Diagnostic events in thermal systems

Abstract

Embodiments of this disclosure provide a thermal model based on an adaptive filter bank for characterizing the volumetric heat transfer of a thermal system. In one embodiment, the adaptive filter bank is used to provide diagnostic information related to the conditions of the thermal system. The diagnostic is based on an analysis of the heat transfer characteristics of a dynamic representation of the thermal system. According to an embodiment, thermal coefficients are generated based on the adaptive filter bank. One or more filters are applied to the thermal coefficients based on a sampling rate, and one or more estimated thermal coefficient thresholds are generated based on the sampling rate. It is determined whether at least one of the filtered thermal coefficients satisfies at least one of the estimated thermal coefficient thresholds. Therefore, based on this determination, alarm information indicating a diagnostic event is provided.

Description

[0001] Cross-reference of related applications

[0002] This application is a continuation-to-file of U.S. Patent Application No. 16 / 519,751, filed July 23, 2019, the entire contents of which are incorporated herein by reference. Technical Field

[0003] This disclosure generally relates to thermal models, and more specifically, to the detection of diagnostic events in thermal systems. Background Technology

[0004] Users frequently use devices to control or modify the temperature in certain locations, such as residences. However, such devices may not be able to learn or represent the heat transfer characteristics of a thermal system. Thermal models utilize observed physical relationships related to device properties and weather estimates to characterize heat transfer and predict temperature and power consumption, which helps in optimal temperature control while improving thermal comfort and reducing power consumption. Summary of the Invention

[0005] According to this disclosure, a thermal modeling solution is provided. The thermal modeling solution implements an adaptive filter bank to characterize heat transfer associated with one or more volumes of a thermal system. In an implementation, the thermal model interacts with one or more thermal devices in a zone and a site-associated weather model to construct, adapt, and validate a set of thermal coefficients that characterize heat transfer and utilize observed relationships between environmental conditions, including temperature and power consumption. The thermal model does not require commissioning information to characterize heat transfer; instead, it extracts information entirely from passive observations of data provided by the thermal devices and weather models. For example, the thermal model uses passive observations to determine site-associated heat transfer data based on relationships between environmental conditions, temperature, power consumption, and other types of physical properties associated with the zone. In some cases, certain information about the zone may be unknown or reported as unreliable, such as the specific geometry of each zone and the thermal mass and heat transfer characteristics of the boundary materials. Such unknown or unreliable information may lead to errors in the characterization of heat transfer, potentially resulting in inefficient or wasteful energy use when the system manages the site based on comfort considerations associated with the system architecture.

[0006] Embodiments of this disclosure address the aforementioned and other drawbacks by implementing an adaptive filter bank to minimize estimation errors that may occur when characterizing heat transfer associated with one or more volumes of a thermal system. An example application of the adaptive filter bank can be used to provide diagnostics related to the conditions of a thermal system. Diagnostics can be based on analysis of the heat transfer characteristics of a dynamic representation of the thermal system. In this regard, diagnostic events can indicate the detection of changes in observed or estimated characteristics of the thermal system, which may be related to defective or anomalous operation.

[0007] In one embodiment, a method is provided. The method includes: generating thermal coefficients at an adaptive filter bank by a controller device to characterize heat transfer of a volume associated with a thermal system; applying one or more filters to the thermal coefficients by the controller device based on a sampling rate; generating one or more estimated thermal coefficient thresholds by the controller device based on the sampling rate in response to the application of the filters; determining whether at least one of the filtered thermal coefficients satisfies at least one of the estimated thermal coefficient thresholds; and providing alarm information indicating a diagnostic event associated with the thermal system based on the determination.

[0008] In another embodiment, a system is provided. The system includes: a memory storing data of a plurality of thermal coefficients; and a controller device operatively coupled to the memory for: generating thermal coefficients at an adaptive filter bank to characterize heat transfer of a volume associated with a thermal system; applying one or more filters to the thermal coefficients based on a sampling rate; generating one or more estimated thermal coefficient thresholds based on the sampling rate in response to the application of the filters; determining whether at least one of the filtered thermal coefficients satisfies at least one of the estimated thermal coefficient thresholds; and providing alarm information indicating a diagnostic event associated with the thermal system based on the determination.

[0009] In another embodiment, a non-transitory computer-readable storage medium is provided. The non-transitory computer-readable storage medium includes executable instructions that, when executed by a controller device, cause the controller device to: apply one or more filters to thermal coefficients based on a sampling rate; generate one or more estimated thermal coefficient thresholds based on the sampling rate in response to the application of the filters; determine whether at least one of the filtered thermal coefficients satisfies at least one of the estimated thermal coefficient thresholds; and provide alarm information indicating diagnostic events associated with the thermal system based on the determination.

[0010] In one example, generating thermal coefficients further includes: identifying a thermal model for the application using a thermal system; determining, based on the thermal model, the estimation error of the reference signal relative to the master signal associated with the volume; and adapting the thermal coefficients to satisfy the solution associated with the volume based on an adaptive filter, taking into account the estimation error. In another example, the sampling rate is adapted based on a filter operation associated with at least one of the filters, wherein the filter operation includes at least one infinite impulse response filter. In yet another example, generating estimated thermal coefficient thresholds further includes: determining whether the sequence of thermal coefficient vectors supports the definition of a diagnostic event. Furthermore, it is determined whether at least one thermal coefficient exceeds an upper or lower bound window associated with at least one estimated thermal coefficient threshold. At this point, an alarm message indicates abnormal operation of the device associated with the volume.

[0011] The summary is provided to introduce some concepts in a simplified form, which will be further described in the detailed description below. The summary is not intended to identify key or essential features of the claimed subject matter, nor is it intended to help determine the scope of the claimed subject matter. Other features will be apparent in part and are indicated in part below. Attached Figure Description

[0012] A more detailed description of the present disclosure, which has been briefly outlined above, can be obtained by referring to some of the various embodiments shown in the accompanying drawings. While the drawings illustrate selected embodiments of the present disclosure, they should not be considered as limiting its scope, as the present disclosure may allow for other equally effective embodiments.

[0013] Figure 1 This is a block diagram illustrating a system architecture associated with a thermal system according to an embodiment of the present disclosure.

[0014] Figure 2 This is an example of an adaptive filter according to an embodiment of the present disclosure.

[0015] Figure 3 This is an example of an adaptive filter bank according to embodiments of the present disclosure.

[0016] Figure 4 This is an example graph illustrating the solution of the adaptive filter for the thermal model according to an embodiment of the present disclosure.

[0017] Figure 5 This is a flowchart of an adaptive filter bank method for modeling a thermal system according to an embodiment of the present disclosure.

[0018] Figure 6 This is a flowchart of another method for modeling a thermal system using an adaptive filter bank, according to embodiments of the present disclosure.

[0019] Figure 7 This is a flowchart of a method for detecting diagnostic events in a thermal system based on an adaptive filter bank, according to embodiments of the present disclosure.

[0020] Figures 8A-8D This is an example graph illustrating diagnostic event data according to an embodiment of the present disclosure.

[0021] Figure 9 This is a block diagram illustrating a machine in which embodiments of the present disclosure can be used.

[0022] Where possible, the same reference numerals are used to denote the same elements in the drawings. However, elements disclosed in one embodiment may be advantageously used in other embodiments without specific description. Detailed Implementation

[0023] Various aspects of this disclosure generally relate to implementing thermal models based on adaptive filter banks to characterize the volumetric heat transfer of a thermal system. The techniques of this disclosure can use adaptive filter banks to improve the performance characteristics, diagnostics, energy-saving estimation, and optimal temperature control (including optimal start-up and optimal shutdown) of applications used to control desired temperatures in one or more volumes of a thermal system. According to this disclosure, adaptive filter banks help minimize estimation errors that may occur when characterizing heat transfer. In one embodiment, the adaptive filter bank may consume an incident signal characterizing the heat transfer properties associated with the volume of the thermal system, generate a reference signal to estimate the observed master signal, calculate the estimation error, and adapt thermal coefficients to minimize the estimation error.

[0024] The incident signal comprises observational data and weather estimates related to active heat transfer, passive heat transfer, solar irradiation heat transfer, and unobserved heat transfer in one or more volumes of the thermal system. The thermal coefficient vector represents the heat transfer characteristics associated with a specific volume of the thermal system, including passive, active, solar irradiation, and unobserved heat transfer. The thermal coefficient vector can be initialized to represent the typical or expected heat transfer characteristics of the volume and adapted to improve the reference signal estimate. The reference signal represents the estimated rate of temperature change associated with the volume of the thermal system and is generated based on a differential equation of state based on the incident signal, the thermal coefficient vector, and an adaptive filter bank. The main signal represents the observed rate of temperature change in the volume of the thermal system, which can be calculated based on the temperature received from a thermal control unit monitoring the temperature and other relevant data associated with the volume.

[0025] The estimation error is calculated based on the difference between the master signal and the reference signal, and represents the estimation error associated with estimating the master signal using the reference signal from the adaptive filter bank under a specific state. The thermal coefficient vector is adapted based on the estimation error to minimize subsequent estimation errors according to a specified metric. After a series of adjustments, the thermal model coefficients converge to represent the heat transfer characteristics of the volume of the thermal system. Depending on the operating conditions of the thermal system, the adaptation can be continuous, as convergence can be stationary, quasi-stationary, or dynamic, and the estimation error generally improves with increasing diversity of incident signal observations.

