Water chilling unit AI energy-saving group control module and method based on IOE

Abstract

The invention discloses a water chilling unit AI energy-saving group control module and method based on IOE, and the module comprises a prediction model building unit, a water chilling unit screening unit and a screening equipment energy-saving adjustment unit, relates to the technical field of water chilling unit control, and solves the technical problems that the real-time load demand cannot be met, and the energy consumption is wasted or the comfort is reduced. According to the method, the number of equipment is dynamically adjusted according to the predicted load, a traditional extensive mode of'fixed number + manual start and stop 'is avoided, for example, a high-energy-efficiency unit is started in advance before a high-load period is predicted, and energy consumption impact caused by equipment overload or temporary start and stop when the load is sharply increased is avoided; meanwhile, high-energy-efficiency equipment is screened as equipment to be analyzed by setting a COP threshold value, it is ensured that follow-up adjustment only aims at an efficient unit, the comprehensive energy efficiency ratio of the system is increased through accurate load matching and a high-energy-efficiency equipment priority strategy, and the frequent starting and stopping time and the low-load operation time of the equipment are shortened.

Description

Technical Field

[0001] The present invention relates to the technical field of chiller control, and specifically to an AI energy-saving group control module and method for chillers based on IOE. Background Technique

[0002] Chillers are not only the "heart" of the central air-conditioning system, but also the "big energy consumers" of the central air-conditioning system. Energy-saving control of chillers is very important and is a shortcut to "energy conservation and consumption reduction". Therefore, the chiller group control system has emerged as the times require.

[0003] According to the patent application with the publication number CN104566787A, an energy-saving control method and control system for a chiller are disclosed. In the present invention, by collecting the average load percentage of each running chiller, the average liquid outlet temperature, or the actual liquid supply temperature of the liquid supply main pipe of the heat exchange system, and comparing the size of the average load percentage with the current stage loading setting threshold and unloading setting threshold, and the size of the average liquid outlet temperature or the actual liquid supply temperature with the set liquid supply temperature, to determine whether to execute the loading, unloading or maintaining the current state of the chiller, which not only simplifies the data collection in the chiller loading and unloading control process, but also is beneficial to further saving the energy consumption of the chiller. It has been verified that the control method of this embodiment can save 10%-20% of the energy consumption.

[0004] However, the traditional chiller control system mainly relies on manual experience or fixed logic for equipment start-stop and parameter adjustment, which cannot meet the real-time load demand, resulting in energy consumption waste or comfort degradation. At the same time, it lacks a mechanism for dynamically screening equipment according to the COP value and cannot give priority to using high-energy efficiency equipment on the premise of meeting the load demand, resulting in a low overall energy efficiency of the system. Summary of the Invention

[0005] In view of the deficiencies of the prior art, the present invention provides an AI energy-saving group control module and method for chillers based on IOE, which solves the problems of inability to meet the real-time load demand, resulting in energy consumption waste or comfort degradation.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: An AI energy-saving group control module for chillers based on IOE, comprising:

[0007] A prediction model establishment unit, which is used to clean the multi-source sensor data to obtain preprocessed data, divide it into a training set, a validation set and a test set, build an LSTM model with PyTorch, optimize the prediction model through the validation set and the test set, use this model to predict the short-term load, calculate the optimal number of equipment operations according to the formula, and generate the required equipment information and then transmit it to the chiller screening unit;

[0008] The chiller screening unit is used to screen a corresponding number of devices as the devices to be analyzed according to the device COP value, classify the usage periods based on historical data, generate working period information and transmit it to the screening device energy-saving regulation unit;

[0009] The screening device energy-saving regulation unit is used to determine the load period according to the real-time load demand, compare the real-time load period with the total load of the devices to be analyzed, and generate a device satisfaction or device increase signal;

[0010] Analyze the device satisfaction signal, calculate the difference value to obtain the load difference value, generate a device reduction regulation signal after comparing it with the load threshold, turn off some devices according to the real-time operating power and working period information of the devices to be analyzed, and generate regulation information;

[0011] Analyze the device increase signal, calculate the load difference value, screen the pre-regulated devices based on this, increase the devices according to the real-time operating power and working period information of the devices to be analyzed, and generate regulation information and transmit it to the energy-saving control information output unit.

