A method and apparatus for load distribution of a water chiller

Abstract

The application discloses a water chilling unit load distribution method and device, and relates to the technical field of water chilling units. The method comprises the following steps: obtaining target working condition data of multiple working moments associated with multiple water chilling units at a current running moment; calculating corresponding cooling loads according to the target working condition data, and constructing a total cooling load sequence according to each working moment based on the cooling loads; decomposing and reconstructing the total cooling load sequence by using a singular spectrum analysis algorithm to determine a reconstructed total cooling load sequence and a noise component; performing load prediction according to the reconstructed total cooling load sequence by using a preset total cooling load prediction model, and outputting a predicted total cooling load in combination with the noise component; and outputting the load rate of each water chilling unit by using a preset load distribution control reinforcement learning model based on the target working condition data, the cooling load and the predicted total cooling load at the current running moment. The load distribution flexibility of the water chilling unit can be improved in the load distribution process of the whole water chilling unit.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of water chiller, and particularly relates to a water chiller load distribution method and device. BACKGROUND

[0002] According to the data of the United Nations Environment Programme, the energy consumption of buildings accounts for about 40% of the total global energy consumption, among which the heating, ventilation and air conditioning system (HVAC) for heating, refrigeration and ventilation accounts for about 50% of the total energy consumption of buildings, and the energy consumption of water chiller as the main cooling equipment of the heating, ventilation and air conditioning system usually accounts for more than half of the total energy consumption of the heating, ventilation and air conditioning system, so optimizing the load distribution of water chiller has important energy-saving significance.

[0003] The traditional water chiller load distribution method is mainly based on meta-heuristic algorithm, and in order to ensure that the result does not fall into local optimum, an additional mechanism is usually designed to ensure the effectiveness of global search, thereby leading to exponential growth of overall complexity, and when the environmental conditions change, there is obvious hysteresis problem from the issuance of control instruction to the chilled water reaching the end device and realizing temperature regulation, resulting in low flexibility of water chiller load distribution. SUMMARY

[0004] The present application provides a water chiller load distribution method and device, which solves the technical problem that the traditional water chiller load distribution method is mainly based on meta-heuristic algorithm, and an additional mechanism is usually designed to ensure the effectiveness of global search, thereby leading to exponential growth of overall complexity, resulting in low flexibility of water chiller load distribution.

[0005] The present application provides a water chiller load distribution method and device, which solves the technical problem that the traditional water chiller load distribution method is mainly based on meta-heuristic algorithm, and an additional mechanism is usually designed to ensure the effectiveness of global search, thereby leading to exponential growth of overall complexity, resulting in low flexibility of water chiller load distribution.

[0006] Target working condition data of multiple running time instants associated with multiple water chillers at a current running time instant are respectively acquired;

[0007] The target working condition data are used to calculate corresponding cooling load, and a total cooling load sequence is constructed according to the running time instants based on the cooling load;

[0008] The total cooling load sequence is decomposed and reconstructed by using singular spectrum analysis algorithm to determine a reconstructed total cooling load sequence and a noise component;

[0009] The reconstructed total cooling load sequence is used to perform load prediction by using a preset total cooling load prediction model, and the noise component is combined to output a predicted total cooling load;

[0010] Based on the target working condition data of the current running time instant, the cooling load and the predicted total cooling load, a load distribution control reinforcement learning model is used to output the load rate of each water chiller.

[0011] Optionally, before the step of acquiring target working condition data of multiple running time instants respectively associated with the multiple chiller units at the current running time instant, comprising:

[0012] Collecting initial working condition data of multiple running time instants within a preset time length at the current running time instant respectively for the multiple chiller units;

[0013] Performing data rejection processing and data completion processing on each of the initial working condition data to obtain corresponding target working condition data.

[0014] Optionally, the step of calculating a cooling load for each of the target working condition data and constructing a total cooling load sequence according to each of the running time instants based on the cooling load, comprising:

[0015] Extracting a target chilled water supply temperature, a target chilled water return temperature and a target chilled water flow rate from each of the target working condition data respectively;

[0016] Performing difference operation on each of the target chilled water return temperature and the corresponding target chilled water supply temperature to obtain a plurality of chilled water temperature differences;

[0017] Performing multiplication operation on each of the chilled water temperature difference, the associated target chilled water flow rate and the specific heat of water to output the cooling load of each of the chiller units at each of the running time instants;

[0018] Performing summation operation on the cooling load of each of the chiller units according to each of the running time instants to determine a corresponding total cooling load;

[0019] Constructing a total cooling load sequence according to the corresponding running time instants based on each of the total cooling load.

