Control method and system for energy efficiency optimization of air conditioning system of constant-temperature and constant-humidity factory workshop

Abstract

The invention discloses a control method and system for energy efficiency optimization of a workshop air conditioning system of a constant-temperature and constant-humidity factory. The control method comprises the steps that a load prediction model is trained, and the constant-temperature and constant-humidity factory is partitioned to set importance weights for all air conditioners; based on a reinforcement learning algorithm, multi-objective optimization is carried out on an air conditioning system composed of all air conditioners, the reinforcement learning algorithm takes adjustment of operation parameters of all the air conditioners as actions, and the actions of each iteration and environment parameters, detected in real time, of a constant-temperature and constant-humidity factory workshop are input into a load prediction model; the predicted cold load, the predicted heat load and the predicted dehumidification amount of each air conditioner are obtained, and the performance coefficient of each air conditioner is calculated; and weighting the performance coefficient of each air conditioner according to the importance weight, and constructing a reward function for multi-objective optimization by combining a weighting result and the environment parameters, detected in real time, of the constant-temperature and constant-humidity factory workshop. According to the invention, the optimal optimization of the constant-temperature and constant-humidity factory among the lowest energy consumption, the most stable control and the highest precision is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent control of air-conditioning systems, and more specifically, relates to a control method and system for optimizing the energy efficiency of an air-conditioning system in a constant temperature and humidity factory workshop. Background Art

[0002] With the rapid development of high-end manufacturing industries such as electronics, biomedicine, and precision instruments, constant temperature and humidity factory workshops have become critical infrastructure for ensuring product quality and stable production processes. These workshops require extremely precise temperature and humidity control. For example, semiconductor chip manufacturing workshops require a temperature fluctuation range of ±0.5°C and a relative humidity control range of 40%-60%. Even the slightest deviation in these environmental parameters can lead to a decrease in product yield or even failure of the production process. However, the energy consumption of traditional air conditioning systems is becoming increasingly prominent. According to statistics, air conditioning energy consumption in constant temperature and humidity workshops accounts for 30%-60% of the total energy consumption of the entire factory.

[0003] At present, the control technology of the constant temperature and humidity workshop air-conditioning system still has many limitations. Traditional control methods mostly use fixed parameter control strategies based on experience, such as timed start and stop equipment, fixed flow adjustment, etc., which cannot adapt to the dynamically changing load requirements of the workshop. Factors such as the number of people in the workshop, the operating status of the equipment, and outdoor meteorological conditions will cause frequent fluctuations in heat and humidity loads. Traditional control methods are prone to energy waste such as "overcooling and overheating" and "large flow and small load". At the same time, existing systems often rely on a single sensor or a few parameters for control, lacking comprehensive analysis of multi-dimensional data such as the environment, equipment, and energy, making it difficult to achieve refined management. In addition, the lack of fault diagnosis and early warning mechanisms makes it impossible for the system to respond in a timely manner when equipment anomalies (such as sensor failure and valve blockage) occur, further exacerbating energy efficiency reduction and production risks. Summary of the Invention

[0004] In order to solve the above problems, the purpose of the present invention is to provide a control method and system for optimizing the energy efficiency of a constant temperature and humidity factory workshop air-conditioning system, thereby realizing precise control and maximizing energy efficiency of the constant temperature and humidity factory workshop air-conditioning system.

[0005] The present invention adopts the following technical solutions.

[0006] A first aspect of the present invention provides a control method for optimizing the energy efficiency of a constant temperature and humidity factory workshop air conditioning system, comprising the following:

[0007] Obtain the environmental parameters of different constant temperature and humidity factory workshops and the operating parameters of different constant temperature and humidity air conditioners, as well as the corresponding constant temperature and humidity air conditioner cooling, heating load and dehumidification capacity, and use them as samples to train the load prediction model;

[0008] The constant temperature and humidity factory is divided into zones based on the density of personnel and the heat output of each device. The location of each constant temperature and humidity air conditioner is obtained, and the importance weight of each constant temperature and humidity air conditioner is set according to the location and zone.

[0009] Based on the reinforcement learning algorithm, multi-objective optimization is performed on the air-conditioning system composed of all constant temperature and humidity air conditioners. The adjustment of the operating parameters of each constant temperature and humidity air conditioner is taken as an action. The action of each iteration and the environmental parameters of the constant temperature and humidity factory workshop detected in real time are input into the load prediction model to obtain the predicted cooling, heating load and dehumidification capacity of each constant temperature and humidity air conditioner. The performance coefficient of each constant temperature and humidity air conditioner is calculated based on the cooling, heating load and dehumidification capacity and the real-time detected external environmental parameters; the performance coefficient of each constant temperature and humidity air conditioner is weighted according to the importance weight, and the reward function is constructed by combining the weighted result and the real-time detected environmental parameters of the constant temperature and humidity factory workshop to perform multi-objective optimization.