[0026] Example applications of adaptive filter banks can be used to provide diagnostics of information related to the conditions of a thermal system. This diagnostic can be based on an analysis of the heat transfer characteristics of a dynamic representation of the thermal system. In this regard, diagnostic events can indicate the detection of changes in observed or estimated characteristics of the thermal system, which may be related to defective or anomalous operation. In some embodiments, diagnostic events can be generated from the evaluation of observations and estimates over various time intervals supporting, for example, transient or continuous classification. For example, transient diagnostic event detection may include pump, valve, or relay failures in a circulating heating thermal system, or relay failures or rapid refrigerant leaks in an air-based furnace, heat pump, or air conditioning system. In other examples, continuous diagnostic event detection includes slow leaks in a circulating boiler, pipe, or radiator, or sludge buildup due to oxidation, or slow refrigerant leaks, damaged heat exchangers, or contaminated pipes and filters in an air-based system. In some examples, transient diagnostic events may exhibit periodic, quasi-periodic, or persistent behavior, which could be caused by conditions including slow leaks in the circulation system, as the occupant may become aware of the defect and choose to occasionally change the water and restore normal operation, or allow the defect to evolve into persistent and increasingly abnormal operation.

[0027] To perform diagnostics on a thermal system, a thermal model is constructed, initialized, and adapted on a zone-by-zone basis (e.g., using an adaptive filter bank) to generate dynamic estimates of thermal coefficient data. This thermal coefficient data is used to describe and estimate the behavior of the thermal system. A filter is then applied to the thermal coefficient data based on the sampling rate. For example, the sampling rate can be adapted and selectively combined with filtering operations to reduce aliasing and implement rate adaptation, thereby transforming the thermal coefficient data into an effective sampling period. In response to the application of the filter, one or more estimated thermal coefficient thresholds are generated. Therefore, based on whether the thermal coefficient data meets the corresponding estimated thermal coefficient threshold, a diagnostic event associated with the thermal system is determined. In this case, an alarm or warning message is generated, indicating that some components of the thermal system may have failed due to abnormal conditions or suffered abnormal operation.

[0028] System Architecture

[0029] Figure 1 This is a block diagram illustrating a system architecture 100 associated with a thermal system according to an embodiment of the present disclosure. The thermal system (also referred to as site 101) can be described as a collection of interdependent volumes whose thermal behavior is described by the transfer of mass, work, and heat across volume boundaries. Each volume is defined as a zone 103 comprising a continuous area of ​​uniform thermal control. In some embodiments, system architecture 100 can implement a system for comfort-based management of the thermal system, including residential and commercial buildings described with active cooling and / or heating, with an emphasis on practical applications.

[0030] In some embodiments, comfort-based management associated with system architecture 100 adaptively and continuously learns the heat transfer characteristics of the thermal system and the thermal comfort characteristics of the occupant to facilitate optimal temperature control, i.e., minimizing power consumption while maintaining thermal comfort. Although embodiments of this disclosure are described with respect to a particular type of system, this should not be construed as limiting the scope or usefulness of the features of this disclosure. For example, the features and techniques described herein can be used with other types of systems, system architectures, local embedded controllers, and / or distributed cloud computing environments.

[0031] like Figure 1As shown, an illustrative example of system architecture 100 is implemented in a distributed cloud computing environment 105. The distributed cloud computing environment 105 supports scalable distributed computing and storage, where one or more remote computing nodes 107 and relatively simple distributed devices may include, for example, one or more thermal control units 102 that interact with one or more Heating, Ventilation, Air Conditioning (HVAC) units 104. The one or more remote computing nodes 107 can execute software and / or other processes. Each computing node 107 may refer to a virtual server, such as a physical machine, a memory partition, or any other type of computing environment, and provides a fully or partially isolated execution environment for the execution of applications.

[0032] The cloud computing environment 105 achieves scalability through flexible resource allocation, typically virtualizing servers 130 and compute nodes 107, which are abstracted from the physical hardware they reside on. For example, resources can be dynamically allocated and disposed of on demand, abstracting infrastructure complexity from compute nodes 107. In some embodiments, compute nodes 107 are virtual machines hosted on physical machines. Compute nodes 107 interact via network 106 with distributed devices such as thermal control units 102 and HVAC units 104 to facilitate data transport, archiving, and synchronization, and to manage compute and storage resources, thereby dynamically allocating and disposing of other services and jobs to support system architecture 100. In some embodiments, network 106 can be a private network (e.g., a local area network (LAN), Wi-Fi, Bluetooth, radio frequency, wide area network (WAN), intranet, etc.) or a public network (e.g., the Internet).

[0033] Server 109 manages the allocation and disposal of computing and storage resources, facilitates data transport and synchronization, and interacts with one or more weather services 110, one or more locations 101, and one or more computing nodes 107. Computing nodes 107 are collections of entities that implement analytical and control capabilities and interact with server 109 to perform comfort-based thermal management of one or more locations 101. Location 101 contains one or more zones 103. Each zone 103 contains one or more thermal control units 102, which interact with server 109 and one or more HVAC units 104. For each supported location 101, computing nodes 107 include a comfort agent 120, a comfort model 130 for each zone 103, a thermal model 140 for each zone 103, one or more thermal devices 150 for each zone 103, and a weather model 160.

[0034] Weather models

[0035] Weather model 160 interacts with server 109 to exchange information with weather service 110. In some embodiments, weather model 140 is a representation of weather service 110 that defines predictive estimates of attributes of weather conditions. These weather conditions may be associated with a specific region of interest, including the environment surrounding location 101. For example, weather estimates provided by weather service 110 may indicate cloud cover, humidity, solar irradiance, and / or temperature for one or more areas. Weather model 160 may indirectly estimate future conditions from weather service 110, which provides attributes at a default resolution (e.g., a nominal value of 1 hour) over a weather duration (e.g., a nominal value of 24 hours) and a limited geographic range supported, including location 101, location, city, region, etc. In some embodiments, weather model 160 may have a unique association with a specific region of interest, and attributes of weather service 110 may require spatial or temporal interpolation to achieve a specified effective resolution or improve accuracy.

[0036] Thermal equipment

[0037] Thermal device 150 interacts with server 109 to exchange information with thermal control unit 102 in zone 103. In some embodiments, thermal device 150 may be represented as a specific thermal control unit that observes and controls the temperature of a volume (e.g., zone 103) of a thermal system (e.g., location 101). For example, thermal control unit 102 may support active cooling and / or heating associated with the volume. For example, active cooling employs cooling-type HVAC unit 104 (e.g., including air conditioners, refrigerators, or freezers) to cool zone 103, while active heating employs heating-type HVAC unit 104 (e.g., including furnaces, heat pumps, resistance heating, or electric or circulating radiant heating) to heat zone 103. HVAC unit 104 may include any device capable of controlling temperature by generating, consuming, or transferring heat in zone 103. In this respect, active cooling and heating are mutually exclusive at a given time. In some embodiments, the thermal control unit 102 may support occupant interaction indicating a tendency to lower or raise the temperature associated with zone 103 of location 101. The thermal device 150 may provide an immediate and transient response in response to occupant interaction by lowering or raising the temperature by a temperature offset over a finite duration to increase the thermal comfort of the occupant.

[0038] Comfort Agent

[0039] Comfort agent 120 interacts with server 109 and observes one or more comfort models 130 and one or more thermal models 140 associated with location 101. Comfort agent 120 interacts with comfort models 130 to establish constraints for temperature control and with thermal models 140 to define physically achievable states and associated sets of energy or power transitions. This allows comfort agent 120 to contribute to optimal temperature control in each zone 103 of location 101, i.e., minimizing power consumption while maintaining thermal comfort. Optimal temperature control contributes to optimal start-up and shutdown settings for thermal equipment 150, i.e., dynamically advancing control temperature transitions associated with active cooling or heating to compensate for heat transfer delays. This ensures that advancing control temperature transitions continues to adhere to defined constraints that are physically achievable relative to expected environmental conditions and has the potential to reduce power consumption.

[0040] Comfort agent 120 interacts with comfort model 130 to establish temperature constraints limiting temperature control, and with thermal model 140 to define a set of physically realizable states and associated energy or power transitions to facilitate optimal temperature control in each zone 103 of location 101, i.e., minimizing power consumption while maintaining thermal comfort. In some embodiments, comfort agent 120 defines a cost function based on energy, power, currency, resource availability, or time, which is associated with an optimal path 125 formed by constrained state transitions over the control duration. Comfort agent 120 uses thermal model 140 in conjunction with comfort model 130 to select an optimal path, which represents a set of state transitions that minimize cost within the context of expected utility and environmental conditions. The specific method used by comfort agent 120 to select the optimal path can be consistent with applications of reinforcement machine learning.