[0020] Optionally, the step of decomposing and reconstructing the total cooling load sequence by using a singular spectrum analysis algorithm to determine a reconstructed total cooling load sequence and a noise component, comprising:

[0021] Projecting the total cooling load sequence into a multi-dimensional lag vector and using the multi-dimensional lag vector to form a trajectory matrix;

[0022] Performing singular value decomposition based on the trajectory matrix to obtain a plurality of singular components;

[0023] Calculating a component contribution value of each of the singular components according to a singular value of each of the singular components;

[0024] Selecting a singular component with a component contribution value greater than or equal to a preset contribution threshold to reconstruct a new trajectory matrix, and performing diagonal line averaging on the new trajectory matrix to output a reconstructed total cooling load sequence;

[0025] The singular component with a component contribution value less than a preset contribution threshold is selected as a noise component.

[0026] Optionally, the step of outputting the load rate of each of the water chiller units based on the target working condition data at the current running time and the cooling load and the predicted total cooling load by a preset load distribution control reinforcement learning model comprises:

[0027] extracting a target chilled water return temperature, a target cooling water return temperature and a target cooling water inlet temperature from the target working condition data at the current running time;

[0028] using the target chilled water return temperature and the target cooling water inlet temperature associated with the current running time and the predicted total cooling load as state space parameters;

[0029] inputting the state space parameters into a preset load distribution control reinforcement learning model for iterative operation, and calculating a reward value based on the target chilled water return temperature and the target cooling water return temperature corresponding to the current running time and the predicted total cooling load by a preset reward function, so as to maximize the reward value and output the load rate of each of the water chiller units.

[0030] Optionally, the reward function comprises:

[0031]

[0032] wherein,

[0033] ;

[0034]

[0035] wherein, is a reward value, is an average performance coefficient, is an exponential function with as a base, is the th water chiller unit, is a load rate, is a penalty value when the th water chiller unit exceeds a specified load rate interval, is a first balance coefficient, is a second balance coefficient, is a third balance coefficient, is a total cooling load, is an actual power consumption of a single water chiller unit, is a cooling load, is a performance coefficient, is an energy efficiency coefficient, is a chilled water return water temperature, is a cooling water return water temperature.

[0036] The second aspect of the present application provides a chiller load distribution device, comprising:

[0037] A working condition acquisition module is configured to acquire target working condition data of multiple operating time points associated with the multiple chillers at a current operating time point.

[0038] A sequence construction module is configured to calculate a cooling load corresponding to each target working condition data and construct a total cooling load sequence according to each operating time point based on the cooling loads.

[0039] A sequence reconstruction module is configured to decompose and reconstruct the total cooling load sequence by using a singular spectrum analysis algorithm, determine a reconstructed total cooling load sequence and a noise component.

[0040] A load prediction module is configured to perform load prediction according to the reconstructed total cooling load sequence by using a preset total cooling load prediction model, and output a predicted total cooling load in combination with the noise component.

[0041] A load distribution module is configured to output a load rate of each chiller by using a preset load distribution control reinforcement learning model based on the target working condition data of the current operating time point, the cooling load and the predicted total cooling load.

[0042] The third aspect of the present application provides a computer device, comprising a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of the chiller load distribution method according to any one of the above aspects.

[0043] The fourth aspect of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed to implement the chiller load distribution method according to any one of the above aspects.

[0044] The fifth aspect of the present application provides a computer program product, comprising computer programs / instructions, and the computer programs / instructions are executed by a processor to implement the chiller load distribution method according to any one of the above aspects.

[0045] From the above technical solutions, the present application has the following advantages:

[0046] The technical scheme of the present application provides a chiller load distribution method, comprising: obtaining target working condition data of multiple operating time points associated with multiple chillers at a current operating time point; calculating corresponding cooling load according to each target working condition data, and constructing a total cooling load sequence according to each operating time point based on each cooling load; decomposing and reconstructing the total cooling load sequence using a singular spectrum analysis algorithm to determine a reconstructed total cooling load sequence and a noise component; predicting the load according to the reconstructed total cooling load sequence through a preset total cooling load prediction model, and outputting a predicted total cooling load in combination with the noise component; and outputting the load rate of each chiller through a preset load distribution control reinforcement learning model based on the target working condition data, the cooling load and the predicted total cooling load at the current operating time point. In the entire chiller load distribution process, the fluctuation characteristics of the total cooling load are analyzed based on the singular spectrum analysis algorithm, the future load change can be better predicted through the total cooling load prediction model, and this change is fed back to the load distribution control reinforcement learning model to realize the optimization control of multiple chillers with the chiller load rate distribution as the control action, which can improve the flexibility of chiller load distribution, and is beneficial to improving the energy efficiency of the chiller system and reducing energy consumption. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art 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.