[0010] Preferably, a temperature and humidity sensor is placed at each set point of the constant temperature and humidity factory workshop, the environmental parameters of the constant temperature and humidity factory workshop include the temperature value and humidity value of each set point of the constant temperature and humidity factory workshop within a set period, and the operating parameters of the constant temperature and humidity air conditioner include the start and stop status of the compressor, the water pump frequency, and the air valve opening;

[0011] The time series features and spatial features of the environmental parameters of the constant temperature and humidity factory workshop are extracted, and the time series features are screened. The screened time series features are spliced ​​with the spatial features, the environmental parameters of the constant temperature and humidity factory workshop, and the operating parameters of the constant temperature and humidity air conditioner as the feature set of the sample. The corresponding cooling, heating loads and dehumidification amounts are used as labels to train a load prediction model. The load prediction model adopts the Transformer model.

[0012] Preferably, the extraction of temporal and spatial features of the environmental parameters of the constant temperature and humidity factory workshop is performed as follows:

[0013] The time series characteristics include: the temperature change rate, humidity change rate, temperature standard deviation, humidity standard deviation, temperature pulsation characteristics, and humidity pulsation characteristics of each set point within a set period; the temperature pulsation characteristics are obtained by performing discrete wavelet transform on the temperature of each set point within a set period to obtain energy density at multiple scales, and the energy density at each scale is used as the temperature pulsation characteristic of the corresponding scale, and each temperature pulsation characteristic is a time series feature; the humidity pulsation characteristics are obtained by performing discrete wavelet transform on the humidity of each set point within a set period to obtain energy density at multiple scales, and the energy density at each scale is used as the humidity pulsation characteristic of the corresponding scale, and each humidity pulsation characteristic is a time series feature;

[0014] The spatial characteristics include temperature uniformity index, temperature stratification index and humidity gradient field;

[0015] The temperature uniformity index is the variance of the temperature at each point, and the temperature stratification index is the difference between the average temperature of the top floor and the average temperature of the bottom floor of the constant temperature and humidity factory workshop divided by the height of the constant temperature and humidity factory workshop; the humidity gradient field is a three-dimensional coordinate, and the three elements represent the rate of change of humidity at each point in the constant temperature and humidity factory workshop on the x-axis, y-axis, and z-axis, respectively.

[0016] Preferably, all the time series features of all the set points are normalized to form a feature matrix, and the element in the u-th row and j-th column of the feature matrix is ​​the v-th time series feature of the u-th set point; the feature matrix is ​​subjected to principal component analysis for dimensionality reduction, and the feature matrix after dimensionality reduction is the screened time series feature.

[0017] Preferably, the constant temperature and humidity factory is partitioned based on the personnel density and the heating power of each device, and the location of each constant temperature and humidity air conditioner is obtained. An importance weight is set for each constant temperature and humidity air conditioner according to the location and partition, specifically:

[0018] If a constant temperature and humidity factory contains temperature-sensitive equipment or humidity-sensitive equipment, and the temperature-sensitive equipment and / or humidity-sensitive equipment respectively have temperature and humidity accuracy requirements higher than a set accuracy threshold, the area where the equipment is installed is set as the core area; outside the core area, areas where the population density exceeds the set density threshold and / or the average heating power of the equipment exceeds the set heating power threshold are set as dense areas; other areas are ordinary areas;

[0019] Set weights for the three areas. The weight of the core area is set to be greater than that of the dense area, and the weight of the dense area is set to be greater than that of the ordinary area.

[0020] Clustering is performed according to the location of each constant temperature and humidity air conditioner to obtain the area where each cluster center is located. The importance weight of all constant temperature and humidity air conditioners in the cluster corresponding to the cluster center is equal to the weight of the area where the cluster center is located.

[0021] Preferably, the performance coefficient of each constant temperature and humidity air conditioner is calculated based on the cooling and heating loads and dehumidification capacity and the external environmental parameters detected in real time, specifically:

[0022] External environmental parameters include external temperature and humidity;

[0023] The performance coefficient COP calculation formula is:

[0024]

[0025] Where Q c , Q h , D are cooling load, heating load and dehumidification capacity respectively, Δh is the air specific enthalpy difference, which is the difference in specific enthalpy between the air entering the air conditioner and leaving the air conditioner, Te 、φ e are the external temperature and humidity respectively, α and β are the set coefficients, T th This setting affects the operating temperature of air-conditioning related hardware.