[0041] The comfort agent 120 selects the optimal path 125 as the set of sequential states, where the epoch of the root state is the endpoint state at the end of the control duration N, and the action value vector from state index u to state index v at sample n, under cooling state c. The vector of state values ​​is selected to correspond to the vector of state values ​​to be allocated in the estimation. A specific transformation. Thus, the optimal path 125 is the one that makes the root state value vector... Minimize the state sequence. In some embodiments, the optimal path includes optimal start-up and optimal shutdown settings for the thermal device 150. Optimal start-up advances the control temperature transition to a higher energy state to ensure the device temperature T at sample n, zone 103i, observed at a specific control temperature transition. i,n Effective temperature T under cooling state c E,i,j,c,nControl offset T C Internally, the optimal stop is to push the control temperature transition to a lower energy state to ensure that the device temperature T at sample n, zone 103i, observed at a specific control temperature transition point is within the specified range. i,n Effective temperature T under cooling state c E,i,j,c,n Control offset T C The optimal start and optimal stop are integral behaviors expressed in optimal path 125, which is extracted from the set of states defining more than one path. However, in simple sequential solutions, the optimal start and optimal stop behaviors are selectable and separable.

[0042] Comfort Model

[0043] Comfort model 130 represents thermal comfort in zone 103. Comfort model 400 can estimate an effective temperature at which occupants are unlikely to object, thereby maintaining occupant comfort while minimizing energy consumption due to active cooling and / or heating. The effective temperature represents the temperature associated with the lowest thermal equipment power consumption at which occupants are unlikely to object. Estimation of the effective temperature facilitates applications including optimal temperature control, demand response, and virtual power plant capacity. In many respects, thermal comfort can be a subjective attribute that occupants in zone 103 may disagree on, and estimates at effective temperatures at which all occupants consistently express similar levels of satisfaction may not be entirely reasonable expectations.

[0044] Comfort model 130 addresses the problem of defining thermal comfort by constructing and adapting a set of temperature curves 137. Through interaction with occupants and by adapting the temperature curves 137 to characterize thermal comfort while minimizing power consumption due to active cooling or heating at location 101, the temperature curves 137 are used to learn behavioral patterns. Comfort model 130 requires no commissioning information, does not utilize occupancy estimates, and extracts the necessary information to characterize thermal comfort entirely from passive observations of data and events provided by associated thermal devices 150. For example, comfort model 130 can learn multiple independent temperature curves 137 through observations of temperature (e.g., the temperature of zone 103) and through occupant interactions, where occupant interactions indicate a tendency to lower or raise the temperature. Control states are associated with comfort model 130 to allow or prevent modifications to associated temperature curves 137, allowing the comfort-based thermal management system to continue observing and learning temperature and power consumption patterns even when the occupant has chosen to disable optimal temperature control. If the control state associated with comfort model 130 prevents modification of the associated temperature curve 137 related to the learned behavior, an alternative temperature curve reflecting a schedule of temperature transitions specified by the occupant is used. An independent temperature curve 137 associated with each day of the week can be defined by the occupant and is retained without modification until it is selectively replaced. The static temperature curve 137 also forms a suitable basis for applying optimal start-up and optimal shutdown behavior.

[0045] thermal model

[0046] Thermal model 140 is a representation of the thermal behavior of volumes in a thermal system, characterizing heat transfer data 142 and estimating energy consumption and temperature within the system. For example, a thermal system can be represented by location 101 and can be generalized as a collection of interdependent volumes defined as zones 103 with boundaries and surrounding environments, whose behavior is described by the transfer of mass, work, and heat across the boundaries. In some embodiments, when determining heat transfer data 142, the thermal system supported by thermal model 140 can reasonably apply the assumptions of open mechanical isolation and simple transient conduction. It should be noted that open mechanical isolation thermal systems do not support volume deformation and associated work transfer, or mass transfer across boundaries. If a thermal system (such as location 101) does not support deformation, and / or if the mass within the volume is quasi-stationary or changes slowly relative to the rate of heat transfer, it can be conveniently described as open and mechanically isolated. In this case, by assuming that heat conduction within the volume is much faster than heat transfer across the boundary, thermal model 140 can adopt simple instantaneous conduction by ignoring the temperature and pressure gradients within the volume (e.g., zone 103) to simplify the analysis.

[0047] Heat transfer

[0048] Regarding the heat transfer data 142 characterized by thermal model 140, the heat transfer to zone 103i at time t... It is a measure of the total rate of change of thermal energy from various sources, including passive heat transfer. Active heat transfer Solar heat transfer and unobserved heat transfer Heat transfer This can be represented by the following equation:

[0049]

[0050] In a site with J-1 zones, at time t, passive heat transfer to zone 103i Equal to the thermal transmittance h Pi,j Surface area A of the shared zone boundary Pi,j and the temperature T of zone j j,t Temperature T in zone j i,t The product of the differences between them, where the heat transfer coefficient h Pi,j It is a property related to the heat transfer characteristics of the boundary material, passive heat transfer. This can be represented by the following equation:

[0051]

[0052] Referring to the foregoing, passive heat transfer relative to the surrounding environment follows Newton's law of cooling in a manner similar to inter-zone passive heat transfer, and therefore can be modeled as independent zones for ease of illustration. In the embodiment, passive heat transfer from adjacent zone j across the boundary to zone 103i is extended by superposition to aggregate passive heat transfer from each zone j, each zone j possibly sharing a boundary with zone 103i, although the relative geometry and connectivity of the zones and the site are unknown. If the observed temperatures of zone j and zone i are sufficiently similar, inter-zone passive heat transfer between zone j and zone i can be selectively ignored, as the resulting temperature difference may be relatively insignificant and insufficiently diverse, leading to poor solution conditions.

[0053] Active heat transfer to zone 103i under time t, cooling state c. Thermal efficiency η is equal to the ratio of the power produced or transported. i,c,k The power P consumed in the cell with index k out of K cells i,c,k,t The total product of active heat transfer This can be represented by the following equation:

[0054]

[0055] Referring to the above equations, if the thermal efficiency of the cell with index k is relatively constant under the observed operating conditions, then the essentially uniform thermal efficiency η relative to power consumption or instantaneous environmental conditions (including the temperature difference between the surrounding environment and zone 103i) is... i,c,k This can be used as a simplifying assumption. Multimode HVAC units, including heat pumps, may exhibit non-uniform thermal efficiency relative to power consumption, typically displaying discontinuous or nonlinear thermal efficiency under observed operating conditions. In such environments, the described thermal model indirectly forms a general estimate of thermal efficiency extracted from specific observations.

[0056] Power estimation P in zone 103i at time t, cell index k, cooling state c. i,c,k,t This presents a dilemma because one or more units can exhibit similar thermal efficiency η. i,c,k Due to similar expected efficiencies, a simplifying assumption can be defined as follows: power is consumed in one or more cells grouped into main cells, and the remaining cells are grouped into one or more auxiliary cells. Alternatively, cells can be prioritized by relative efficiency or available capacity, and these cells can be selectively used to implement alternative temperature control schemes while minimizing power consumption.

[0057] In some environments, power may not be directly observable or measurable because the average value of the power consumed within the sampling period is T. S In such a system, power can be abstracted into a normalized representation defining the duty cycle, which is indirectly estimated from the relay state. Independent of the source or cell, power should generally be normalized by the maximum available power to improve the numerical accuracy of the solution.

[0058] The cooling state c, indicating whether active cooling or active heating is enabled, implies the thermal efficiency η in the cell at index k. i,c,k Independent estimation is necessary because active cooling and active heating can be performed by physically different units. Even if the same HVAC unit is used to perform active cooling and active heating, it is unreasonable to assume that these functions are performed with similar efficiency. Therefore, the solution must be independently formed and associated with a specific cooling state c, which is derived from observations limited to active cooling or active heating operation.

[0059] At time t, the solar thermal transfer from surface s to zone 103i The heat transfer coefficient h equal to the surface s Si,s The surface area A of the shared boundary exposed to incident solar radiation.Si,s Cloud cover i,t The complement of the available solar irradiance I i,s,t The total product, where the heat transfer coefficient h Si,s It is a property related to the heat transfer characteristics of surface materials, cloud cover c i,t Solar heat transfer is defined as the normalized ratio of solar irradiance obscured by clouds. This can be represented by the following equation:

[0060]

[0061] Since the relative geometries of the various zones within the site are unknown, and the geometry and heat transfer characteristics of the specific surfaces forming the boundary of zone i are also unknown, directly calculating the solar heat transfer of each surface s is impractical. The simplified assumptions of a stable geometry and boundary material in zone i, as well as uniform cloud cover and incident solar irradiance relative to the zone, are useful and generally sufficient to simplify the solution. Furthermore, the location of the site, and taking into account losses due to cloud cover, allows for easy provision of the solar irradiance incident on a single, orthogonal accumulation surface by weather services, further simplifying the understanding.

[0062] Unobserved heat transfer to zone 103i at time t Equal to the unobserved heat q i,t Unobserved heat q i,t This refers to heat generated or transported by unknown, unmeasured, or unobservable sources (including fireplaces, ovens, stoves, lighting, uninstrumented HVAC units 104, or people); unobserved heat transfer. This can be represented by the following equation:

[0063]

[0064] If the unobserved heat is known or determined to be negligible, then the unobserved heat transfer in zone 103i can be reasonably ignored. In addition, the unobserved heat q i The simplification that the memory depth N of the solution changes slowly over time and can be considered quasi-stationary is potentially useful or necessary for forming well-conditional solutions.