[0048] Figure 1 A step flow chart of a chiller load distribution method provided by the embodiment of the present application;

[0049] Figure 2 A structural block diagram of a chiller load distribution device provided by the embodiment of the present application. DETAILED DESCRIPTION

[0050] The embodiment of the present application provides a chiller load distribution method and device, which is used to solve the technical problem that the traditional chiller load distribution method is mainly based on meta-heuristic algorithm, and usually needs to design additional mechanisms to ensure the effectiveness of global search, thereby causing the overall complexity to increase exponentially, and resulting in low flexibility of chiller load distribution.

[0051] In order to make the purpose, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0052] See also Figure 1 , Figure 1 A flow chart of the steps of a chiller load distribution method provided by an embodiment of the present invention.

[0053] The present invention provides a method for distributing loads of a chiller, comprising:

[0054] Step 101: respectively obtain target operating condition data of multiple chillers at multiple operating moments associated with the current operating moment.

[0055] Before step 101, the following steps are included:

[0056] Collecting initial operating condition data of multiple chillers at multiple operating moments within a preset time period at the current operating moment;

[0057] The initial working condition data are processed by data elimination and data completion to obtain the corresponding target working condition data.

[0058] It should be noted that the initial operating condition data refers to data that can reflect the operating conditions of the chiller, including but not limited to the chilled water supply temperature, chilled water return temperature, chilled water flow, cooling water return temperature and cooling water inlet temperature of the chiller, while the target operating condition data refers to the data after data preprocessing such as data elimination and data completion of the initial operating condition data, and correspondingly includes the target chilled water supply temperature, target chilled water return temperature, target chilled water flow, target cooling water return temperature and target cooling water inlet temperature.

[0059] When it is necessary to allocate the load of multiple units, the initial working condition data of each unit at the current running time and multiple running times within a preset time period before the current running time are obtained to establish a working condition data set, and the data set is preprocessed to ensure the accuracy of the data. The data preprocessing includes data rejection processing and data completion processing. The data rejection processing refers to rejecting logical error data and outlier data. The logical error data refers to invalid data that does not conform to the physical law or business logic, such as the case where the chilled water supply temperature is higher than the chilled water return temperature due to sensor failure, data recording error or system configuration error, etc. The outlier data, also known as abnormal value, refers to a value that deviates significantly from other data points in the working condition data set, which may be caused by sensor noise, network delay or data transmission error, etc. For example, the water temperature increases or decreases extremely in a short time, which is much higher or lower than the historical data of the same period, and there is no reasonable explanation or background factor. The data completion processing can include linear interpolation completion and spline interpolation completion. In specific implementation, when there is missing data of a certain chilled water unit in the working condition data set, and the data of the missing data belongs to a data type that is in a relatively stable running phase, the linear interpolation method can be used to complete the missing data. This method is simple and effective and is suitable for cases where the data is not missing seriously and the change is gentle. When there is non-linear trend in a certain type of data of a certain chilled water unit in the working condition data set, the spline interpolation method can be used to complete the missing data. This method can better fit the non-linear data change.

[0060] Step 102, calculate the corresponding cooling load according to each target working condition data, and construct a total cooling load sequence according to each running time based on each cooling load.

[0061] It should be noted that the total cooling load sequence refers to a sequence formed by arranging multiple total cooling loads in chronological order.

[0062] Step 102 includes the following sub-steps:

[0063] S11, respectively extract the target chilled water supply temperature, the target chilled water return temperature and the target chilled water flow from each target working condition data.

[0064] S12, difference operation is performed on each target chilled water return temperature and the corresponding target chilled water supply temperature to obtain a plurality of chilled water temperature differences.

[0065] S13, multiply each chilled water temperature difference with the associated target chilled water flow and the specific heat of water to output the cooling load of each chilled water unit at each running time.