[0012] As a further solution of the present invention, it further includes a multi-source sensor data acquisition unit and an energy-saving control information output unit;

[0013] The multi-source sensor data acquisition unit is used to collect the operation data and load data of the chiller through different sensors, and transmit them to the prediction model establishment unit. The operation data includes temperature, pressure, flow rate, voltage and current parameters, and the load data includes the operating power and operating load of the chiller;

[0014] The energy-saving control information output unit is used to display the obtained regulation information to the corresponding operator.

[0015] As a further solution of the present invention, the specific way for the prediction model establishment unit to generate the demand device information is as follows:

[0016] Obtain the short-term load of the chiller according to the obtained prediction model and record it as Q load , and the short-term load here represents the load within 24 hours, and substitute the obtained short-term load Q load into the formula Calculate to obtain the optimal number of device operations N opt , and satisfy

[0017] where argmin is a mathematical operation, representing the value of the independent variable when the expression takes the minimum value, N is the total number of chiller devices, P i (Q) represents the power of the i-th device at a load of Q, represents the sum of the powers of N devices at their respective loads, Q load represents the load demand, and at the same time generates the demand device information.

[0018] As a further solution of the present invention, the specific way for the chiller screening unit to generate working period information is as follows:

[0019] Obtain the equipment labels i of all chillers and their corresponding coefficient of performance (COP), where i = 1, 2, …, N. Compare the COP with a preset value set by the operator, and select the equipment with a COP greater than the preset value as the equipment to be analyzed, labeled as a, where a = 1, 2, …, b, and b is the number of equipment to be analyzed;

[0020] Obtain the historical operation data of the equipment to be analyzed a within the time T set by the operator. Based on these data, obtain the operation records and equipment loads in different periods, classify the equipment loads, distinguish the high-load periods and normal-load periods, and generate working period information.

[0021] As a further solution of the present invention, the specific way for the screening equipment energy-saving regulation unit to generate an equipment satisfaction or equipment increase signal is as follows:

[0022] Obtain the real-time load demand. At the same time, classify the load in the current period according to the load demand to obtain the high-load period and the normal-load period, and determine the real-time load period;

[0023] Obtain the number of operating equipment and the total load in the real-time load period, and compare it with the total load in the current high-load period: if the former is greater than the latter, generate an equipment satisfaction signal; if the former is less than the latter, generate an equipment increase signal.

[0024] As a further solution of the present invention, the specific way for the screening equipment regulation unit to analyze the equipment satisfaction signal is as follows:

[0025] Calculate the difference between the real-time total load and the current total load, denoted as the load difference, and compare it with the load threshold set by the operator. If the load difference is greater than the threshold, generate an equipment reduction regulation signal; if it is less than the threshold, do not process it, and at the same time analyze the equipment reduction signal.

[0026] As a further solution of the present invention, the specific way for the screening equipment regulation unit to analyze the equipment reduction signal is as follows:

[0027] Obtain all the equipment a to be analyzed and their real-time operating power and working period information. Match the normal-load period in the working period information with the current period, and select the equipment with the same load condition as the screening equipment;

[0028] Obtain the COP of the screening equipment and sort it from largest to smallest. Then obtain its real-time operating power. Calculate the number of screening equipment required based on the current total load. Select the corresponding number of equipment in sequence according to the COP sorting, and turn off the remaining equipment to generate energy-saving regulation information.

[0029] As a further solution of the present invention, the specific method for the screening device adjustment unit to analyze the device increase signal is as follows:

[0030] Obtain the total load of the real-time operating devices, calculate the difference between it and the current total load and record it as the load difference. Based on the load difference, obtain all normal devices, and screen out the pre-adjustment devices according to the matching situation between the historical operation records and the current time period;

[0031] Obtain the real-time power and COP of the pre-adjustment devices, calculate the required number of devices in descending order of COP, and select the corresponding devices to generate adjustment information.

