteristics and load influencing factors to form an input feature data set, which specifically includes environmental features, historical load features and time sequence features.

[0080] The input feature set is constructed, and the formula is:

[0081] z t =[x′ t ,h t ,Q hist (t-1)]

[0082] Where the target value is y t = Q true (t) (true load).

[0083] The input feature data set is divided according to time sequence to construct the GBT algorithm training data set; XGBoost gradient boosting tree algorithm is selected, and the initialization parameters are set.

[0084] Gradient boosting tree algorithm is used to iteratively learn the training data set; regression trees are constructed on the training set to predict the initial value of the heat pump load, and the initial prediction result is obtained.

[0085] Initialize the model:

[0086]

[0087] (Usually L takes the mean square error Calculate the residual error between the predicted value and the true value after each iteration.

[0088] Train new trees to fit the residual error, and in the next iteration, train the next regression tree with the current residual error as the target value to reduce the prediction error of the last iteration, forming a model ensemble learning.

[0089] Update the model prediction result, update the prediction result using the new regression tree, and continuously reduce the prediction error to improve the overall prediction accuracy of the model.

[0090] Iteratively construct regression trees, and for the mth tree (m = 1, …, M), calculate the residual error, which is expressed as:

[0091]

[0092] Fit new trees to find regression trees f m (z) that minimize the residual error:

[0093]

[0094] Update the model, which is expressed as:

[0095] F m (z) = F m-1 (z) + ηf m (z)

[0096] Where η represents the learning rate (e.g. 0.1).

[0097] Repeat the iteration training until the model reaches the set stopping condition.

[0098] After training is completed, the model performance is preliminarily evaluated through the validation set to determine whether hyperparameter adjustment is needed; the model accuracy is calculated using the validation set; cross-validation and grid search methods are used to adjust key hyperparameters; according to the evaluation results, the optimal model hyperparameter combination is selected.

[0099] The heat pump load prediction model is constructed by continuously collecting the latest data from the environmental sensors and the heat pump controller, inputting the preprocessed real-time data into the trained gradient boosting tree model, and outputting a sequence of heat load prediction values within a time period.

[0100] The load prediction sequence is exponentially weighted smoothed and corrected with the latest actual load measurement.

[0101] The optimization goal is to minimize the total energy consumption under the conditions of indoor comfort and system safe operation.

[0102] The constraint conditions include that the room temperature must be maintained within the user set range.

[0103] The minimum interval for equipment protection must be met between each start and stop.

[0104] The total power must not exceed the capacity of the backup power supply.

[0105] According to the smoothed and corrected load prediction results, the entire prediction period is divided into several control periods; in each control period, the optimal start-stop period of the heat pump and the target outlet water temperature are calculated to match the predicted load and the heating capacity; an optimization algorithm is used to traverse the limited start-stop and temperature setting combinations to select a set of schemes that meet the constraints and have the minimum energy consumption.

[0106] It should be noted that the finally determined gradient boosting tree model is deployed to the real-time data processing unit, continuously receives the latest preprocessed data from the sensor, and outputs the heat load prediction value sequence for one or more future periods in real time, and performs exponential weighted smoothing processing on the prediction sequence, and error correction combined with the latest actual load measurement. Through the iterative residual fitting mechanism of the gradient boosting tree algorithm and the strict cross-validation and hyperparameter optimization strategy, high-precision prediction of the heat pump load demand can be realized while ensuring the generalization ability of the model.

[0107] S3: Obtain the collaborative heating strategy through the optimization algorithm combined with the load prediction results and the heat pump operation data.

[0108] Further, the load prediction value sequence obtained in step S2 and the available heat capacity of the waste heat recovery system are used as inputs to the optimization algorithm.

[0109] The proportion of the heat pump in each prediction period is regarded as the particle position vector of the particle swarm optimization algorithm, and the position and speed of each particle are randomly initialized.