[0066] It should be noted that the calculation process of the cooling load specifically includes:

[0067]

[0068] In the formula, is the first air-cooled water chiller unit, is the cooling load, is the specific heat of water, is the chilled water flow rate, is the chilled water return temperature, is the chilled water supply temperature.

[0069] S14, summing up the cooling loads of each air-cooled water chiller unit according to each operating time to determine the corresponding total cooling load.

[0070] It should be noted that the total cooling load is calculated by the following process:

[0071]

[0072] S15, constructing a total cooling load sequence based on each total cooling load according to the corresponding operating time.

[0073] It should be noted that according to the target operating condition data of each unit at each operating time, the cooling load of each unit at each operating time is determined according to the above calculation process, and then the cooling load of each unit at each operating time is calculated according to the above calculation process. The total cooling load corresponding to each operating time is calculated according to the above calculation process of the total cooling load, and each total cooling load is sorted according to its corresponding operating time to construct a total cooling load sequence.

[0074] Step 103, decompose and reconstruct the total cooling load sequence using singular spectrum analysis algorithm to determine the reconstructed total cooling load sequence and noise component.

[0075] It should be noted that the singular spectrum analysis (SSA) algorithm is a time series analysis and prediction technology, which is usually regarded as a method for identifying and extracting oscillation components from original sequences, so as to decompose the original sequence into the sum of interpretable components, including trend, periodic, quasi-periodic components or noise information,

[0076] Step 103 includes the following sub-steps:

[0077] S21, projecting the total cooling load sequence into a multi-dimensional lag vector, and using the multi-dimensional lag vector to form a trajectory matrix.

[0078] S22, singular value decomposition based on the trajectory matrix to obtain a plurality of singular components.

[0079] It should be noted that singular value decomposition of the trajectory matrix can be understood as obtaining a plurality of singular components composed of the product of left singular vectors, right singular vectors and singular values, and different singular components can represent noise components, trend components and possible periodic components.

[0080] S23, calculate the component contribution value of each singular component according to the singular value of each singular component.

[0081] It should be noted that the contribution value of each singular component constituting the total cooling load sequence can be measured by the following formula:

[0082]

[0083] In the formula, is the component contribution value, is the singular component, is the number of singular components, is the singular value, is the singular value of the singular component.

[0084] S24, select the singular component with a component contribution value greater than or equal to a preset contribution threshold to reconstruct a new trajectory matrix, perform diagonal line averaging on the new trajectory matrix, and output the reconstructed total cooling load sequence.

[0085] S25, select the singular component with a component contribution value less than the preset contribution threshold as a noise component.

[0086] It should be noted that the singular components are sorted from large to small according to the component contribution value, the singular component with a component contribution value less than the preset contribution threshold is selected as a noise component, and the singular component with a component contribution value greater than or equal to the preset contribution threshold, which can represent periodic data and trend data, is selected to reconstruct a new trajectory matrix. Diagonal line averaging is performed based on the new trajectory matrix, thereby restoring a one-dimensional time sequence component to obtain a reconstructed total cooling load sequence, which can exclude the influence of noise and is beneficial to improve the prediction accuracy.

[0087] Step 104, perform load prediction according to the reconstructed total cooling load sequence through a preset total cooling load prediction model, and output a predicted total cooling load in combination with the noise component.

[0088] It should be noted that, by training and testing an existing time series prediction model, such as a Transformer model, to obtain a preset total cooling load prediction model, in specific implementation, a similar process to steps 101 to 103 is followed. First, a plurality of historical reconstructed total cooling load sequences corresponding to the current operating moments are obtained. Then, based on the maximum and minimum values ​​of each historical reconstructed total cooling load sequence, normalization is performed so that the normalized value is between [0, 1]. The normalization process is as follows:

[0089]

[0090] Where, is the original data, is the original data after standardization, is the minimum value of the sequence where the original data is located, is the maximum value of the sequence of the original data;

[0091] The standardized historical reconstructed total cooling load series are divided into a training set and a test set. The training set is used to train the time series prediction model, and the test set is used to verify the accuracy of the time series prediction model. The total cooling load prediction model is then obtained. The prediction accuracy can be evaluated using the root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE), including:

[0092]

[0093] Where, For the Running time, is the number of running times, is the true value, is the predicted value;

[0094] The reconstructed total cooling load sequence corresponding to the current operating time is input into the total cooling load prediction model for load prediction. After the total cooling load to be adjusted is output, the noise component decomposed by the singular spectrum analysis algorithm is added back to the total cooling load to be adjusted, thereby obtaining the predicted total cooling load to ensure the authenticity of the data.