[0065] Heat transfer to zone 103i at time t and under cooling state c. This is usually not directly observable. The transformation from continuous differential equations to equivalent causal discrete-time difference equations, and the substitution of heat transfer at point n in zone i. Defines a practical and useful relationship for heat transfer. Equal to thermal mass Ci and rate of temperature change The product of the product of the temperature change rate (T i,n It can be represented by the following equation:

[0066]

[0067] It is not necessary to explicitly determine the rate of temperature change. Several relevant thermal system properties, including the passive heat transfer coefficient h Pi,j Surface area A of the shared zone boundary Pi,j Thermal efficiency η i,c,k Solar heat transfer coefficient hs i,j The surface area A of the shared boundary exposed to incident solar radiation. Si,s and thermal mass C i While specific parameter resolution may have potential significance in a diagnostic context, it is neither necessary nor useful for characterizing the observed relationship between heat transfer and environmental conditions, including temperature and power consumption.

[0068] The thermal coefficient vector at sample n, under cooling state c, in zone 103i The relevant characteristics of the system are abstracted from the above equations. For conceptual convenience, the thermal coefficient vector is used. It can be represented by the following equation:

[0069]

[0070] In some embodiments, the thermal coefficient vector at sample n, under cooling state c, in zone 103i This includes thermal coefficients related to passive heat transfer, active heat transfer, solar heat transfer, and unobserved heat transfer. Passive heat transfer is indexed by zone j out of J zones sharing a boundary with zone i, while active heat transfer is indexed by cell index k out of K cells. Passive thermal coefficients represent heat transfer from zone j to zone i, excluding the condition j = i and including the surrounding environment. Active thermal coefficients must be evaluated independently in the context of active cooling or active heating, which typically requires independent evaluation of the thermal coefficient vector.

[0071] Based on the temperature T in zone j at sample n-1 j,n-1 Temperature T in zone 103i i,n-1 Power P at cell index k i,c,k,n and solar irradiance I i,n-1 The incident vector at sample n, under cooling state c, and in zone i. It can be represented by the following equation:

[0072]

[0073] Referring to the equations above, by applying an appropriate low-pass filter, the power P in zone 103i at sample n, cell index k, cooling state c, can be optimized. i,c,k,n Filtering is often useful. A representative power filter is the second-order Butterworth IIR filter, with a -3dB normalized frequency of 0.2250790799 and a nominal group delay of 2.0 samples. Typically, various units of measurement are abstracted by scaling the constituent to ensure approximation of a common range, to represent the incident matrix at sample n, under cooling state c, and in zone i. Normalization is convenient. This technique can be used to improve the numerical accuracy of the solution and indirectly scale and normalize the thermal coefficient vector extracted in the solution. Representative normalization scales the temperature difference by 10°C, the power by 1kW, unless the power has already been normalized to a unit range based on the relay status, and the solar irradiance is scaled by 1kW / m². 2 Although alternative embodiments are feasible and may be preferred.

[0074] The main signal y at sample n, under cooling state c, in zone 103i i,c,n Defined as sampling period T S and the observed rate of temperature change The product of the main signal y i,c,n It can be calculated in various forms, including based on temperature T. i,n Previous temperature T i,n-m The first, second, and fourth discrete derivatives are defined by m in [1,4], and the main signal y is given by y. i,c,n equal to the thermal coefficient vector and incident vector The product of the main signal y i,c,n It can be represented by the following equation:

[0075]

[0076] The choice of the order of the discrete derivative is crucial because phase deviation and delay compensation must be considered, and the higher-order discrete derivative approximation is related to the rate of temperature change at sample n in zone 103i. and sampling period T SIt may not be compact enough. The derivative representation itself is noisy because the differentiation process effectively amplifies the high-frequency noise content in the signal related to errors and uncertainties in the observations. By applying an appropriate low-pass filter, the main signal y at sample n, under cooling state c, and in zone 103i... i,c,n Filtering is useful. A representative master signal filter is a second-order Butterworth IIR filter with a -3dB normalized frequency of 0.2250790799 and a nominal group delay of 2.0 samples. The delay of each filter must be addressed by applying delay compensation, which effectively aligns the components of the differential equation in time by inserting filters or delays into the appropriate signal path, minimizing the time alignment error associated with the asymmetric processing of a particular signal path, and ensuring that the differential equation is faithfully implemented. Delay compensation can take various forms, or the delay can be introduced into the incident vector associated with zone 103i at sample n, in cooling state c. Or the main signal y i,c,n This ensures that each path has an equivalent total delay.

[0077] In some embodiments, the thermal coefficient vector Solutions can be extracted from, among others, generalized linear inversion or adaptive filter banks, and expressed independently in various forms. Each solution has associated advantages, and the appropriate choice depends on the application. Solutions for generalized linear inversion and adaptive filter banks are described, although alternative embodiments are feasible and may be desirable.

[0078] Generalized linear inversion

[0079] In some cases, generalized linear inversion is a statistical solution that can be applied to extract the thermal coefficient vector at sample n, under cooling state c, and in zone i. The estimated value makes the main signal y at memory depth N corresponding to a set of independent observations. i,c,n The overall estimation error is minimized. The generalized linear inversion solution can be periodic, and the observations do not need to be ordered or continuous. Compared to alternative methods, this solution is computationally intensive but is typically performed less frequently. Storage requirements are proportional to the memory depth N and can be significant. Numerical accuracy depends heavily on the diversity of observations and the memory depth N. Changes in dynamic thermal systems may be indistinguishable when the time resolution is significantly smaller than the memory depth N.

[0080] Regarding the generalized linear inversion solution, the principal vector at sample n, under cooling state c, and in zone 103i is... Equal to the incident matrix and thermal coefficient vector The product of. For a nominal memory depth N of 7 days, the main vector and the incident matrix Having thermal coefficient vectors of dimension [J+K+2,1] Consistent dimensions [N,1] and [N,J+K+2], principal vector This can be represented by the following equation:

[0081]

[0082] Principal Vector and the incident matrix Each row is composed of independent incident vectors in zone i. and the main signal y i,c,m Observations are formed, where sample m is in [Nn,n], assuming a sampling period T. S Synchronous and continuous observations. The order of observations at memory depth N is independent and does not need to be continuous or sequential. Maximizing the diversity of observations is advantageous, although the chance of doing so may be constrained by limitations on extracting solutions in an environment employing completely passive observations. Solutions can be formed with fewer observations, although memory depth N and the diversity of observations largely determine the numerical accuracy of the solution.

[0083] The thermal coefficient vector at sample n, under cooling state c, and in zone i. It is the incident matrix and principal vector The solution to the defined linear system of equations, the vector of thermal coefficients It can be expressed by the following equation:

[0084]

[0085] The thermal coefficient vector at sample n, under cooling state c, and in zone i. It exhibits the same characteristics as the incident matrix. and principal vector Numerical accuracy is closely related to the diversity or independence of observations in a dataset. Diversity can be measured through the source matrix. The condition number is used to quantify the condition of a matrix, defined as the ratio of its largest to smallest eigenvalue. A well-conditioned matrix has a condition number close to 1 (unity), and if the matrix is ​​singular, the condition number is infinite, and a solution is impossible. Thermal coefficient vector The numerical accuracy depends on the diversity of the observations that form it, because the approximate accuracy is proportional to the logarithm of the condition number. This applies to extracting the thermal coefficient vector. Previously, explicit computation of the condition number significantly increased the computational complexity of the solution and may be impractical or impossible in many environments. A practical and advantageous alternative rank reduction method is defined, which dynamically tests and excludes incident matrices. Specific columns in the solution and the corresponding thermal coefficient vectors in the solution The elements must be sufficiently diverse to form a solution with adequate numerical accuracy; otherwise, the computation of a complete solution would be neither possible nor useful.

[0086] Adaptive filter bank

[0087] In some embodiments of this disclosure, filter bank 145, filter bank 935 may be implemented.

[0088] It is an iterative solution used to extract estimates of coefficients and state vectors, indirectly approximating the thermal coefficient vector at sample n, under cooling state c, in zone 103i. Conversely, this minimizes the reference signal v i,c,n Relative to the observed main signal y i,c,n The instantaneous estimation error.

[0089] The adaptive filter bank solution is periodic, and although not continuous, the observations are required to be ordered. Compared to alternative methods, this solution is computationally less expensive but is typically performed more frequently with each observation. Storage requirements are minimal and independent of the convergence rate or effective memory depth. Numerical accuracy depends heavily on the diversity of observations and the rate of adaptation μ. j Changes in dynamic thermal systems are typically discernible at a time resolution significantly lower than that of alternative methods (including generalized linear inversion). In some implementations, the adaptive filter bank 145 is constructed as a set of J filters, each consumed at sample n, filter index j, cooling state c, and zone i. j The incident signal x in i,c,j,n And generate a reference signal v i,c,j,n The effective transfer function of each filter is iteratively adapted to a function representing the estimation error.