[0110] Define the fitness function: based on the consideration of electricity price, waste heat utilization cost and temperature comfort, etc., the weighted total energy consumption cost is taken as the target, and a penalty function is applied to the particles that violate the minimum continuous operation time, temperature upper and lower limits and energy balance constraints.

[0111] The optimization algorithm includes, within the scope of meeting the constraints, randomly generating a batch of initial scheme set; for each scheme, calculating its operation cost index as the fitness value; if the scheme temperature is not up to standard, excessive switching and exceeding energy balance appear violation, then superimpose penalty points on the fitness.

[0112] Each scheme records the parameter combination at the lowest cost; in each iteration, each scheme is adjusted from two aspects: one is to approach the best parameter combination in the history of the scheme itself, and the other is to approach the best parameter combination in the current global, continuously generating new scheme candidates.

[0113] After generating a new scheme each time, if the heat pump proportion of some time period exceeds the reasonable interval, it is immediately trimmed to the boundary value; if the scheme still violates the continuous operation or energy balance coupling constraint, the scheme is adjusted back to the nearest feasible solution region through projection, and the iteration process is continued until the preset maximum iteration number is reached.

[0114] The obtained collaborative heating strategy includes obtaining the heat pump load prediction value and the current available heat capacity value of the heat recovery system based on the gradient boosting tree model.

[0115] The heat pump allocation proportion vector X = [x1, x2, …, x T ] of each period is regarded as a “particle”. Each component x t represents the proportion of load borne by the heat pump in the t period, with a value range of [0, 1].

[0116] Randomly generate N particles in the feasible region (the region where all constraints are met at the same time), and randomly initialize the corresponding velocity vector V for each particle.

[0117] Perform fitness evaluation (set a constraint penalty function), for each particle X (i) , calculate its objective function value:

[0118]

[0119] Where L t is the total load prediction in the t period, c elec , c heat are the electricity price and waste heat cost respectively.

[0120] For each constraint (such as minimum / maximum temperature, minimum continuous operation time, energy balance, etc.), construct a penalty function term, when the particle X (i) violates the constraint, a large penalty value is added to the fitness, to ensure that the feasible solution is selected first.

[0121] Update the historical optimum and global optimum, for each particle, record the position with the lowest (optimal) fitness in its history

[0122] Find the position with the lowest fitness among all particles, denoted as global optimum G best .

[0123] Update the velocity and position of the i-th particle using the following formula:

[0124]

[0125] X (i) ←X (i) +V (i)

[0126] where w is the inertia weight, c1, c2 are learning factors, and r1, r2 are random numbers.

[0127] After updating, if a component of X exceeds [0, 1], it is directly clipped to the interval boundary.

[0128] If the overall vector X (i) still violates the energy balance or the minimum continuous operation time coupling constraints, use projection repair: project X (i) to the nearest feasible region point to ensure that all constraints are satisfied.

[0129] Repeat the above steps of "fitness evaluation - update optimum - update velocity and position - constraint repair" until the maximum number of iterations is reached, or the fitness of the global optimum G best converges (changes very little) within a certain number of generations.

[0130] Output the optimal scheme, and take the final G best as the optimal allocation ratio of the heat pump and waste heat recovery system in each period, generate scheduling instructions and issue them to the field controller.

[0131] For each prediction period, calculate the maximum heat supply that can be met by the waste heat recovery system in the waste heat priority mode, and determine the initial allocation ratio of the heat pump and waste heat recovery; the initial allocation ratio is determined according to the degree of satisfaction and economy.

[0132] According to the optimization objective, the power consumption of the heat pump system and the waste heat utilization cost are taken as the optimization items; the objective function considers the electricity price, system conversion efficiency, and waste heat utilization rate parameters in each period to quantitatively evaluate the allocation ratio.

[0133] Under the premise of satisfying the constraint conditions, an optimization algorithm is used to repeatedly adjust the allocation ratio of the heat pump and waste heat recovery, and the objective function value is recalculated each time. According to the descending amplitude of the objective function, it is judged whether to accept the current solution. When the objective function no longer significantly decreases, it is determined that the optimal allocation scheme is found.