[0026] Preferably, the reward function is constructed by combining the weighted results with the environmental parameters of the constant temperature and humidity factory workshop detected in real time to perform multi-objective optimization, specifically:

[0027] The reward function R is calculated as follows:

[0028]

[0029] Where, COP b is the set benchmark energy efficiency, COP′ is the weighted result, k c is a coefficient with a range of (0,1], ω T 、ω H is the set temperature weight and humidity weight, U is the total number of set points, T u 、H u are the temperature and humidity detected at the u-th set point respectively; T u 、H set are the target temperature and humidity values ​​respectively, f is the constraint penalty term, λ is the set penalty term weight, and exp is the exponential operation with the natural exponent e as the base.

[0030] Preferably, the constraint penalty item is specifically:

[0031] Real-time detection of the current actual water pump frequency, air valve opening, and temperature and humidity of each set point;

[0032] Obtain all set points whose temperatures are greater than the maximum allowable temperature or less than the minimum allowable temperature, calculate the difference between their temperatures and the corresponding maximum allowable temperature or minimum allowable temperature, and calculate the average of these differences to obtain the temperature penalty term; obtain all set points whose humidity is greater than the maximum allowable humidity or less than the minimum allowable humidity, calculate the difference between their humidity and the corresponding maximum allowable temperature or minimum allowable temperature, and calculate the average of these differences to obtain the humidity penalty term;

[0033] Calculate the difference between the adjusted water pump frequency and air valve opening of the current iterative setting of all set points and the actual current water pump frequency and air valve opening to obtain the water pump frequency difference and air valve opening difference. If the water pump frequency difference is greater than the water pump frequency error threshold, calculate the difference between all corresponding water pump frequency differences and the water pump frequency error threshold, and calculate the average value as the water pump penalty item; if the air valve opening difference is greater than the air valve opening error threshold, calculate the difference between all corresponding air valve opening differences and the air valve opening error threshold, and calculate the average value as the air valve penalty item;

[0034] The constraint penalty items are obtained by normalizing each penalty item and performing weighted summation. The weights of the temperature penalty item and the water pump penalty item are both the set temperature weight; the weights of the humidity penalty item and the air valve penalty item are both the set humidity weight.

[0035] Preferably, when the reinforcement learning algorithm iteratively adjusts the operating parameters of each constant temperature and humidity air conditioner each time, the current performance coefficients of all constant temperature and humidity air conditioners are normalized, and the normalization result is the probability of adjusting the operating parameters of the corresponding constant temperature and humidity air conditioner.

[0036] The second aspect of the present invention provides a control system for optimizing the energy efficiency of a constant temperature and humidity factory workshop air conditioning system based on the method described in the first aspect of the present invention, comprising a load forecasting model construction module, a weight setting module, and a multi-objective optimization module, specifically:

[0037] Load forecasting model construction module: This module obtains the cooling and heating loads and dehumidification capacity of constant temperature and humidity air conditioners under different constant temperature and humidity factory workshop environmental parameters and different constant temperature and humidity air conditioner operating parameters, and uses these as samples to train the load forecasting model;

[0038] Weight setting module: This module divides the constant temperature and humidity factory into zones based on the density of personnel and the heat output of each device, obtains the location of each constant temperature and humidity air conditioner, and sets the importance weight for each constant temperature and humidity air conditioner based on the location and zone.

[0039] Multi-objective optimization module: Based on the reinforcement learning algorithm, multi-objective optimization is performed on the air-conditioning system composed of all constant temperature and humidity air conditioners. The reinforcement learning algorithm adjusts the operating parameters of each constant temperature and humidity air conditioner as an action, and inputs the action of each iteration and the environmental parameters of the constant temperature and humidity factory workshop detected in real time into the load prediction model to obtain the predicted cooling, heating load and dehumidification capacity of each constant temperature and humidity air conditioner. The performance coefficient of each constant temperature and humidity air conditioner is calculated based on the cooling, heating load and dehumidification capacity and the real-time detected external environmental parameters; the performance coefficient of each constant temperature and humidity air conditioner is weighted according to the importance weight, and the reward function is constructed by combining the weighted result and the real-time detected environmental parameters of the constant temperature and humidity factory workshop to perform multi-objective optimization.

[0040] The beneficial effect of the present invention is that, compared with the prior art, the constant temperature and humidity factory is partitioned based on the personnel density and the heating power of each equipment, and the location of each constant temperature and humidity air conditioner is obtained, and the importance weight is set for each constant temperature and humidity air conditioner according to the location and partition. It is suitable for various factory workshops with different requirements and pays attention to sensitive areas of the factory; the present invention inputs the action and environmental parameters of each iteration into the load prediction model to obtain the predicted cooling and heating loads and dehumidification amount of each constant temperature and humidity air conditioner, calculates the performance coefficient of each constant temperature and humidity air conditioner, performs energy efficiency evaluation according to the predicted load dynamic changes and eliminates the evaluation deviation caused by climate factors, so that the energy efficiency calculation is accurate; multi-objective optimization is performed to achieve the optimal optimization of the constant temperature and humidity factory among the lowest energy consumption, the most stable control and the highest accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 Flowchart of the present invention. DETAILED DESCRIPTION

[0042] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. The embodiments described in this application are only part of the embodiments of the present invention, not all of them. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0043] The present invention adopts the following technical solutions.