[0095] Step 105: Based on the target operating condition data at the current operating moment, the cooling load, and the predicted total cooling load, the load rate of each chiller is output through a preset load distribution control reinforcement learning model.

[0096] Step 105 includes the following sub-steps:

[0097] S31, extracting a target chilled water return temperature, a target cooling water return temperature, and a target cooling water inlet temperature from the target operating condition data at the current operating time;

[0098] S32, using the target chilled water return temperature associated with the current running time, the target cooling water inlet temperature and the predicted total cooling load as state space parameters;

[0099] S33, inputting the state space parameters into the preset load distribution control reinforcement learning model for iterative operation, and calculating a reward value based on the target chilled water return temperature and the target cooling water return temperature and the predicted total cooling load corresponding to the current running time through a preset reward function, so as to output the load rate of each water chiller unit with the maximum reward value as the target.

[0100] It should be noted that in the present embodiment, the optimal water chiller load distribution problem is converted into a Markov (MDP) decision process based on deep reinforcement learning, and the MDP model of the water chiller system composed of multiple water chillers can be expressed as a four-tuple, wherein S represents the state parameters related to the water chiller system, S represents the control variables of the water chiller system, S represents the state S represents the action S represents the reward obtained, S represents the state transition probability;

[0101] In the load distribution control reinforcement learning model, the design of the state space needs to include a complete description of the problem, so that the agent can make optimal decisions according to the state changes, at the same time, the dimension of the state space should be simplified as much as possible, but the key information of the problem cannot be lost, and for the water chiller system, the cooling water inlet temperature can reflect the change of outdoor meteorological parameters, the chilled water return temperature can reflect the change of indoor load to a certain extent, and the cooling load can greatly affect the operation efficiency of the water chiller and is not controlled by the system, which is one of the most important state parameters of the cold source system, therefore, the state space parameters are specifically represented as follows:

[0102]

[0103] In the formula, S represents the predicted total cooling load, S represents the chilled water return temperature, S represents the cooling water inlet temperature;

[0104] Meanwhile, in the load distribution control reinforcement learning model, the design of the action space needs to consider the feasibility of the problem and operation, and the action space should have sufficient precision so that the agent can find the optimal solution in it. The discretization precision of the action should consider the exploration cost. To achieve higher precision, a larger action space is needed, and the agent also needs a longer time to converge. Therefore, the action space includes: in each control time step, the agent takes the load rate PLR of the chiller as the control action. The lower limit of the load rate is set to 30%. Because when the chiller runs at a low load rate, it will cause the equipment to run unstably, and even cause equipment failure. The upper limit of the load rate is 100%. The discrete step length is set according to the task requirements.

[0105] In addition, the reward function of the load distribution control reinforcement learning model includes:

[0106]

[0107] Wherein,

[0108]

[0109] In the formula, is the reward value; is the average performance coefficient, which represents the average COP of multiple chillers; is an exponential function with as the base, and the growth amount becomes larger and larger as increases; is the th chiller; is the load rate; is the penalty value when the th chiller exceeds the specified load rate range. The higher the degree of excess, the greater the punishment. A positive reward is given when it is within the specified range; is the first balance coefficient, is the second balance coefficient, is the third balance coefficient. The balance coefficient is mainly used to balance the rewards obtained by the agent between the operating efficiency of the chiller and the constraints;

[0110] In specific implementation, the agent calculates the reward function based on different state space values through iterative calculation with the environment, takes the highest reward value obtained as the optimal control point in this state and outputs the corresponding control action, and then issues the optimal control instruction to each chiller, including: creating two neural networks, one for estimating Q value and one as a target network; initializing the experience replay pool, setting the replay pool capacity, and defining the number of samples extracted from the pool each time training; obtaining the initial state of the system, taking the chilled water return water temperature , and the cooling water inlet water temperature ​and total cooling load as input states; by policy selection action ; execute the action and observe the result, the selected action applied to the environment, the system responds and moves to a new state , collect the immediate reward of the action and the new state of the system ; store the current experience into the experience replay pool; periodically randomly sample a batch of samples from the experience replay pool for training; update the weights of the Q network through the back propagation algorithm (such as gradient descent) to minimize the loss function; periodically copy the weights of the Q network to the target Q network; issue the control action as the optimal control instruction to the chiller;