[0032] An AI energy-saving group control method for chillers based on IOE, which specifically includes the following steps:

[0033] Step 1: Collect multi-source sensor data and perform data cleaning to obtain preprocessed data, and at the same time perform data partitioning on it to obtain a training set, a validation set, and a test set;

[0034] Step 2: Use PyTorch to build an LSTM model, and optimize it with the validation set and the test set to obtain a prediction model;

[0035] Step 3: Predict the short-term load through the prediction model, calculate the optimal number of device operations according to the formula, and generate demand device information;

[0036] Step 4: Screen out the corresponding number of devices according to the COP value of the devices and record them as the devices to be analyzed. At the same time, classify the usage time periods according to the corresponding historical data to generate working time period information;

[0037] Step 5: Determine the corresponding load time period according to the real-time load demand, judge the size between the real-time load time period and the total load of the devices to be analyzed, and generate a device satisfaction or device increase signal;

[0038] Step 6: Analyze the device satisfaction signal, calculate the difference between the two and record it as the load difference, compare it with the load threshold, generate a device reduction adjustment signal, and at the same time perform corresponding device shutdown adjustment according to the real-time operating power and working time period information of the devices to be analyzed to generate adjustment information;

[0039] Step 7: Analyze the device increase signal, calculate the load difference, and screen out the pre-adjustment devices based on it in combination with the operation records of normal devices. At the same time, perform corresponding device increase adjustment according to the real-time operating power and working time period information of the devices to be analyzed to generate adjustment information.

[0040] The present invention provides an AI energy-saving group control module and method for chillers based on IOE. Compared with the prior art, it has the following beneficial effects:

[0041] The present invention avoids the extensive mode of the traditional "fixed number of units + manual start / stop" by dynamically adjusting the number of devices according to the predicted load. For example, before the high-load period is predicted, high-efficiency units are started in advance to avoid energy consumption shocks caused by equipment overload or temporary start / stop when the load surges.

[0042] At the same time, high-efficiency devices are selected as the devices to be analyzed by setting the COP threshold, ensuring that subsequent adjustments are only for high-efficiency units. Through accurate load matching and the priority strategy of high-efficiency devices, the system's comprehensive energy efficiency ratio (SCOP) is improved, and the frequent start / stop of equipment and the low-load operation time are reduced. Brief Description of the Drawings

[0043] Figure 1 It is a block diagram of the functional units of the present invention;

[0044] Figure 2 It is a flowchart of the steps of the present invention. Detailed Embodiments

[0045] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0046] Embodiment 1

[0047] Please refer to Figure 1 , this application provides an AI energy-saving group control module for chillers based on IOE. The module includes a multi-source sensor data acquisition unit, a prediction model establishment unit, a chiller screening unit, a screened equipment energy-saving adjustment unit, and an energy-saving control information output unit, and in combination with Figure 1 it can be known that the above functional units are connected in a one-way electrical connection.

[0048] The multi-source sensor data acquisition unit is used to collect the operation data and load data of the chiller through different sensors and transmit them to the prediction model establishment unit. The operation data includes temperature, pressure, flow rate, voltage, and current parameters, while the load data includes the operation power and operation load of the chiller.

[0049] The prediction model establishment unit is used to preprocess the acquired operation data and load data. Here, the preprocessing includes handling missing values and outliers to obtain preprocessed data. At the same time, the preprocessed data is divided into a training set, a validation set, and a test set. An LSTM model is constructed using a deep learning framework (such as TensorFlow or PyTorch). Meanwhile, the divided validation set and test set are substituted into the LSTM model to optimize the model. During this process, indicators such as mean squared error (MSE), root mean squared error (RMSE), and mean absolute error (MAE) are used to evaluate the performance of the model and optimize it. Finally, a prediction model is obtained;

[0050] And the example code for constructing the LSTM model is as follows:

[0051]

[0052]

[0053]

[0054] Obtain the short-term load of the chiller according to the obtained prediction model, denoted as Q load , and here the short-term load represents the load within 24 hours. And substitute the obtained short-term load Q load into the formula to calculate the optimal number of operating devices N opt , and satisfy where argmin is a mathematical operation representing the value of the independent variable when the following expression takes the minimum value, N is the total number of chiller devices, P i (Q) represents the power of the i-th device at a load of Q, represents the sum of the powers of N devices at their respective loads, Q load represents the load demand, and at the same time, generate demand device information;

[0055] Transmit the generated demand device information to the chiller screening unit.