[0044] like Figure 1 As shown, embodiment 1 of the present invention proposes a control method for optimizing the energy efficiency of a constant temperature and humidity factory workshop air conditioning system, including the following contents:

[0045] Obtain the environmental parameters of different constant temperature and humidity factory workshops and the operating parameters of different constant temperature and humidity air conditioners, as well as the corresponding constant temperature and humidity air conditioner cooling, heating load and dehumidification capacity, and use them as samples to train the load prediction model;

[0046] The constant temperature and humidity factory is divided into zones based on the density of personnel and the heat output of each device. The location of each constant temperature and humidity air conditioner is obtained, and the importance weight of each constant temperature and humidity air conditioner is set according to the location and zone.

[0047] Based on the reinforcement learning algorithm, multi-objective optimization is performed on the air-conditioning system composed of all constant temperature and humidity air conditioners. The adjustment of the operating parameters of each constant temperature and humidity air conditioner is taken as an action. The action of each iteration and the environmental parameters of the constant temperature and humidity factory workshop detected in real time are input into the load prediction model to obtain the predicted cooling, heating load and dehumidification capacity of each constant temperature and humidity air conditioner. The performance coefficient of each constant temperature and humidity air conditioner is calculated based on the cooling, heating load and dehumidification capacity and the real-time detected external environmental parameters; the performance coefficient of each constant temperature and humidity air conditioner is weighted according to the importance weight, and the reward function is constructed by combining the weighted result and the real-time detected environmental parameters of the constant temperature and humidity factory workshop to perform multi-objective optimization.

[0048] In this embodiment, preferably, the cooling and heating loads and dehumidification amounts of the constant temperature and humidity air conditioners under different constant temperature and humidity factory workshop environmental parameters and different constant temperature and humidity air conditioner operating parameters are obtained and used as samples to train the load prediction model, specifically:

[0049] Temperature and humidity sensors are placed at various set points in the constant temperature and humidity factory workshop. The environmental parameters of the constant temperature and humidity factory workshop include the temperature and humidity values ​​of each set point in the constant temperature and humidity factory workshop within a set period. The operating parameters of the constant temperature and humidity air conditioner include the start and stop status of the compressor, the water pump frequency, and the air valve opening.

[0050] It should be noted that the set points include the air outlets of all air conditioners and several locations on the top and bottom floors of the constant temperature and humidity factory workshop;

[0051] The time series features and spatial features of the environmental parameters of the constant temperature and humidity factory workshop are extracted, and the time series features are screened. The screened time series features are spliced ​​with the spatial features, the environmental parameters of the constant temperature and humidity factory workshop, and the operating parameters of the constant temperature and humidity air conditioner as the feature set of the sample. The corresponding cooling, heating loads and dehumidification amounts are used as labels to train a load prediction model. The load prediction model adopts the Transformer model.

[0052] In this embodiment, preferably, the extraction of temporal and spatial features of the environmental parameters of the constant temperature and humidity factory workshop is performed as follows:

[0053] The time series characteristics include: the temperature change rate, humidity change rate, temperature standard deviation, humidity standard deviation, temperature pulsation characteristics, and humidity pulsation characteristics of each set point within a set period; the temperature pulsation characteristics are obtained by performing discrete wavelet transform on the temperature of each set point within a set period to obtain energy density at multiple scales, and the energy density at each scale is used as the temperature pulsation characteristic of the corresponding scale, and each temperature pulsation characteristic is a time series feature; the humidity pulsation characteristics are obtained by performing discrete wavelet transform on the humidity of each set point within a set period to obtain energy density at multiple scales, and the energy density at each scale is used as the humidity pulsation characteristic of the corresponding scale, and each humidity pulsation characteristic is a time series feature;

[0054] The spatial characteristics include temperature uniformity index, temperature stratification index and humidity gradient field;

[0055] The temperature uniformity index is the variance of the temperature at each point, and the temperature stratification index is the difference between the average temperature of the top floor and the average temperature of the bottom floor of the constant temperature and humidity factory workshop divided by the height of the constant temperature and humidity factory workshop; the humidity gradient field is a three-dimensional coordinate, and the three elements represent the rate of change of humidity at each point in the constant temperature and humidity factory workshop on the x-axis, y-axis, and z-axis, respectively.