[0090] This disclosure describes an architecture for an Infinite Impulse Response (IIR) filter based on a Direct II design, although alternative embodiments including DirectI, Lattice, and Parallel are discussed and may be preferred. A Finite Impulse Response (FIR) filter may be a viable alternative to an IIR filter, as it is architecturally equivalent, with the additional constraint being the recursive coefficient vector. It is static and consists of elements with values ​​equal to zero. It describes an adaptive filter with dynamic coefficients, although it may be advantageous to initially define one or more filters in an adaptive filter bank 145 with static coefficients based on prior knowledge, or to switch to a static solution after achieving specific performance.

[0091] In some embodiments, the adaptive filter 200 is one of a set of adaptive filters that can constitute an adaptive filter bank, such as... Figure 1 The system architecture 100 may implement the adaptive filter bank 145 as processing logic 143, which may include hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, etc.), software, firmware, or a combination thereof. In one embodiment, the adaptive filter bank 145 may be implemented as part of a service within the system's cloud computing environment 105 for comfort-based management. For example, the service associated with the adaptive filter bank 145 may be used to adapt a thermal model 140 to further help characterize heat transfer associated with comfort-based management of the venue 101. In other embodiments, the service associated with the adaptive filter bank 145 may be installed and used for comfort-based management by various other components of the system architecture 100, which may or may not be geographically distributed.

[0092] Figure 2 This is an example of an adaptive IIR filter 200 in the Direct II form according to an embodiment of the present disclosure. The adaptive filter 200 is implemented as an incident signal x in zone 103i at sample n, filter index j, cooling state c. i,c,j,n 230 recursive coefficient vectors and forward coefficient vector 240 The difference equation of the function is used to generate the reference signal 220V. i,c,j,n State vector 250s i,c,j, n is implemented to maintain the state of the recursive inner filter and is delayed by 260 Hz in discrete time units. -1 and filter order M j Limited. For example, at sample n, filter index j, cooling state c, and zone 103i, the recursive coefficient vector 230 Forward coefficient vector 240 and state vector 250s i,c,j,n The dimension is [M] j [1]. For ease of explanation, the zero-index recursion coefficient is 230. It has an implicit persistence value equal to zero and a zero-indexed state coefficient s. i,c,j,0,n It consumes no storage space and updates the state vector every 250 seconds. i,c,j,n It is useful in that context.

[0093] In some embodiments, stability considerations exist in the adaptive IIR filter 200 because the poles or roots of the denominator of the discrete transfer function derived from the z-transform of the difference equation are constrained to the interior of the unit circle to form a stable causal solution, and the zeros or roots of the numerator of the transfer function can be similarly constrained only if a minimum phase solution is desired. In many environments, root calculation may be computationally limited, although the process depends on the filter order M. j .

[0094] In some embodiments, this is achieved by using a stable recursive coefficient vector 230 in zone 103i at sample n equal to zero, filter index j, cooling state c. Assign values ​​determined by previous solutions or small random values, and for the forward coefficient vector 240. The adaptive IIR filter 200 is initialized by assigning values ​​determined by the previous solution or small random values.

[0095] By assuming continuous operation under these conditions to reduce edge effects, the state coefficient 250s will be obtained at sample n, coefficient index m, filter index j, cooling state c, and in zone 103i. i,c,j,m,0 Initialized to the initial incident signal 210x i,c,j,0 230 recursive coefficient vectors and forward coefficient vector 240 A function, where M is in [1, M]. j In the middle, s i,c,j,m,0 This can be represented by the following equation:

[0096]

[0097] The reference signal 220V is defined at sample n, filter index j, cooling state c, and zone 103i. i,c,j,n And modify the state vector 250 The adaptive IIR filter 200 is evaluated synchronously during the sampling period. In one embodiment, this is achieved by modifying the recursive coefficient vector 230. and forward coefficient vector 240 The adaptive filter 200 is selectively adapted after each evaluation. In this disclosure, the coefficients of the adaptive filter are described as being modified for least mean square (LMS) adaptation, although alternative embodiments including other variations of normalized least mean square (NLMS), recursive least square (RLS), and gradient descent adaptation are discussed and may be preferred.

[0098] At sample n, filter index j, cooling state c, reference signal 220V in zone 103i i,c,j,n Equal to the initial forward coefficient 240b i,c, The recursive coefficient vector of j,0,n scaling is 230. and forward coefficient vector 240 The transpose and state vector 250 The product of the incident signal 210x i,c,j,n and initial forward coefficient b i,c, The sum of the products of j, 0, and n, with a reference signal of 220V. i,c,j,n This can be represented by the following equation:

[0099]

[0100] Referring to the above equation, at sample n, with coefficient index 0, filter index j, and under cooling state c, the state coefficient 250s in zone 103i. i,c,j,0,n It is the recursive coefficient vector 230 The transpose and state coefficient vector 250 The product of the incident signal 210x i,c,j,n The sum, state coefficient 250s i,c,j,0,n This can be represented by the following equation:

[0101]

[0102] In some embodiments, at sample n+1, with coefficient index m, filter index j, cooling state c, and state coefficient 250s in zone 103i i,c,j,m,n+1 The state coefficient s was assigned at sample n with coefficient index m-1. i,c,j,m-1,n The value of m, where m is in [1, M]. j In the [section], the state coefficient is 250s. i,c,j,m,n+1 This can be represented by the following equation:

[0103]

[0104] In some embodiments, at sample n, in cooling state c, in zone 103i, the reference signal 220v i,c,n The reference signal 220V is in the filter index [0, J-1]. i,c,j,n The sum, reference signal 220V i,c,n This can be represented by the following equation:

[0105]

[0106] In some embodiments, the estimation error e at sample n, under cooling state c, in zone 103i i,c,n Defined as the main signal y i,c,n and reference signal v i,c,n The difference, and representing the error associated with e in estimating the main signal using a reference signal from an adaptive filter bank of a specific state. i,c,n This can be represented by the following equation:

[0107] e i，c，n =y i，c，n v i，c，n

[0108] The estimation error e at sample n, under cooling state c, in zone 103i i,c,n It is an instantaneous measure of the numerical precision of the solution, which is not adequately suited to the recursive coefficient vector 230. The forward coefficient vector of filter j in the index [0, J-1] is 240. In certain situations, this is necessary. The estimation error e i,c,n Visualized as M j A continuously differentiable surface of dimension 230, as a recursive coefficient vector and forward coefficient vector 240 The function may be useful.

[0109] The recursive coefficient vector at sample n+1, filter index j, cooling state c, and zone 103i Defined as the recursive coefficient vector 230 at sample index n With adaptation rate μ j Gradient estimation of recursive coefficients (Estimation error e) 2 i,c,n L1 or L2 represents the difference of the products of partial derivatives with respect to the recursive coefficients, where L1 or L2 is the recursive coefficient vector. This can be represented by the following equation:

[0110]

[0111] In the above equation, the adaptation rate μ j It is related to the convergence rate exponent and is proportional to the misadjustment or noise injected by the adaptive process, and must be appropriately chosen to ensure that the stable system has sufficient convergence and numerical accuracy relative to the thermal system.

[0112] refer to Figure 2 The recursive coefficient vector at sample n+1, filter index j, cooling state c, and zone 103i The recursive coefficient vector at sample n is 230. With adaptation rate μ j Estimation error e i,c,n The reference signal 220V with filter index j in [0, J-1] i,c,j,n The sum of the products of partial derivatives, the recursive coefficient vector This can be represented by the following equation:

[0113]

[0114] For ease of explanation, consider the state gradient vector at sample n, filter index j, cooling state c, and zone 103i. The recursive coefficient vector at sample index nm is 230. and state gradient vector Inner product and state coefficients The sum of all elements, where m is in [1, M]. j In the context of the state gradient vector... This can be represented by the following equation:

[0115]

[0116] Referring to the equation above, at sample n+1, filter index j, cooling state c, and in zone 103i, the recursive coefficient vector is 230. The recursive coefficient vector at sample index n is 230. With adaptation rate μ j Estimation error e i,c,n And regarding the recursive coefficient vector 230 Forward coefficient vector 240 and state gradient vector The sum of the products of the expressions. This can be represented by the following equation:

[0117]

[0118] In some embodiments, at sample n+1, filter index j, cooling state c, and in zone 103i, the forward coefficient vector 240 Defined as the forward coefficient vector 240 at sample n With adaptation rate μ j Gradient estimation of recursive coefficients (Estimation error e) 2 i,c,n L1 or L2 represents the difference of the products of the partial derivatives with respect to the forward coefficients. The equivalent expression is the forward coefficient vector at sample n, which is 240. With adaptation rate μ jEstimation error e i,c,n The reference signal 220v with filter index j in [0, J–1] i,c,j,n The sum of the products of the partial derivatives, the forward coefficient vector 240 This can be represented by the following equation:

[0119]

[0120] At sample n+1, with coefficient index zero, filter index j, and cooling state c, the forward coefficient b in zone 103i i,c,j,0,n+1 The forward coefficient b at sample n i,c,j,0,n With adaptation rate μ j Estimation error e i,c,n The reference signal 220v with filter index j in [0, J–1] i,c,j,n The sum of the products of the partial derivatives, the forward coefficient b i,c,j,0,n+1 This can be represented by the following equation:

[0121]

[0122] In other embodiments, at sample n+1, filter index j, cooling state c, and in zone 103i, the forward coefficient vector 240 Defined as the forward coefficient vector 240 at sample n With adaptation rate μ j Estimation error e i,c,n and state vector s i,c,j,n The sum of the products, forward coefficient vector 240 This can be represented by the following equation:

[0123]

[0124] Turn Figure 3 This illustrates an example architecture of an adaptive filter bank 300. In some embodiments, the adaptive filter bank 300 can be coupled with... Figure 1 The adaptive filter bank 145 is the same. Adaptive filter bank 300 can be constructed as adaptive filters 311-315 (such as...). Figure 2 A set of adaptive IIR filters (200).