[0111] Based on the above steps, in the actual iterative operation process of the load distribution control reinforcement learning model, after determining the predicted cooling load allocated to each chiller based on the predicted total cooling load, the predicted cooling load of each chiller is further used to determine the load rate of each chiller. In specific implementation, the load rate can be calculated by the ratio of the cooling load to the rated cooling load. In this process:

[0112] The performance coefficient of each chiller is calculated using the predicted cooling load, the target chilled water return temperature and the associated target cooling water return temperature at the current running moment; wherein the performance coefficient of the chiller can be calculated by establishing an energy efficiency model of the chiller, taking the cooling load, the chilled water return temperature and the cooling water return temperature as input data and the performance coefficient as output data, establishing the energy efficiency model of the chiller, determining the functional relationship between the chilled water return temperature, the cooling water return temperature, the cooling load and the performance coefficient by training the energy efficiency model, and calculating the energy efficiency coefficient of the functional relationship, so as to determine the energy efficiency model as follows:

[0113]

[0114] In the formula, is the i-th chiller; is the performance coefficient; is the energy efficiency coefficient; is the cooling load; is the chilled water return temperature; is the cooling water return temperature; Then, the predicted cooling load of each chiller is respectively multiplied by the corresponding performance coefficient to obtain the actual power consumption of each chiller, as shown in the following formula:

[0115]

[0116] ​

[0117] wherein, is the actual power consumption of the i-th cooling water unit; cooling water unit; is the actual power consumption of the single cooling water unit; is the cooling load; is the performance coefficient;

[0118] Subsequently, the average performance coefficient is determined by performing ratio operation on the sum of the predicted total cooling load and each actual power consumption, and then the reward value is calculated based on the average performance coefficient and the load rate by using the above-mentioned preset reward function, and the load rate of each cooling water unit is output as the control action with the maximum reward value as the target; wherein the calculation process of the average performance coefficient can refer to the following:

[0119]

[0120] wherein, is the total cooling load; is the actual power consumption of the i-th cooling water unit; cooling water unit; is the actual power consumption of the single cooling water unit; is the average performance coefficient.

[0121] In this embodiment, in the load distribution process of the entire cooling water unit, the fluctuation characteristics of the total cooling load are analyzed based on the singular spectrum analysis algorithm, the future load change can be better predicted through the total cooling load prediction model, and this change is fed back to the load distribution control reinforcement learning model to realize the optimization control of the multiple cooling water units with the cooling water unit load rate distribution as the control action, which can improve the load distribution flexibility of the cooling water unit, is beneficial to improve the energy efficiency of the cooling water unit system and reduce energy consumption.

[0122] Please refer to Figure 2 , Figure 2 is a structural block diagram of a cooling water unit load distribution device provided by the embodiment of the present application.

[0123] The cooling water unit load distribution device provided by the present application comprises:

[0124] A working condition acquisition module 201 is configured to acquire target working condition data of multiple running time points associated with multiple cooling water units at a current running time point respectively;

[0125] A sequence construction module 202 is configured to calculate corresponding cooling loads according to each target working condition data, and construct a total cooling load sequence according to each running time point based on each cooling load;

[0126] A sequence reconstruction module 203 is configured to decompose and reconstruct the total cooling load sequence by using a singular spectrum analysis algorithm, to determine a reconstructed total cooling load sequence and a noise component;

[0127] The load prediction module 204 is configured to perform load prediction according to the reconstructed total cooling load sequence by using a preset total cooling load prediction model, and output a predicted total cooling load in combination with a noise component.

[0128] The load distribution module 205 is configured to output a load rate of each chiller unit by using a preset load distribution control reinforcement learning model based on the target working condition data of the current operation time, the cooling load and the predicted total cooling load.

[0129] Further, the working condition acquisition module 201 is further configured to:

[0130] acquire initial working condition data of each chiller unit at multiple operation times within a preset time length at the current operation time, respectively;

[0131] perform data rejection processing and data completion processing on each initial working condition data to obtain corresponding target working condition data.

[0132] Further, the sequence construction module 202 is specifically configured to:

[0133] extract a target chilled water supply temperature, a target chilled water return temperature and a target chilled water flow rate from each target working condition data, respectively;

[0134] perform difference operation on each target chilled water return temperature and the corresponding target chilled water supply temperature to obtain a plurality of chilled water temperature differences;

[0135] perform multiplication operation on each chilled water temperature difference, the associated target chilled water flow rate and the specific heat of water to output a cooling load of each chiller unit at each operation time;

[0136] perform summation operation on the cooling loads of each chiller unit according to each operation time to determine a corresponding total cooling load;

[0137] construct a total cooling load sequence based on each total cooling load according to the corresponding operation time.