[0056] It should be noted that the temperature and humidity of each set point in the set period are detected and the corresponding slopes obtained by linear fitting are the temperature change rates and humidity change rates of each set point in the set period.

[0057] The calculation formulas for the temperature pulsation characteristics and humidity pulsation characteristics are:

[0058]

[0059] in, When it is 0, the corresponding parameter is temperature, and when it is 1, the corresponding parameter is humidity; for The corresponding parameter is the wavelet coefficient when the scale is j and the time period is k, j∈[1,J max ], J max is the maximum scale, and its value is Round down, L is the support length of the mother wavelet, N is the total number of samples in the set period, k∈[0,K j ], K j is the maximum translation parameter, whose value is Round down; for The nth sampling data of the corresponding parameter within the set period; for The corresponding parameters are the wavelet basis functions when the scale is j and the time period is k; is the set mother wavelet function; Δt is the time interval between the two sampling data; for The corresponding parameter is the energy density at scale j.

[0060] In this embodiment, preferably, all time series features of all set points are normalized to form a feature matrix, and the element in the u-th row and j-th column of the feature matrix is ​​the v-th time series feature of the u-th set point; the feature matrix is ​​subjected to principal component analysis for dimensionality reduction, and the feature matrix after dimensionality reduction is the filtered time series feature.

[0061] Preferably, in this embodiment, the constant temperature and humidity factory is partitioned based on the personnel density and the heating power of each device, and the location of each constant temperature and humidity air conditioner is obtained. An importance weight is set for each constant temperature and humidity air conditioner according to the location and partition, specifically:

[0062] If a constant temperature and humidity factory contains temperature-sensitive equipment or humidity-sensitive equipment, and the temperature-sensitive equipment and / or humidity-sensitive equipment respectively have temperature and humidity accuracy requirements higher than a set accuracy threshold, the area where the equipment is installed is set as the core area; outside the core area, areas where the population density exceeds the set density threshold and / or the average heating power of the equipment exceeds the set heating power threshold are set as dense areas; other areas are ordinary areas;

[0063] Set weights for the three areas. The weight of the core area is set to be greater than that of the dense area, and the weight of the dense area is set to be greater than that of the ordinary area.

[0064] Clustering is performed according to the location of each constant temperature and humidity air conditioner to obtain the area where each cluster center is located. The importance weight of all constant temperature and humidity air conditioners in the cluster corresponding to the cluster center is equal to the weight of the area where the cluster center is located.

[0065] In this embodiment, the performance coefficient of each constant temperature and humidity air conditioner is preferably calculated based on the cooling and heating loads and dehumidification capacity and the real-time detected external environmental parameters, specifically:

[0066] External environmental parameters include external temperature and humidity;

[0067] The performance coefficient COP calculation formula is:

[0068]

[0069] Where Q c , Q h , D are cooling load, heating load and dehumidification capacity respectively, Δh is the air specific enthalpy difference, which is the difference in specific enthalpy between the air entering the air conditioner and leaving the air conditioner, T e 、φ e are the external temperature and humidity respectively, α and β are the set coefficients, Tth The temperature of the hardware that affects the air conditioner is set. Specifically, in this embodiment, α and β are set to 0.015 and 0.005 respectively. th Set to 35°C.

[0070] It should be noted that the calculation formula for the specific enthalpy h of the air is:

[0071] h=1.006T D +(2501+1.86T D )·σ H

[0072] Among them, T D is the dry bulb temperature, σ H For moisture content.

[0073] It should be noted that the unit of Δh is usually kJ / kg, which must be converted into kW when calculating the performance coefficient.

[0074] In this embodiment, preferably, the reward function is constructed by combining the weighted results and the environmental parameters of the constant temperature and humidity factory workshop detected in real time to perform multi-objective optimization, specifically:

[0075] The reward function R is calculated as follows:

[0076]

[0077] Where, COP b is the set benchmark energy efficiency, COP′ is the weighted result, k c is a coefficient with a range of (0,1], ω T 、ω H For the set temperature weight and humidity weight, it should be noted that ω T 、ω H are all (0,1] and add up to 1, U is the total number of set points, T u 、H u are the temperature and humidity detected at the u-th set point respectively; T u 、H set are the target temperature and humidity values ​​respectively, f is the constraint penalty term, λ is the set penalty term weight, and exp is the exponential operation with the natural exponent e as the base.