[0056] The chiller screening unit first analyzes the acquired demand device information to obtain the labels of all chiller devices, denoted as unit i, where i = 1, 2,..., N (N is the total number of chiller devices). At the same time, obtain the coefficient of performance COP corresponding to each device i. COP is used to measure the energy utilization efficiency of the chiller. The higher the value, the less energy the unit consumes when providing the same cooling capacity;

[0057] Compare the COP value of each device with the preset value set by the operator. Select the devices with a COP greater than the preset value, record them as the devices to be analyzed, labeled as unit a, where a = 1, 2, …, b, (b represents the number of devices to be analyzed, and b ≤ N). For the selected devices to be analyzed a, obtain all their operation data within time T. The specific duration of time T is set by the operator according to actual needs. These historical data cover various operation parameters of the device during this time period, such as operation status (start, stop), cooling capacity output, input power, operation duration, etc.;

[0058] According to the obtained historical data, sort out the operation records of the device to be analyzed a at different time periods. The operation record contains the specific operation parameter values and status of the device at each time point or time period. Extract the device load data for different time periods from the operation record. The device load reflects the amount of cooling task undertaken by the chiller during operation. Based on the size of the device load, classify these time periods:

[0059] Classify the time periods with a device load higher than a specific threshold (this threshold can be set by the operator according to the rated load of the device and the actual operation situation, such as 80% of the rated load of the device) as high-load time periods. During high-load time periods, the chiller needs to operate at full capacity to meet the cooling demand, and the energy consumption is relatively high;

[0060] The time periods when the device load is within the normal range (lower than the high-load threshold and higher than the low-load threshold. The low-load threshold can be set to 30% of the rated load of the device, and the specific value can be adjusted according to the actual situation) are classified as normal-load time periods. During normal-load time periods, the device operates relatively smoothly, and the energy consumption is at a relatively reasonable level;

[0061] Based on the above analysis, obtain the corresponding working time period information and transmit it to the energy-saving adjustment unit of the screening device.

[0062] The energy-saving adjustment unit of the screening device, which is used to analyze the obtained working time period information, obtain the real-time load demand, and at the same time classify the load of the current time period according to the load demand to obtain high-load time periods and normal-load time periods, and conduct device adjustment analysis on the two respectively. The specific adjustment analysis methods are as follows:

[0063] For high-load time periods, obtain the number of chiller devices in operation in real time, as well as the total real-time operation load corresponding to these devices. At the same time, clarify the total load demand of the current high-load time period, and compare the total real-time operation load with the total load of the current high-load time period:

[0064] If the total real-time operation load is greater than the total load of the current high-load time period, it means that the existing operating devices can meet or even exceed the load demand. At this time, generate a device satisfaction signal;

[0065] If the total real-time operating load is less than the total load during the current high-load period, it indicates that the existing equipment cannot meet the load demand, and equipment needs to be added, thereby generating an equipment addition signal. Subsequently, different generated signals are analyzed and processed respectively;

[0066] When the equipment satisfaction signal is received, calculate the difference between the total real-time operating load and the total load during the current high-load period, record it as the load difference, and compare the load difference with the load threshold preset by the operator:

[0067] If the load difference is greater than the load threshold, it means that the current number of real-time operating equipment is too large and there is a situation of load waste. At this time, the number of equipment needs to be reduced and adjusted, and an equipment reduction adjustment signal is generated;

[0068] If the load difference is less than the load threshold, it indicates that the current equipment operating state is basically reasonable and no adjustment is required, so no additional processing is done;

[0069] When the equipment reduction adjustment signal is received, obtain all equipment a in the to-be-analyzed state (that is, the equipment with a coefficient of performance (COP) greater than the preset value screened out in the early stage). At the same time, obtain the real-time operating power of each piece of equipment a to be analyzed and its corresponding working period information, and match the normal load period in the working period information with the current high-load period. Here, the matching refers to the consistency of the load situation within a specific period, such as both being in a high-load state or both being in a normal load state. Screen out the equipment to be analyzed whose working period information matches the current period, and record it as the screened equipment;

[0070] Obtain the COP values corresponding to the screened equipment and sort them in descending order according to the COP values. The higher the COP value, the higher the energy utilization efficiency of the equipment. Obtain the real-time operating power corresponding to the screened equipment. Taking the total load during the current high-load period as the standard, calculate and determine the number of screened equipment required to meet the load demand. In the order of COP from large to small, select the corresponding number of screened equipment to continue running in sequence, and for the remaining screened equipment, perform the shutdown operation. At the same time, generate energy-saving adjustment information.