[0134] It should be noted that in the iteration process, the particles update the speed and position according to the inertia weight, individual historical optimum and global optimum direction, and clip the components exceeding the range, project and repair the overall position vector of the coupling constraint, and ensure that each iteration is in the feasible region. Repeat the steps of "fitness evaluation - update the optimum - speed and position update - constraint repair" until the maximum number of iterations or the global optimum fitness converges, and output the final optimal load distribution ratio of the heat pump and waste heat recovery system. It should be noted that the introduction of the particle swarm optimization algorithm can quickly search for the global or approximate optimal collaborative heating scheme under the premise of meeting multiple coupling constraints with lower computational complexity, and adapt to system operation changes.

[0135] S4: According to the load prediction result and the current waste heat recovery condition, waste heat recovery scheduling is performed.

[0136] Further, the optimal distribution ratio output by step S3 is simulated and verified. If the simulation result shows that the room temperature may exceed the user set range, the corresponding period is fine-tuned for the outlet water temperature setting or the running time is extended, and re-evaluation is performed. The final scheduling instruction is generated, including the heat pump start-stop time sequence, target outlet water temperature and flow regulation parameter of the waste heat recovery system of each control period.

[0137] The waste heat recovery scheduling includes simulating and verifying the preliminary generated scheduling scheme. If it is found that the scheme will cause the room temperature to exceed the set range, the outlet water temperature of the scheme period is fine-tuned and the running time is extended.

[0138] The final determined start-stop time sequence and temperature setting are issued to the on-site heat pump controller through the control bus; at the corresponding time, automatic on-off operation and temperature adjustment are performed to ensure that the load and the predicted scheduling are kept in synchronization.

[0139] According to the final determined distribution ratio, a scheduling instruction is generated, including the start-stop time of the heat pump in each period, the output temperature setting, and the flow regulation instruction of the waste heat recovery system, which is issued to the on-site controller in the form of time sequence for adjustment.

[0140] It should be noted that the scheduling instruction is issued to the heat pump controller and the waste heat recovery device through the on-site control bus, and the on-off and temperature / flow adjustment operations are automatically performed according to the instruction to ensure that the system output is kept in synchronization with the predicted scheduling; during the scheduling execution process, the indoor temperature and system running state are continuously monitored, and feedback data is sent back to the data processing unit in time to provide a basis for the next round of load prediction and scheduling optimization. Through the combination of simulation verification and on-site real-time execution, it can be ensured that the final scheduling scheme can not only meet the user's requirements for room temperature comfort, but also ensure the stable operation of the system under safe and economic conditions.

[0141] The embodiment also provides a heat pump waste heat recovery and heat supply system, which comprises a data acquisition and preprocessing module, a load prediction module, a collaborative heat supply strategy optimization module and a waste heat recovery scheduling module.

[0142] The data acquisition and preprocessing module acquires environmental and equipment data in real time through a running state acquisition unit of the heat pump and temperature, humidity, pressure and flow sensors arranged indoors and outdoors; and the data acquired by the sensors is transmitted to a data processing unit through an Internet of Things device.

[0143] The load prediction module comprises a data processing unit, which predicts the heat consumption load in each control period based on the preprocessed environmental characteristics, historical load sequence and time characteristics.

[0144] The collaborative heat supply strategy optimization module automatically searches for the optimal combination of heat pump start and stop time periods and outlet water temperature through a modified swarm intelligence optimization algorithm.

[0145] The waste heat recovery scheduling module schedules the waste heat recovery system according to the heat pump scheduling result and the available heat of the waste heat recovery system.

[0146] The embodiment also provides a computer device suitable for the heat pump waste heat recovery and heat supply method, which comprises a memory and a processor; the memory is used for storing computer executable instructions, and the processor is used for executing the computer executable instructions to realize the heat pump waste heat recovery and heat supply method proposed in the above embodiment.