[0078] In this embodiment, preferably, the constraint penalty item is specifically:

[0079] Real-time detection of the current actual water pump frequency, air valve opening, and temperature and humidity of each set point;

[0080] Obtain all set points whose temperatures are greater than the maximum allowable temperature or less than the minimum allowable temperature, calculate the difference between their temperatures and the corresponding maximum allowable temperature or minimum allowable temperature, and calculate the average of these differences to obtain the temperature penalty term; obtain all set points whose humidity is greater than the maximum allowable humidity or less than the minimum allowable humidity, calculate the difference between their humidity and the corresponding maximum allowable temperature or minimum allowable temperature, and calculate the average of these differences to obtain the humidity penalty term;

[0081] Calculate the difference between the adjusted water pump frequency and air valve opening of the current iterative setting of all set points and the actual current water pump frequency and air valve opening to obtain the water pump frequency difference and air valve opening difference. If the water pump frequency difference is greater than the water pump frequency error threshold, calculate the difference between all corresponding water pump frequency differences and the water pump frequency error threshold, and calculate the average value as the water pump penalty item; if the air valve opening difference is greater than the air valve opening error threshold, calculate the difference between all corresponding air valve opening differences and the air valve opening error threshold, and calculate the average value as the air valve penalty item;

[0082] The constraint penalty items are obtained by normalizing each penalty item and performing weighted summation. The weights of the temperature penalty item and the water pump penalty item are both the set temperature weight; the weights of the humidity penalty item and the air valve penalty item are both the set humidity weight.

[0083] It should be noted that this embodiment also detects the current actual compressor start / stop status in real time. If the actual compressor start / stop status adjusted by the current iterative setting of the set point is different from the actual compressor start / stop status, an alarm is issued.

[0084] In this embodiment, it should be noted that, during each iteration of the reinforcement learning algorithm, not all operating parameters of each constant temperature and humidity air conditioner are updated and adjusted; when adjusting the operating parameters of each constant temperature and humidity air conditioner, the current performance coefficients of all constant temperature and humidity air conditioners are normalized, and the normalization result is the probability of adjusting the operating parameters of the corresponding constant temperature and humidity air conditioner.

[0085] Example 2 of the present invention proposes a control system for optimizing the energy efficiency of a constant temperature and humidity factory workshop air conditioning system based on the method described in Example 1 of the present invention, including a load forecasting model construction module, a weight setting module, and a multi-objective optimization module, specifically:

[0086] Load forecasting model construction module: This module obtains the cooling and heating loads and dehumidification capacity of constant temperature and humidity air conditioners under different constant temperature and humidity factory workshop environmental parameters and different constant temperature and humidity air conditioner operating parameters, and uses these as samples to train the load forecasting model;

[0087] Weight setting module: This module divides the constant temperature and humidity factory into zones based on the density of personnel and the heat output of each device, obtains the location of each constant temperature and humidity air conditioner, and sets the importance weight for each constant temperature and humidity air conditioner based on the location and zone.

[0088] Multi-objective optimization module: Based on the reinforcement learning algorithm, multi-objective optimization is performed on the air-conditioning system composed of all constant temperature and humidity air conditioners. The reinforcement learning algorithm adjusts the operating parameters of each constant temperature and humidity air conditioner as an action, and inputs the action of each iteration and the environmental parameters of the constant temperature and humidity factory workshop detected in real time into the load prediction model to obtain the predicted cooling, heating load and dehumidification capacity of each constant temperature and humidity air conditioner. The performance coefficient of each constant temperature and humidity air conditioner is calculated based on the cooling, heating load and dehumidification capacity and the real-time detected external environmental parameters; the performance coefficient of each constant temperature and humidity air conditioner is weighted according to the importance weight, and the reward function is constructed by combining the weighted result and the real-time detected environmental parameters of the constant temperature and humidity factory workshop to perform multi-objective optimization.

[0089] The present disclosure may be a system, method and / or computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A control method for optimizing the energy efficiency of a constant temperature and humidity factory workshop air conditioning system, characterized in that: Includes the following: Obtain the environmental parameters of different constant temperature and humidity factory workshops and the operating parameters of different constant temperature and humidity air conditioners, as well as the corresponding constant temperature and humidity air conditioner cooling, heating load and dehumidification capacity, and use them as samples to train the load prediction model; The constant temperature and humidity factory is divided into zones based on the density of personnel and the heat output of each device. The location of each constant temperature and humidity air conditioner is obtained, and the importance weight of each constant temperature and humidity air conditioner is set according to the location and zone. Based on the reinforcement learning algorithm, multi-objective optimization is performed on the air-conditioning system composed of all constant temperature and humidity air conditioners. The adjustment of the operating parameters of each constant temperature and humidity air conditioner is taken as an action. The action of each iteration and the environmental parameters of the constant temperature and humidity factory workshop detected in real time are input into the load prediction model to obtain the predicted cooling, heating load and dehumidification capacity of each constant temperature and humidity air conditioner. The performance coefficient of each constant temperature and humidity air conditioner is calculated based on the cooling, heating load and dehumidification capacity and the real-time detected external environmental parameters; the performance coefficient of each constant temperature and humidity air conditioner is weighted according to the importance weight, and the reward function is constructed by combining the weighted result and the real-time detected environmental parameters of the constant temperature and humidity factory workshop to perform multi-objective optimization.