[0071] It is known that there are a total of 6 pieces of equipment a to be analyzed screened out in the early stage (that is, the COPs of these 6 chillers are greater than the preset value). Obtain the real-time operating powers of these 6 pieces of equipment, which are Equipment A: 800 kW, Equipment B: 750 kW, Equipment C: 850 kW, Equipment D: 700 kW, Equipment E: 900 kW, and Equipment F: 650 kW respectively. At the same time, obtain their working period information and find that the load situations of Equipment B, D, and F during the past normal load periods are similar to the current high-load period (such as similar load fluctuation ranges), and record these 3 pieces of equipment as the screened equipment;

[0072] Obtain the COP values of the screening devices. Assuming they are sorted from largest to smallest as: Device D (COP = 4.2), Device B (COP = 4.0), Device F (COP = 3.8). Taking the current total load of 3500 kW as the standard, after calculation (considering factors such as device operation efficiency), it is determined that only 4 devices need to operate to meet the load demand. Sorting by COP from largest to smallest, select Device D, B, and another 2 devices with higher COP (assumed to be Device A and C) to continue operating, and turn off Device F and the remaining unselected devices (such as Device E). Generate energy-saving adjustment information, record that Device F and E are turned off this time, and further analyze the energy-saving effect and system operation stability after this adjustment.

[0073] Analyze the generated device addition signals, obtain the total load corresponding to all real-time operating devices, and then compare it with the total load during the current high-load period, and calculate the difference between the two. This difference is denoted as the load difference. The calculation formula is: Load difference = Total load during the current high-load period - Total load of real-time operating devices. This difference represents the additional cooling capacity required to meet the load demand.

[0074] Obtain normal devices. Normal devices refer to those in a standby state, without faults in the device itself, and meeting the requirements in the previous energy efficiency screening (for example, COP is greater than the preset value). The system will obtain information on all such normal devices. According to historical data, the system will obtain the operation records corresponding to each normal device. These operation records include operation parameters of the device at different times, such as cooling capacity, power, COP, etc. Then, the system will match these operation records with the current period information. The matching basis includes period type (such as whether it is a high-load period), date, time range, etc., to ensure that the selected devices have good operation performance under similar working conditions. Select the normal devices whose operation records match the current period information, and record these devices as pre-adjustment devices.

[0075] The system will obtain the corresponding real-time operating power and COP value of each pre-adjustment device. The COP value reflects the energy efficiency level of the device. The higher the COP value, the less energy the device consumes when providing the same cooling capacity. Taking the load difference as the standard, combined with the cooling capacity of each pre-adjustment device (which can be estimated according to historical data), calculate the number of pre-adjustment devices to be added in the order of COP from largest to smallest.

[0076] During specific calculation, first estimate the cooling capacity of each pre-adjustment device under the current working condition according to historical data, then divide the load difference by the cooling capacity of a single device, round up to get the number of devices to be added, select the corresponding number of pre-adjustment devices in the order of COP from largest to smallest, generate adjustment information including device numbers, start instructions, etc., and transmit it to the energy-saving control information output unit, which executes the device start operation.

[0077] For the normal period, in the same way as the analysis method for the high-load period, all devices are adjusted, adjustment information is generated, and at the same time, it is transmitted to the energy-saving control information output unit.

[0078] The energy-saving control information output unit is used to display the obtained adjustment information to the corresponding operator.

[0079] Embodiment 2

[0080] Please refer to Figure 2 , as Embodiment 2 of the present invention, a method for AI energy-saving group control of a chiller based on IOE is provided. The method specifically includes the following steps:

[0081] Step 1: Collect multi-source sensor data and perform data cleaning to obtain preprocessed data, and at the same time perform data division on it to obtain a training set, a validation set, and a test set;

[0082] Step 2: Use PyTorch to build an LSTM model, and optimize it with the validation set and the test set to obtain a prediction model, and the processing method here is the same as that of the prediction model establishment unit in Embodiment 1;