[0147] The computer device can be a terminal, and the computer device comprises a processor, a memory, a communication interface, a display screen and an input device which are connected through a system bus. The processor of the computer device is used for providing computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used for wired or wireless communication with external terminals. The wireless communication can be realized through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse.

[0148] The embodiment also provides a storage medium on which a computer program is stored, the program being executed by a processor to implement the method for realizing heat pump waste heat recovery and heat supply cooperation according to the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk.

[0149] To sum up, the method for realizing heat pump waste heat recovery and heat supply cooperation according to the present application can significantly improve the accuracy and real-time performance of load prediction, make the heat source configuration highly match the heat supply demand, maximize the waste heat utilization rate and reduce the power consumption, while ensuring the indoor comfort and system operation stability, thereby significantly reducing the operation cost and prolonging the equipment life.

[0150] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application, and although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and all of them should be covered in the scope of the claims of the present application.

Claims

1. A heat pump waste heat recovery and heating method, characterized in that: The method comprises the following steps: collecting environmental data and heat pump operation data through sensors, and performing data preprocessing; The operation state of the heat pump system is modeled using a gradient boosting tree algorithm, a heat pump load prediction model is constructed, and a load prediction is output; A coordinated heating strategy is obtained by combining the load prediction result with the heat pump operation data through an optimization algorithm; According to the load prediction result and the current waste heat recovery condition, the waste heat recovery is dispatched.

2. The heat pump waste heat recovery and heating method of claim 1, wherein: The data preprocessing includes: the data collected by the sensors is transmitted to the data processing unit through the Internet of Things device; the data collected by the sensors is denoised and missing value is interpolated, and is standardized; The periodic time characteristics of the heat pump operation are extracted from the sensor data according to the time stamp, and the heat pump operation characteristics and load influencing factors are analyzed.

3. The heat pump waste heat recovery and heating method of claim 2, wherein: The gradient boosting tree algorithm includes: constructing input features according to the heat pump operation characteristics and load influencing factors to form an input feature data set, which specifically includes environmental features, historical load features and time sequence features; The input feature data set is divided according to the time sequence to construct a GBT algorithm training data set; The XGBoost gradient boosting tree algorithm is selected, and the initialization parameters are set; The gradient boosting tree algorithm is used to iteratively learn the training data set; A regression tree is constructed on the training set to predict the initial value of the heat pump load and obtain an initial prediction result; The residual error between the predicted value after each iteration and the true value is calculated; A new tree is trained to fit the residual error, and in the next iteration, the current residual error is taken as the target value to train the next regression tree to reduce the prediction error of the last iteration, forming a model ensemble learning; The model prediction result is updated, the prediction result is updated using the new regression tree, and the prediction accuracy of the model is improved by continuously reducing the prediction error; Repeat the iteration training until the model reaches the set stop condition; After the training is completed, the model performance is preliminarily evaluated through the validation set to determine whether the hyperparameter adjustment is needed; the model accuracy is calculated using the validation set; the key hyperparameters are adjusted using cross-validation and grid search methods; according to the evaluation result, the optimal model hyperparameter combination is selected.

4. The heat pump waste heat recovery and heating method of claim 3, wherein: The heat pump load prediction model is constructed by continuously collecting the latest data from the environmental sensors and the heat pump controller, inputting the preprocessed real-time data into the trained gradient boosting tree model, and outputting a heat load prediction value sequence within a time period; The load prediction sequence is exponentially weighted and smoothed, and is corrected in combination with the latest actual load measurement value; The optimization objective is to minimize the total energy consumption under the conditions of indoor comfort and safe operation of the system; The constraint conditions include: the room temperature must be kept within the user's set range; The minimum interval for device protection must be met between each start and stop; The total power must not exceed the capacity of the standby power supply; According to the smoothed and corrected load prediction result, the entire prediction period is divided into several control periods; in each control period, the optimal start-stop period and the target outlet water temperature of the heat pump are calculated to match the predicted load with the heating capacity; an optimization algorithm is used to traverse the limited start-stop and temperature setting combinations, and a set of schemes that meet the constraints and have the minimum energy consumption is selected.