2. The control method for optimizing energy efficiency of a constant temperature and humidity factory workshop air conditioning system according to claim 1, characterized in that: Temperature and humidity sensors are placed at various set points in the constant temperature and humidity factory workshop. The environmental parameters of the constant temperature and humidity factory workshop include the temperature and humidity values ​​of each set point in the constant temperature and humidity factory workshop within a set period. The operating parameters of the constant temperature and humidity air conditioner include the start and stop status of the compressor, the water pump frequency, and the air valve opening. The time series features and spatial features of the environmental parameters of the constant temperature and humidity factory workshop are extracted, and the time series features are screened. The screened time series features are spliced ​​with the spatial features, the environmental parameters of the constant temperature and humidity factory workshop, and the operating parameters of the constant temperature and humidity air conditioner as the feature set of the sample. The corresponding cooling, heating loads and dehumidification amounts are used as labels to train a load prediction model. The load prediction model adopts the Transformer model.

3. The control method for optimizing energy efficiency of a constant temperature and humidity factory workshop air conditioning system according to claim 2, characterized in that: The temporal and spatial features of the environmental parameters of the constant temperature and humidity factory workshop are extracted as follows: The time series characteristics include: the temperature change rate, humidity change rate, temperature standard deviation, humidity standard deviation, temperature pulsation characteristics, and humidity pulsation characteristics of each set point within a set period; the temperature pulsation characteristics are obtained by performing discrete wavelet transform on the temperature of each set point within a set period to obtain energy density at multiple scales, and the energy density at each scale is used as the temperature pulsation characteristic of the corresponding scale, and each temperature pulsation characteristic is a time series feature; the humidity pulsation characteristics are obtained by performing discrete wavelet transform on the humidity of each set point within a set period to obtain energy density at multiple scales, and the energy density at each scale is used as the humidity pulsation characteristic of the corresponding scale, and each humidity pulsation characteristic is a time series feature; The spatial characteristics include a temperature uniformity index, a temperature stratification index, and a humidity gradient field; the temperature uniformity index is the variance of the temperature at each point, and the temperature stratification index is the difference between the average temperature at the top floor and the average temperature at the bottom floor of the constant temperature and humidity factory workshop divided by the height of the constant temperature and humidity factory workshop; the humidity gradient field is a three-dimensional coordinate, and the three elements represent the rate of change of humidity at each point in the constant temperature and humidity factory workshop on the x-axis, y-axis, and z-axis, respectively.

4. The control method for optimizing energy efficiency of a constant temperature and humidity factory workshop air conditioning system according to claim 2, characterized in that: All the time series features of all the set points are normalized to form a feature matrix, and the element in the u-th row and j-th column of the feature matrix is ​​the v-th time series feature of the u-th set point; The feature matrix is ​​subjected to principal component analysis to reduce its dimension, and the feature matrix after dimension reduction is the filtered time series feature.

5. The control method for optimizing energy efficiency of a constant temperature and humidity factory workshop air conditioning system according to claim 2, characterized in that: The constant temperature and humidity factory is partitioned based on the personnel density and the heating power of each device, and the location of each constant temperature and humidity air conditioner is obtained. The importance weight is set for each constant temperature and humidity air conditioner according to the location and partition, specifically: If a constant temperature and humidity factory contains temperature-sensitive equipment or humidity-sensitive equipment, and the temperature-sensitive equipment and / or humidity-sensitive equipment respectively have temperature and humidity accuracy requirements higher than a set accuracy threshold, the area where the equipment is installed is set as the core area; outside the core area, areas where the population density exceeds the set density threshold and / or the average heating power of the equipment exceeds the set heating power threshold are set as dense areas; other areas are ordinary areas; Set weights for the three areas. The weight of the core area is set to be greater than that of the dense area, and the weight of the dense area is set to be greater than that of the ordinary area. Clustering is performed according to the location of each constant temperature and humidity air conditioner to obtain the area where each cluster center is located. The importance weight of all constant temperature and humidity air conditioners in the cluster corresponding to the cluster center is equal to the weight of the area where the cluster center is located.

6. The control method for optimizing energy efficiency of a constant temperature and humidity factory workshop air conditioning system according to claim 2, characterized in that: The performance coefficient of each constant temperature and humidity air conditioner is calculated based on the cooling and heating loads, dehumidification capacity and real-time detected external environmental parameters. Specifically: External environmental parameters include external temperature and humidity; The performance coefficient COP calculation formula is: Where Q c , Q h , D are cooling load, heating load and dehumidification capacity respectively, Δh is the air specific enthalpy difference, which is the difference in specific enthalpy between the air entering the air conditioner and leaving the air conditioner, T e 、φ e are the external temperature and humidity respectively, α and β are the set coefficients, T th This setting affects the operating temperature of air-conditioning related hardware.