[0083] Step 3: Predict the short-term load through the prediction model, calculate the optimal number of device operations according to the formula, and generate demand device information, and the processing method here is the same as that of the prediction model establishment unit in Embodiment 1;

[0084] Step 4: Select the corresponding number of devices according to the device COP value and record them as devices to be analyzed. At the same time, classify the usage time period according to the corresponding historical data to generate working time period information, and the processing method here is the same as that of the chiller selection unit in Embodiment 1;

[0085] Step 5: Determine the corresponding load period according to the real-time load demand, judge the size between the real-time load period and the total load of the devices to be analyzed, and generate a device satisfaction or device increase signal, and the processing method here is the same as that of the device selection and energy-saving adjustment unit in Embodiment 1;

[0086] Step 6: Analyze the device satisfaction signal, calculate the difference between the two and record it as the load difference, compare it with the load threshold, generate a device reduction adjustment signal, and at the same time perform corresponding device shutdown adjustment according to the real-time operating power and working time period information of the devices to be analyzed, and generate adjustment information, and the processing method here is the same as that of the device selection and energy-saving adjustment unit in Embodiment 1;

[0087] Step 7: Add signal analysis to the device, calculate the load difference, and use it as a criterion to screen the pre-adjustment device in combination with the operation records of normal devices. At the same time, perform corresponding device addition adjustment according to the real-time operating power and working period information of the device to be analyzed, generate adjustment information, and the processing method here is the same as that of screening the device energy-saving adjustment unit in Embodiment 1.

[0088] For some data in the above formula, only their numerical values are taken for calculation, and the parameter units are not substituted for calculation. At the same time, the content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

[0089] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An AI energy-saving group control module for a water chiller based on IOE, characterized in that Including: A prediction model establishment unit, which is used to clean multi-source sensor data to obtain preprocessed data, divide it into a training set, a validation set and a test set, build an LSTM model using PyTorch, optimize the prediction model through the validation set and the test set, predict the short-term load using this model, calculate the optimal number of equipment operations according to the formula, generate demand equipment information and transmit it to the chiller screening unit; A chiller screening unit, which is used to screen the corresponding number of equipment as the equipment to be analyzed according to the equipment COP value, classify the usage time periods based on historical data, generate working time period information and transmit it to the screening equipment energy-saving regulation unit; A screening equipment energy-saving regulation unit, which is used to determine the load time period according to the real-time load demand, compare the real-time load time period with the total load of the equipment to be analyzed, and generate an equipment satisfaction or equipment increase signal; Analyze the equipment satisfaction signal, calculate the difference value to obtain the load difference value, compare it with the load threshold value and then generate an equipment reduction regulation signal, turn off some equipment according to the real-time operating power and working time period information of the equipment to be analyzed, and generate regulation information; Analyze the equipment increase signal, calculate the load difference value, use this to screen the pre-regulated equipment, increase the equipment according to the real-time operating power and working time period information of the equipment to be analyzed, and generate regulation information and transmit it to the energy-saving control information output unit.

2. The AI energy-saving group control module for a chiller based on IOE according to claim 1, characterized in that, It also includes a multi-source sensor data acquisition unit and an energy-saving control information output unit; The multi-source sensor data acquisition unit is used to collect the operation data and load data of the chiller through different sensors, and transmit them to the prediction model establishment unit, and the operation data includes temperature, pressure, flow, voltage and current parameters, while the load data includes the operating power and operating load of the chiller; The energy-saving control information output unit is used to display the obtained regulation information to the corresponding operator.

3. The AI energy-saving group control module for a water chiller based on IOE according to claim 1, wherein The specific way for the prediction model establishment unit to generate demand equipment information is: Obtain the short-term load of the chiller according to the obtained prediction model, denoted as Q load , and the short-term load here represents the load within 24 hours, and substitute the obtained short-term load Q load into the formula to calculate the optimal number of equipment operations N opt , and satisfy Among them, argmin is a mathematical operation, representing the value of the independent variable when the expression takes the minimum value. N is the total number of chiller units, and P i (Q) represents the power of the i-th unit at a load of Q, represents the total power of N units at their respective loads, and Q load represents the load demand and generates demand equipment information at the same time.