[0125] The set of adaptive filters 311-315 is defined as processing heat transfer related to passive heat transfer, active heat transfer, assisted heat transfer, solar heat transfer, and unobserved heat transfer at sample n, under cooling state c, and zone i. j The incident vector 310 And the incident signal 210 with filter index j is x i,c,j,n Reference signal 320Vi,c,n Applying an appropriate 340Hz delay -L (Applying delay compensation, which minimizes timing alignment error by inserting a delay of L cycles into the reference signal path) followed by the main signal 330y i,c,n The total estimated value. Estimation error: 370e i,c,n Defined as different adaptation rates μ j Selectively adapt the recursive coefficient vector and forward coefficient vector

[0126] Each adaptive filter 311-315 of the adaptive filter bank 300 consumes the incident vector 310 at sample n, under cooling state c, in zone i. To generate a reference signal of 320V i,c,n The incident signal 210 with filter index j is x. i,c,j,n For example, the incident signal 210x i,c,j,n Data can be received from the thermal model 140 and includes thermal coefficient data associated with at least one of the following: active heat transfer, passive heat transfer, solar irradiation heat transfer, and unobserved heat transfer in zone 103i. At this point, the adaptive filter bank 300 defines thermal coefficients associated with the corresponding heat transfer properties, allowing them to be evaluated individually and modified accordingly.

[0127] Each adaptive filter 311-315 of the adaptive filter bank 300 includes a corresponding filter section 311p-315p associated with a transfer function. The effective transfer function of the filter section 311p-315p is iteratively adapted to be a function of an estimation error 370, which is represented by the estimation of the main signal 340y using a reference signal 230 from the adaptive filter bank 300 in a particular state. i,c,n The associated error. Each of the filter sections 311p-315p is defined as processing the incident vector 310 at sample n, under cooling state c, in zone i, related to passive heat transfer, active heat transfer, assisted heat transfer, solar heat transfer, and unobserved heat transfer. To generate a reference signal of 320V i,c,n The incident signal 210 with filter index j is x. i,c,j,n For example, the transfer function can represent the corresponding incident signal 210x. i,c,j,n As input, the recursion coefficient vector Forward coefficient vector and self-adaptation rate μ j Return it as output.

[0128] In some embodiments, the reference signal is 320V.i,c,n Applying an appropriate delay z L (For example, the main signal 330y after applying delay compensation 340) i,c,n The total estimate (generated by each of the adaptive filters 311-315). In an illustrative example, delay compensation 340 minimizes the time alignment error by inserting a delay of L cycles (e.g., nominally [0,2]) into the reference signal path. The delay of each of the adaptive filters 311-315 must be addressed by applying delay compensation 340, which effectively aligns the components of the difference equation in time by inserting filters or delays into the appropriate signal path. In this way, any time alignment error associated with the asymmetric processing of a particular signal path will be minimized, and the difference equation will be faithfully implemented. Delay compensation 340 can take many forms, for example, by introducing a delay into the incident vector 310 associated with sample n, cooling state c, and zone i. Or main signal 330y i,c,n This ensures that each path has an equivalent total delay.

[0129] In some embodiments, the adaptive filters 311-315 of the adaptive filter bank 300 can be based on the estimation error 370e. i,c,n To adapt. For example, the estimation error is 370e. i,c,n Defined by adaptive filter logic 132, to achieve various adaptation rates μ j Selectively adapt the recursive coefficient vector and forward coefficient vector For example, this is achieved by using adaptive filter logic 132 to solve a system of linear equations using adaptive filters 311-315 (as referenced). Figure 2 The adaptive filter 200 is used to implement this, wherein each equation is formed by independent observations of the incident signal 210 associated with the incident vector 310 and the thermal coefficient vector. Formation. In some embodiments, adaptive filters 311-315 are adaptively updated until a solution associated with zone 103 is satisfied. For example, this solution may indicate numerical convergence between reference signal 320 and main signal 330 associated with zone 103, which helps minimize estimation errors and ensures the numerical accuracy of the comfort-based managed solution for zone 103.

[0130] Figure 4 This is an example diagram 400 illustrating an adaptive filter solution for a thermal model according to an embodiment of the present disclosure. Figure 4 In the sample n, under cooling state c, in zone i, the recursive coefficient vector Forward coefficient vector A commercial establishment with 12 zones (e.g., Figure 1 Location 101 is shown, illustrating a specific zone (e.g., zone 103) comprising five HVAC units (e.g., HVAC unit 104) in UTC-5, which have been actively heated over several consecutive days. Recursive and forward coefficient vectors are associated with a thermal model (e.g., thermal model 140) implementing an adaptive filter bank solution, where the adaptive filter bank 145 is initially reset to nominal conditions and iteratively adapted over intervals of continuous operation.

[0131] Figure 400 illustrates the process based on an adaptive filter bank (whose processing of the incident vector). The estimated reference signal v i,c,n and the main signal y, which is calculated as a discrete rate of temperature change. i,c,n A sequence of (inherent noise signals). Estimation error e i,c,n Defined as the difference between the master signal and the reference signal, it is used to adapt the recursive coefficient vector and the forward coefficient vector to continuously and opportunistically improve the reference signal estimate. From January 5th to January 7th, 2019, significant changes were observed in the components of the forward coefficient vector associated with passive heat transfer, active heat transfer, and solar heat transfer, corresponding to significant changes in air temperature and solar irradiance outside the thermal system boundary, and corresponding to increased power consumption in this zone. The increased diversity of the incident vectors reduced the estimation error and improved the accuracy of the reference signal estimate relative to the master signal, indicating convergence of the thermal model.

[0132] estimate

[0133] In an illustrative example, the utility of finding a thermal model (e.g., thermal model 140) in the estimation of temperature and power, and its use for extracting and validating thermal coefficient vectors. The rationale behind this is that it facilitates applications including optimal start-up, optimal shutdown, savings estimation, diagnostics, demand response, and virtual power plant capacity. For example, the temperature estimate T at sample n, in zone i. i,n The previous temperature T at sample n-1 i,n-1 The reference signal v in the solution of the adaptive filter bank i,c,n The sum, temperature estimate T i,n This can be represented by the following equation:

[0134]

[0135] Referring to the equation above, temperature estimation is useful for iteratively estimating temperature series, requiring only previously estimated or observed temperatures T. i,n-1 Expected power consumption vector P i,c,n The sequence and the weather models (such as Figure 1The weather model 160 provides predicted temperature and solar irradiance estimates. In some embodiments, the temperature estimates can be used to predict possible behaviors associated with various scenarios that differ in power consumption, availability, or cost.

[0136] Assuming the power consumption in associated cell k is zero, the estimated power P at sample n, cell index k, cooling state c, and zone 103i is... i,c,k,n It is the main signal y i,c,n With the incident vector and thermal coefficient vector The difference of the product and the active thermal coefficient with index m. The ratio, power estimate P i,c,k,n It can be represented by the following equation.

[0137]

[0138] The adaptive filter used in the absence of an adaptive filter bank solution (such as...) Figure 3 Under the specific definition of the adaptive filter (311-315), for the power estimate P at sample n, cell index k, cooling state c, and zone i i,c,k,n A specific solution is not explicitly available. In some embodiments, the process for defining a solution for the power estimate begins with a reference signal v indexed as j to the filter associated with cell k. i,c,j,n With estimation error e i,c,n Definitions that are related.

[0139] In the adaptive filter bank solution, assuming the power consumption in the correlated unit k is zero, at sample n, filter index j, cooling state c, and the reference signal v in zone i... i,c,j,n It is the main signal y i,c,n and reference signal v i,c,n The difference, or equivalently the estimation error e i,c,n Reference signal v i,c, j, n are represented by the following equations:

[0140]

[0141] When the solution of the adaptive filter bank is determined by the filter order M j When an adaptive IIR filter with a value equal to zero is used, the recursive coefficient vector in zone i is given at sample n, filter index j, cooling state c. It is a unit-length vector with zero values. The forward coefficient vector of filter index j. It has a forward coefficient b i,c,j,0,n The adaptive value is a unit-length vector, with forward coefficients b. i,c,j,0,nIn representation, it is equivalent to the thermal coefficient with coefficient index m. The power estimate P in this system i,c,k,n This is directly derived from the power estimation equation above. In some embodiments, the power estimation is useful for iteratively estimating a sequence of power consumption, requiring only a deterministic temperature estimate T. i,n The sequence of values, along with predicted temperature and solar irradiance estimates provided by weather models, is used. Power estimates can also be used to predict possible behaviors associated with alternative temperature control scenarios.