7. The control method for optimizing energy efficiency of a constant temperature and humidity factory workshop air conditioning system according to claim 2, characterized in that: The reward function is constructed by combining the weighted results and the environmental parameters of the constant temperature and humidity factory workshop detected in real time to perform multi-objective optimization, specifically: The reward function R is calculated as follows: Where, COP b is the set benchmark energy efficiency, COP′ is the weighted result, k c is a coefficient with a range of (0,1], ω T 、ω H is the set temperature weight and humidity weight, U is the total number of set points, T u 、H u are the temperature and humidity detected at the u-th set point respectively; T u 、H set are the target temperature and humidity values ​​respectively, f is the constraint penalty term, λ is the set penalty term weight, and exp is the exponential operation with the natural exponent e as the base.

8. The control method for optimizing energy efficiency of a constant temperature and humidity factory workshop air conditioning system according to claim 7, characterized in that: The constraint penalty items are specifically: Real-time detection of the current actual water pump frequency, air valve opening, and temperature and humidity of each set point; Obtain all set points whose temperatures are greater than the maximum allowable temperature or less than the minimum allowable temperature, calculate the difference between their temperatures and the corresponding maximum allowable temperature or minimum allowable temperature, and calculate the average of these differences to obtain the temperature penalty term; obtain all set points whose humidity is greater than the maximum allowable humidity or less than the minimum allowable humidity, calculate the difference between their humidity and the corresponding maximum allowable temperature or minimum allowable temperature, and calculate the average of these differences to obtain the humidity penalty term; Calculate the difference between the adjusted water pump frequency and air valve opening of the current iterative setting of all set points and the actual current water pump frequency and air valve opening to obtain the water pump frequency difference and air valve opening difference. If the water pump frequency difference is greater than the water pump frequency error threshold, calculate the difference between all corresponding water pump frequency differences and the water pump frequency error threshold, and calculate the average value as the water pump penalty item; if the air valve opening difference is greater than the air valve opening error threshold, calculate the difference between all corresponding air valve opening differences and the air valve opening error threshold, and calculate the average value as the air valve penalty item; The constraint penalty items are obtained by normalizing each penalty item and performing weighted summation. The weights of the temperature penalty item and the water pump penalty item are both the set temperature weight; the weights of the humidity penalty item and the air valve penalty item are both the set humidity weight.

9. The control method for optimizing energy efficiency of a constant temperature and humidity factory workshop air conditioning system according to claim 1, characterized in that: When the reinforcement learning algorithm iteratively adjusts the operating parameters of each constant temperature and humidity air conditioner, the current performance coefficients of all constant temperature and humidity air conditioners are normalized, and the normalized result is the probability of adjusting the operating parameters of the corresponding constant temperature and humidity air conditioner.

10. A control system for optimizing energy efficiency of a constant temperature and humidity factory workshop air conditioning system based on the method according to any one of claims 1 to 9, comprising a load forecasting model building module, a weight setting module, and a multi-objective optimization module, characterized in that: Load forecasting model construction module: This module obtains the environmental parameters of different constant temperature and humidity factory workshops and the operating parameters of different constant temperature and humidity air conditioners, as well as the corresponding constant temperature and humidity air conditioners' cooling and heating loads and dehumidification capacity, and uses them as samples to train the load forecasting model. Weight setting module: This module divides the constant temperature and humidity factory into zones based on the density of personnel and the heat output of each device, obtains the location of each constant temperature and humidity air conditioner, and sets the importance weight for each constant temperature and humidity air conditioner based on the location and zone. Multi-objective optimization module: Based on the reinforcement learning algorithm, multi-objective optimization is performed on the air-conditioning system composed of all constant temperature and humidity air conditioners. The operating parameters of each constant temperature and humidity air conditioner are adjusted as actions. The actions of each iteration and the environmental parameters of the constant temperature and humidity factory workshop detected in real time are input into the load prediction model to obtain the predicted cooling, heating load and dehumidification capacity of each constant temperature and humidity air conditioner. The performance coefficient of each constant temperature and humidity air conditioner is calculated based on the cooling, heating load and dehumidification capacity and the real-time detected external environmental parameters; the performance coefficient of each constant temperature and humidity air conditioner is weighted according to the importance weight, and the reward function is constructed by combining the weighted results and the real-time detected environmental parameters of the constant temperature and humidity factory workshop to perform multi-objective optimization.

Images