4. An AI energy-saving group control module for a water chiller based on IOE according to claim 1, characterized in that, The specific way for the chiller screening unit to generate working time period information is: Obtain all chiller equipment numbers i and their corresponding energy efficiency ratios COP, where i = 1, 2,..., N, compare the COP with the preset value set by the operator, and screen out the equipment with COP greater than the preset value as the equipment to be analyzed, with the number a, where a = 1, 2,..., b, and b is the number of equipment to be analyzed; Obtain the historical operation data of the equipment to be analyzed a within the time T set by the operator, obtain the operation records and equipment loads in different time periods based on these data, classify the equipment loads, distinguish the high-load time periods and normal-load time periods, and generate working time period information.

5. An AI energy-saving group control module for a water chiller based on IOE according to claim 1, characterized in that, The specific way for the screening equipment energy-saving regulation unit to generate an equipment satisfaction or equipment increase signal is: Obtain the real-time load demand, and at the same time classify the load in the current time period according to the load demand to obtain the high-load time periods and normal-load time periods, and determine the real-time load time period; Obtain the number of operating equipment and the total load in the real-time load time period, and compare it with the total load in the current high-load time period: if the former is greater than the latter, generate an equipment satisfaction signal; If the former is less than the latter, generate an equipment increase signal.

6. The AI energy-saving group control module for a water chiller based on IOE according to claim 1, characterized in that The specific way for the screening equipment regulation unit to analyze the equipment satisfaction signal is: Calculate the difference between the real-time running total load and the current total load, which is denoted as the load difference. Compare it with the load threshold set by the operator. If the load difference is greater than the threshold, generate a device reduction adjustment signal; if it is less than the threshold, do not process it. At the same time, analyze the device reduction signal.

7. An AI energy-saving group control module for a water chiller based on IOE according to claim 6, characterized in that, The specific method for the screening device adjustment unit to analyze the device reduction signal is as follows: Obtain all devices a to be analyzed and their real-time running power and working period information. Match the normal load period in the working period information with the current period, and screen out the devices with the same load situation as the screening devices; Obtain the COP of the screening devices and sort them from large to small. Then obtain their real-time running power. Calculate the required number of screening devices based on the current total load. Select the corresponding number of devices in the order of COP sorting, turn off the remaining devices, and generate energy-saving adjustment information.

8. The AI energy-saving group control module for a water chiller based on IOE according to claim 1, characterized in that The specific method for the screening device adjustment unit to analyze the device increase signal is as follows: Obtain the total load of the real-time running devices, calculate the difference between it and the current total load, which is denoted as the load difference. Based on the load difference, obtain all normal devices, and screen out the pre-adjustment devices according to the matching situation between the historical running records and the current period; Obtain the real-time power and COP of the pre-adjustment devices, calculate the required number of devices from large to small according to the COP, and select the corresponding devices to generate adjustment information.

9. An AI energy-saving group control method for a chiller based on IOE, which is used to execute an AI energy-saving group control module for a chiller based on IOE according to any one of claims 1-8, characterized in that, This method specifically includes the following steps: Step 1: Collect multi-source sensor data and perform data cleaning to obtain preprocessed data. At the same time, perform data partitioning on it to obtain a training set, a validation set, and a test set; Step 2: Use PyTorch to build an LSTM model, and optimize it using the validation set and the test set to obtain a prediction model; Step 3: Predict the short-term load through the prediction model, calculate the optimal number of device operations according to the formula, and generate demand device information; Step 4: Screen out the corresponding number of devices according to the device COP value, which are denoted as devices to be analyzed. At the same time, classify their usage periods according to the corresponding historical data to generate working period information; Step 5: Determine the corresponding load period according to the real-time load demand, judge the size between the real-time load period and the total load of the devices to be analyzed, and generate a device satisfaction or device increase signal; Step 6: Analyze the device satisfaction signal, calculate the difference between the two, which is denoted as the load difference, and compare it with the load threshold to generate a device reduction adjustment signal. At the same time, perform corresponding device shutdown adjustment according to the real-time running power and working period information of the devices to be analyzed, and generate adjustment information; Step 7: Analyze the device increase signal, calculate the load difference, and screen out the pre-adjustment devices based on it in combination with the running records of normal devices. At the same time, perform corresponding device increase adjustment according to the real-time running power and working period information of the devices to be analyzed, and generate adjustment information.

Images