[0142] If passive heat transfer between zones is selectively considered in a thermal system such as site 101, the estimation of temperature and power must typically be performed recursively through interaction with other thermal models, since passive heat transfer from each zone j that may share a boundary with zone 103i will affect the temperature and power consumption of zone 103i. Alternatively, in many thermal systems, the following assumption may be reasonable or even necessary: ​​that passive heat transfer between zones is negligible with respect to temperature and power estimation. For example, in thermal systems where the temperature is similar across zones sharing a boundary, it may be necessary to ignore passive heat transfer between zones because the temperature difference between zones may be insufficient to extract an accurate solution using the techniques disclosed herein.

[0143] Example Flowchart

[0144] Figure 5 This is a flowchart of a method 500 for modeling an adaptive filter bank for a thermal system according to embodiments of the present disclosure. Method 500 can be executed by processing logic that may include hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, etc.), software, firmware, or a combination thereof. In one embodiment, adaptive filter logic 132 executed by a processor of system architecture 100 can execute method 500. Although shown in a specific order or sequence, the order of processes can be modified unless otherwise stated. Therefore, the illustrated embodiments should be understood as examples only, and the illustrated processes can be executed in different orders, and some processes can be executed in parallel. Furthermore, one or more processes may be omitted in various embodiments. Therefore, not all processes are necessary in every embodiment. Other processing flows are also possible.

[0145] Method 500 begins at block 510, where an adaptive filter bank and associated thermal coefficient data are identified to characterize the heat transfer in the volume of the thermal system. For example, an adaptive filter bank and associated thermal coefficients are identified to define the architecture, order, and delay of component filters to characterize the heat transfer in the volume of the thermal system. In block 520, reference signal data indicating an estimate of the rate of temperature change in the volume are generated, based at least on the adaptive filter bank. In block 530, master signal data is received, based at least on the observed rate of temperature change in the volume. In block 540, the estimation error of the reference signal data relative to the master signal data associated with the volume is determined. In block 550, the thermal coefficient data are modified, given the estimation error, to satisfy the solution associated with the volume. For example, the thermal coefficients are modified to reduce the estimation error and improve the approximation of the rate of temperature change in the reference signal.

[0146] Figure 6 This is a flowchart of another method 600 for modeling an adaptive filter for a thermal system according to embodiments of the present disclosure. Method 600 can be executed by processing logic including hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, etc.), software, firmware, or a combination thereof. In one embodiment, adaptive filter logic 132 executed by a processor of system architecture 100 can execute method 600. Although shown in a specific order or sequence, the order of processes can be modified unless otherwise stated. Therefore, the illustrated embodiments should be understood as examples only, and the illustrated processes can be executed in different orders, and some processes can be executed in parallel. Furthermore, one or more processes may be omitted in various embodiments. Therefore, not all processes are necessary in every embodiment. Other processing flows are also possible.

[0147] Method 600 begins at block 610, where an incident signal including thermal coefficients associated with the volume of the thermal system is received at an adaptive filter bank. In block 620, a reference signal indicating an estimate of the rate of temperature change in the volume is generated, based at least on the adaptive filter bank. In block 630, an estimation error of the reference signal relative to a master signal associated with the volume is determined. In block 640, the adaptive filter bank is updated based on the estimation error to satisfy an approximation of the rate of temperature change in the volume relative to the reference signal.

[0148] diagnosis

[0149] Turn Figure 7 This illustrates the use of an adaptive filter bank (e.g., adaptive filter bank 145) to detect, for example,... Figure 1A flowchart of a method 700 for diagnosing events in the thermal system of location 101 is provided. For example, method 700 defines diagnostic events in diagnostic model 701 to detect state changes in a representation of the thermal system such as location 101 (e.g., adaptive thermal model 702) over various time intervals, which can indicate abnormal operation. Method 700 can be executed by processing logic including hardware (e.g., circuitry, dedicated logic, programmable logic, microcode, etc.), software, firmware, or combinations thereof. In one embodiment, adaptive filter bank logic 155 executed by the processor of system architecture 100 can execute method 700. Although shown in a specific order or sequence, the order of processes can be modified unless otherwise stated. Therefore, the illustrated embodiments should be understood as examples only, and the illustrated processes can be executed in different orders, and some processes can be executed in parallel. Furthermore, one or more processes may be omitted in various embodiments. Therefore, not all processes are necessary in every embodiment. Other processing flows are also possible.

[0150] In this example, the diagnostic provides information related to the conditions of a thermal system (such as location 101). The diagnostic can be based on the analysis of characteristics of the dynamic representation of thermal system 101. At this point, a diagnostic event can be described as instantaneous or continuous, and defined based on continuous monitoring, local analysis, or overall analysis of independent thermal systems with similar design, construction, or geographical location. A diagnostic event can indicate the detection of a change in observed or estimated characteristics of thermal system 101, which may be related to defective or abnormal operation. Without independent means to examine the thermal system, the correlation between the detection of a diagnostic event and the presence or cause of abnormal operation cannot be confirmed. In this context, the objective of diagnostic method 700 is to determine whether the thermal system is operating in a manner inconsistent with normal or expected behavior.

[0151] In some embodiments, diagnostic events may be generated from the evaluation of observations and estimates over various time intervals supporting, for example, transient or persistent classifications. These classifications are not exclusive or unique, as transient diagnostic events may eventually become persistent, and in many environments, cyclic or quasi-cyclic diagnostic events are reliable. Diagnostic evaluations over various time intervals have practical utility because anomalous operation may evolve over different time intervals (durations ranging from seconds to months). In some examples, operating modes of thermal equipment 150 of thermal system 101 that could lead to the detection of transient diagnostic events include pump, valve, or relay failures in cyclic heating thermal systems, or relay failures or rapid refrigerant leaks in gas-based boilers, heat pumps, or air conditioning systems. In other examples, operating modes of thermal equipment 150 that could lead to the detection of persistent diagnostic events include slow boiler, pipe, or radiator leaks in cyclic systems, or sludge buildup due to oxidation, or slow refrigerant leaks, damaged heat exchangers, or contaminated pipes and filters in gas-based systems. Transient diagnostic events may exhibit periodic, quasi-periodic, or persistent behavior, which could be caused by conditions including slow leaks in the circulation system, as the occupant may become aware of the defect and choose to occasionally change the water and restore normal operation, or allow the defect to evolve into persistent and increasingly abnormal operation.

[0152] refer to Figure 7 The diagnostic model 701 detects changes in the state of a thermal system that may indicate abnormal operation. A dynamic representation of the state of a thermal system (including physical or abstract thermal models) is at least sufficient and necessary to facilitate diagnosis.

[0153] Representative thermal models suitable for diagnostic applications include physical thermal models described by the following heat transfer difference equations:

[0154]

[0155] The heat transfer difference equation above defines a physical relationship that will be the rate of temperature change at sample n in zone 103i. By relating the observed and estimated values, a convenient vector of thermal coefficients under cooling state c can be derived. In this context, thermal coefficient vectors are dynamic physical representations of the state of a specific zone in a thermal system, as these components conveniently define the normalized rates of passive heat transfer, active heat transfer, solar heat transfer, and unobserved heat transfer, as disclosed herein.

[0156] In typical locations (such as location 101), the passive heat transfer coefficient at sample n, zone 103i It is relatively stable because it relates to the geometry and material properties associated with a specific zone relative to the temperature difference. At this point, in systems including heat pumps, multi-mode systems, and systems with variable compressor speeds, the active heat transfer coefficient at cooling state c... With the efficiency η of the HVAC unit i,c,n Proportional and dynamic based on changing environmental conditions or design, or it may change due to abnormal operation. Solar thermal transfer coefficient. It is relatively stable because it relates to the geometry and material properties associated with a specific region of the incident solar energy. The unobserved heat transfer coefficient... It is relatively stable because it typically aggregates over a sufficiently long period of time to satisfy this simplifying assumption.

[0157] In some embodiments, continuous monitoring can opportunistically or periodically evaluate solutions used for observing and controlling the thermal system to ensure that installation, configuration, and operation are consistent with expected behavior. Diagnostic events can be defined if communication with one or more thermal devices or cloud-based resources (including threads, worker roles, remote compute nodes, virtual machines, caches, tables, dictionaries, databases, data lakes, hubs, or services) is inaccessible or exhibits anomalous operation (including excessive latency or data corruption). Continuous monitoring can be implemented as a supervisory agent with sufficient visibility and access to the components of the solution that are crucial for ensuring that observations and estimates are defined and preserved with sufficient quality, quantity, and density for both local and aggregate analysis.