[0138] Further, the sequence reconstruction module 203 is specifically configured to:

[0139] project the total cooling load sequence into a multi-dimensional lag vector, and use the multi-dimensional lag vector to form a trajectory matrix;

[0140] perform singular value decomposition based on the trajectory matrix to obtain a plurality of singular components;

[0141] calculate a component contribution value of each singular component according to a singular value of the singular component;

[0142] Select singular components whose component contribution values ​​are greater than or equal to a preset contribution threshold to reconstruct them into a new trajectory matrix, perform diagonal averaging on the new trajectory matrix, and output the reconstructed total cooling load sequence;

[0143] The singular component whose component contribution value is less than a preset contribution threshold is selected as the noise component.

[0144] Furthermore, the load distribution module 205 is specifically configured to:

[0145] Extracting target chilled water return temperature, target cooling water return temperature and target cooling water inlet temperature from target operating condition data at the current operating time;

[0146] The target chilled water return temperature, target cooling water inlet temperature and predicted total cooling load associated with the current operating time are used as state space parameters;

[0147] The state space parameters are input into the preset load distribution control reinforcement learning model for iterative calculation. The reward value is calculated based on the target chilled water return temperature and target cooling water return temperature corresponding to the current operating time and the predicted total cooling load through the preset reward function. With the goal of maximizing the reward value, the load rate of each chiller is output.

[0148] Furthermore, the reward function includes:

[0149]

[0150] in, ;

[0151] ;

[0152] Where, is the reward value, is the average performance coefficient, For The exponential function with base , For the chillers, is the load rate, For the The penalty value when the chiller exceeds the specified load rate range, is the first balance coefficient, is the second balance coefficient, is the third balance coefficient, is the total cooling load, is the actual power consumption of a single chiller, is the cooling load, is the performance coefficient, is the energy efficiency coefficient, is the chilled water return temperature, is the return water temperature of the cooling water.

[0153] The embodiment of the present application further provides a computer device, comprising a memory and a processor, and the memory stores a computer program; the computer program is executed by the processor, so that the processor executes the steps of the water chiller load distribution method according to any one of the above-mentioned embodiments.

[0154] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program / instruction, and the computer program / instruction is executed by a processor to realize the steps of the water chiller load distribution method according to any one of the above-mentioned embodiments.

[0155] The embodiment of the present application further provides a computer program product, which comprises a computer program / instruction, and the computer program / instruction is executed by a processor to realize the steps of the water chiller load distribution method according to any one of the above-mentioned embodiments.

[0156] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-mentioned devices and modules can refer to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0157] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0158] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. According to actual needs, part or all of the units can be selected to achieve the purpose of the embodiment scheme.

[0159] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically independently, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of software functional unit.

[0160] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0161] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features; and these modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method of load distribution for a water chiller, the method comprising: The method comprises the following steps: Respectively acquiring target working condition data of multiple running time points associated with the current running time point of multiple water chilling units; According to each target working condition data, the corresponding cooling load is calculated, and based on each cooling load, the total cooling load sequence is constructed according to each running time point; The singular spectrum analysis algorithm is used to decompose and reconstruct the total cooling load sequence to determine the reconstructed total cooling load sequence and the noise component; Through the preset total cooling load prediction model, the load prediction is carried out according to the reconstructed total cooling load sequence, and the predicted total cooling load is output in combination with the noise component; Based on the target working condition data, the cooling load and the predicted total cooling load of the current running time point, the load rate of each water chilling unit is output through the preset load distribution control reinforcement learning model; The singular spectrum analysis algorithm is used to decompose and reconstruct the total cooling load sequence to determine the reconstructed total cooling load sequence and the noise component, which comprises: The total cooling load sequence is projected into a multi-dimensional lag vector, and the multi-dimensional lag vector is used to form a trajectory matrix; Based on the trajectory matrix, singular value decomposition is carried out to obtain multiple singular components; According to the singular value of each singular component, the component contribution value of each singular component is calculated; The singular components with a component contribution value greater than or equal to a preset contribution threshold are selected to reconstruct a new trajectory matrix, and the diagonal line of the new trajectory matrix is averaged to output the reconstructed total cooling load sequence; The singular components with a component contribution value less than the preset contribution threshold are selected as the noise component.