5. The heat pump waste heat recovery and heating method of claim 4, wherein: The optimization algorithm includes, within the scope of meeting the constraints, randomly generating a batch of initial scheme set; for each scheme, calculate its running cost index as fitness value; if the scheme temperature is not up to standard, excessive switching and energy balance violation occurs, then superimpose penalty points on the fitness; Each scheme records the parameter combination at the lowest cost; in each iteration, adjust each scheme from two aspects: one is to approach the historical best parameter combination of the scheme itself, and the other is to approach the current global best parameter combination, continuously generating new scheme candidates; After generating a new scheme each time, if the heat pump proportion of some time period exceeds the reasonable interval, it is immediately trimmed to the boundary value; if the scheme still violates the continuous operation or energy balance coupling constraint, adjust the scheme back to the nearest feasible solution region through projection method, continue the iteration process until the preset maximum iteration number is reached.

6. The heat pump waste heat recovery and heating method of claim 5, wherein: The obtained collaborative heating strategy includes obtaining the heat pump load prediction value obtained based on the gradient boosting tree model and the heat capacity value available for the current heat recovery system; For each prediction period, calculate the maximum heat supply that the heat recovery system can meet in the heat recovery priority mode, and determine the initial allocation ratio of the heat pump and the heat recovery; The initial allocation ratio is determined according to the degree of satisfaction and economy; According to the optimization target, the power consumption of the heat pump system and the heat recovery cost are taken as optimization items; the objective function considers the electricity price, system conversion efficiency and heat recovery rate parameters of each period to quantitatively evaluate the allocation ratio; Under the premise of meeting the constraint conditions, an optimization algorithm is used to repeatedly adjust the allocation ratio of the heat pump and the heat recovery, the objective function value is recalculated each time, and whether to accept the current solution is judged according to the descending amplitude of the objective function; when the objective function no longer significantly decreases, it is determined that the optimal allocation scheme is found.

7. The heat pump waste heat recovery and heating method of claim 6, wherein: The heat recovery scheduling includes simulating and checking the preliminary generated scheduling scheme, and if it is found that the scheme will cause the room temperature to exceed the set range, the outlet water temperature of the scheme cycle is fine-tuned and the running time is extended; The finally determined start-stop timing and temperature setting are sent to the on-site heat pump controller through the control bus; Automatic on-off operation and temperature adjustment are performed at the corresponding time to ensure that the load and the prediction scheduling are synchronized; According to the finally determined allocation ratio, generate scheduling instructions, including the start-stop time of the heat pump in each period, the output temperature setting, and the flow regulation instruction of the heat recovery system, and send them to the on-site controller in the form of time sequence for adjustment.

8. A heat pump waste heat recovery and heating system based on the heat pump waste heat recovery and heating method according to any one of claims 1 to 7, characterized in that: It includes a data acquisition and preprocessing module, a load prediction module, a collaborative heating strategy optimization module, and a heat recovery scheduling module. The data acquisition and preprocessing module acquires environmental and equipment data in real time through the running state acquisition unit of the heat pump and the temperature, humidity, pressure and flow sensors arranged indoors and outdoors; the data collected by the sensors is transmitted to the data processing unit through the Internet of Things device; The load prediction module includes a data processing unit, which predicts the heating load in each control period based on the preprocessed environmental characteristics, historical load sequence and time characteristics; The synergistic heat supply strategy optimization module automatically searches for the optimal combination of the heat pump start-stop period and the outlet water temperature through a modified swarm intelligence optimization algorithm. The waste heat recovery scheduling module is based on the heat pump scheduling result and the available heat of the current waste heat recovery system. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is characterized in that: The processor executes the computer program to realize the steps of the heat pump waste heat recovery synergistic heat supply method of any one of claims 1-7.

10. A computer readable storage medium having stored thereon a computer program, characterized in that: The computer program is executed by the processor to realize the steps of the heat pump waste heat recovery synergistic heat supply method of any one of claims 1-7.

Images