2. The water chiller load distribution method of claim 1, wherein, Before the step of respectively acquiring target working condition data of multiple running time points associated with the current running time point of multiple water chilling units, the method comprises the following steps: Respectively collecting initial working condition data of multiple running time points within a preset time length at the current running time point of multiple water chilling units; The data rejection processing and data completion processing are carried out on each initial working condition data to obtain the corresponding target working condition data.

3. The water chiller load distribution method of claim 1, wherein, The step of calculating the corresponding cooling load according to each target working condition data, and constructing the total cooling load sequence based on each cooling load according to each running time point, comprises: Respectively extracting the target chilled water supply temperature, the target chilled water return temperature and the target chilled water flow from each target working condition data; The difference between each target chilled water return temperature and the corresponding target chilled water supply temperature is calculated to obtain a plurality of chilled water temperature differences; Each chilled water temperature difference, the associated target chilled water flow and the specific heat of water are multiplied to output the cooling load of each water chilling unit at each running time point; The cooling loads of each water chilling unit are summed according to each running time point to determine the corresponding total cooling load; Based on each total cooling load, the total cooling load sequence is constructed according to the corresponding running time point.

4. The water chiller load distribution method of claim 1, wherein, The step of outputting the load rate of each water chilling unit through the preset load distribution control reinforcement learning model based on the target working condition data, the cooling load and the predicted total cooling load of the current running time point, comprises: extracting a target chilled water return water temperature, a target cooling water return water temperature and a target cooling water inlet water temperature from the target working condition data of the current running time; using the target chilled water return water temperature and the target cooling water inlet water temperature associated with the current running time and the predicted total cooling load as state space parameters; inputting the state space parameters into a preset load distribution control reinforcement learning model for iterative operation, and calculating a reward value based on the target chilled water return water temperature and the target cooling water return water temperature corresponding to the current running time and the predicted total cooling load through a preset reward function, and outputting the load rate of each chiller unit with the maximum reward value as the target.

5. The water chiller load distribution method according to claim 4, wherein, The reward function comprises: ； wherein ; ； ； ； In the formula, is a reward value, is an average performance coefficient, is an exponential function with as the base, is the first chiller unit, is a load rate, is the first chiller unit exceeds the specified load rate interval, is a first balance coefficient, is a second balance coefficient, is a third balance coefficient, is the total cooling load, is the actual power consumption of a single chiller unit, is the cooling load, is the performance coefficient, is the energy efficiency coefficient, is the chilled water return temperature, is the cooling water return temperature.

6. A chiller load distribution apparatus, comprising: comprises: a working condition acquisition module configured to acquire target working condition data of multiple running times associated with a plurality of chiller units at a current running time; a sequence construction module configured to calculate a corresponding cooling load according to each target working condition data, and construct a total cooling load sequence according to each running time based on each cooling load; a sequence reconstruction module configured to decompose and reconstruct the total cooling load sequence using a singular spectrum analysis algorithm to determine a reconstructed total cooling load sequence and a noise component; a load prediction module configured to perform load prediction according to the reconstructed total cooling load sequence through a preset total cooling load prediction model, and output a predicted total cooling load in combination with the noise component; a load distribution module configured to output the load rate of each chiller unit through a preset load distribution control reinforcement learning model based on the target working condition data of the current running time, the cooling load and the predicted total cooling load. The sequence reconstruction module is specifically configured to: project the total cooling load sequence into a multi-dimensional lag vector, and use the multi-dimensional lag vector to form a trajectory matrix; perform singular value decomposition based on the trajectory matrix to obtain a plurality of singular components; calculate a component contribution value of each singular component according to a singular value of each singular component; select singular components with a component contribution value greater than or equal to a preset contribution threshold to reconstruct a new trajectory matrix, perform diagonal line averaging on the new trajectory matrix, and output a reconstructed total cooling load sequence; select singular components with a component contribution value less than the preset contribution threshold as noise components.

7. A computer device, characterized by The computer program / instructions are executed by the processor to implement the steps of the chiller unit load distribution method according to any one of claims 1-5.

8. A computer readable storage medium having stored thereon computer programs / instructions, characterized in that, The computer program / instructions are executed by the processor to implement the steps of the chiller unit load distribution method according to any one of claims 1-5.

9. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instructions are executed by the processor to implement the steps of the chiller unit load distribution method according to any one of claims 1-5.

Images