[0158] Back Figure 7 Method 700 begins with blocks 710-730, where, for a specific location 101, it is constructed, initialized, and adapted (e.g., using) based on each zone (e.g., zone 103). Figure 1 The adaptive filter bank 145) is a thermal model such as thermal model 140. In some embodiments, a representative thermal model suitable for diagnostic applications includes a physical thermal model described by the heat transfer difference equations described herein. Each adaptive process in each zone produces a vector of thermal coefficients in zone 103i at sample n, under cooling state c. The dynamic estimates are used to describe and estimate the behavior of the thermal system 101. In box 750, thermal model data, including thermal coefficient vectors, is fed into the diagnostic model 701.

[0159] In box 740, one or more filters are applied to the thermal coefficient data based on the sampling rate. For example, diagnostic model 701 is used to selectively apply a set of linear or nonlinear filters to the thermal coefficient vector at sample n, under cooling state c, in zone 103i. Filtering is performed. This can further include morphological operators to reduce noise and volatility while preserving features. In box 750, the sampling rate can be adapted and selectively combined with the filtering operation to reduce aliasing and implement rate adaptation, resulting in a sampling period T. S (e.g., nominal value 600-900 seconds) thermal coefficient vector This is converted to an effective sampling period (e.g., a nominal 24 hours). In some embodiments, a representative filter operation includes a second-order Butterworth IIR filter with a -3dB normalized frequency f. c Less than or equal to Its nominal group delay is Each sample can be effectively integrated with a decimation filter to achieve an efficient sampling period.

[0160] In box 760, in response to the application of a filter, one or more estimated thermal coefficient thresholds are generated. For example, diagnostic model 701 is used to estimate the thermal coefficient thresholds at sample n, under cooling state c, in zone 103i over consecutive intervals having a memory depth N or a window length. This is achieved by evaluating the thermal coefficient vector after applying filters and rate adaptive operations. This is achieved through a sequence to determine whether a trend or pattern of behavior is observed that is sufficient to support the definition of a diagnostic event. Thermal coefficient threshold. It is defined by applying linear or nonlinear operations to the thermal coefficients constrained to a specific window, where each successive window shares [0, N-1] samples with the previous window. This is to determine whether the thermal model 140 has shown convergence to a solution with minimum estimation error on a specific window, and converges with sufficient confidence to at least the boundary samples of the window to prove the threshold of the thermal coefficients at sample n, under cooling state c, in zone 103i. The estimation. If the thermal model 140 does not converge sufficiently over the window, it may be infeasible to identify diagnostic events within the corresponding time interval.

[0161] In box 770, the thermal coefficient data is evaluated to determine if any data meets the corresponding estimated thermal coefficient threshold. In an illustrative example, diagnostic model 710 is used to evaluate the thermal coefficient vector. The active heat transfer coefficient at sample n, under cooling state c, in zone 103i thermal coefficient vector A specific window was defined to determine whether the coefficient fit exhibited sufficient change to indicate that the normalization efficiency threshold ε was exceeded. a,c Efficiency η (for example, a nominal value of 0.025% decrease per day)i,c,n Effective loss. Efficiency threshold ε a,c It significantly affects the sensitivity of diagnostic event detection and can be empirically determined, at least in part, based on the geometry, geography, materials, construction, and HVAC unit type of typical locations in the region.

[0162] For the probability density function P at sample n, under cooling state c, in zone 103i a,i,c,n With thermal coefficient vector Active heat transfer coefficient The estimation of memory depth N allows for the extraction of the associated probability density function P. a,i,c,n The first and third quartile estimates are used as thresholds for thermal coefficients. For the thermal coefficient vector that defines a specific window The active heat transfer coefficient at sample n, under cooling state c, in zone 103i An evaluation is conducted to determine the active heat transfer coefficient. Is it less than or equal to? and active heat transfer coefficient Is it greater than or equal to? Furthermore, asserting the active heat transfer coefficient Also greater than or equal to It could be useful to increase the efficiency of observations η. i,c,n This reduces the confidence level in forming a pattern or trend consistent with continuous losses. Due to efficiency η i,c,n The rate of reduction depends on the operating mode that may lead to the detection of transient or continuous diagnostic events, and therefore has considerable utility in the evaluation of diagnostic models at various memory depths.

[0163] Additional selectivity restrictions can be applied to thermal coefficient vectors with a memory depth of N. The evaluation of the sequence includes determining the passive heat transfer coefficient w. p,i,c,n Or solar thermal transfer coefficient w s,i,c,n To statistically approximate the expected value, or to determine the unobserved heat transfer coefficient |w u,i,c,n It can be considered insignificant compared to other means of heat transfer.

[0164] In box 780, alert information indicating diagnostic events associated with the thermal system can be generated based on an evaluation of a threshold. For example, if the thermal coefficient w is found... i,c,nA diagnostic event can be identified if any of the defined constraints are met and abnormal operation is exhibited within a consecutive interval of memory depth N. Alarm information can indicate, for example, transient diagnostic event detection, continuous diagnostic event detection, or other types of detected diagnostic events. Transient diagnostic event detection may include pump, valve, or relay failures in a circulating heating system, or relay failures or rapid refrigerant leaks in a gas-based boiler, heat pump, or air conditioning system. Continuous diagnostic event detection may include slow boiler, pipe, or radiator leaks in a circulating system, or sludge buildup due to oxidation, or slow refrigerant leaks, damaged heat exchangers, or contaminated pipes and filters in a gas-based system.

[0165] Diagnostic event detection example

[0166] Figures 8A-8D Example graphs 800, 820, 840 and 860 illustrate diagnostic event data according to embodiments of the present disclosure.

[0167] exist Figure 8A In the figure, graph 800 uses plotted point 802 to show the active heat transfer coefficient w at sample n, under cooling state c, in zone i. a,i,c,n And the thermal model temperature estimation error e at a site with an observable zone in UTC, UK, during a 45-day active heating period of the circulating system. i,n A diagnostic event was detected at plotted point 808 on March 26, 2017, using the techniques disclosed herein. In deep shadow 804, transient local classification was based on memory depth N (equal to 3 days). It should be noted that in light shadow 806, efficiency η was also observed over a time interval twice the memory depth N (equal to 6 days). i,c,n The implicit, continuous decreasing trend is observed. In this analysis, a conservative normalized efficiency threshold ε of 0.05 was applied. a,c In a collection of 1872 sites across the UK during a two-month period of cyclical heating, each with an observable zone, at least one transient localized diagnostic event was detected at 38 sites. This was effectively achieved through diagnostic analysis (e.g., utilizing...). Figure 7 According to the method (700), 2.03% of the locations in the set were identified as exhibiting transient anomalous operations.

[0168] exist Figure 8B In the figure, graph 820 uses plotted point 822 to show the active heat transfer coefficient w at sample n, under cooling state c, in zone i. a,i,c,n And the thermal model temperature estimation error e at a site with an observable zone in UTC, UK, during a 45-day active heating period of the circulating system. i,nUsing the techniques disclosed herein, sequences of diagnostic events were detected at plotting points 824 on April 11, 2017, 825 on April 12, 2017, 826 on April 13, 2017, and 828 on April 14, 2017. In deep shadow 829, persistent local classification was based on memory depth N (equal to 14 days). It should be noted that in shallow shadow 827, efficiency η was also observed over time intervals twice the memory depth N (equal to 28 days). i,c,n The implicit, continuous decreasing trend is observed. In this analysis, a conservative normalized efficiency threshold ε of 0.025 was applied. a,c In a collection of 1,872 sites across the UK during a two-month cyclical heating period, with one observable zone at each site, at least one persistent local diagnostic event was detected at 96 sites. Effectively, through diagnostic analysis, 5.13% of the sites in the collection were identified as exhibiting persistent anomalous operation.

[0169] In some embodiments, diagnostic events 824, 825, 826, and 828 can be uniquely identified or verified with increased confidence through aggregate analysis. Aggregate analysis may consider trends or statistical indicators among the site sets and may select sites to include based on factors including similar age, geometry, geographic location, construction, weather, HVAC unit type, energy consumption, population density, or occupant demographics. In some embodiments, probability density functions may be estimated for attributes of interest (including thermal coefficients) of the site set, and persistent clustering diagnostic events can be independently identified in sites exhibiting anomalous behavior relative to statistical specifications. For example, sites with certain zones may define persistent clustering diagnostic events indicating relatively poor insulation or materials exceeding a reporting threshold for anomalous operation, in which the associated thermal model has converged to a passive heat transfer coefficient w at sample n, zone i, above a threshold (nominal value being the 90th percentile of the probability distribution). p,i,n Or solar thermal transfer coefficient w s,i,n The solution. Similarly, the active heat transfer coefficient w below the threshold (the nominal value is the 10th percentile of the probability distribution). a,i,c,n The HVAC unit efficiency η, which indicates the relative poor performance, can be defined. i,c,n The continuous clustering of diagnostic events.

[0170] exist Figure 8C In the study, for a collection of 1872 sites in the UK's UTC during two consecutive months of active heating in an independent circulation system, each site having an observable zone, points 842, 844, and 846 in plot 840 illustrate the conditions at sample n, where {w} p,n ,w a,n ,e nA series of probability density functions P of k in} k,n In this example, the quartiles are defined by vertical line 845, which in turn defines the middle 50th percentile of the corresponding probability distribution within the deep shaded area 843. For example, the active heat transfer coefficient w... a,n The probability distribution has a 25th percentile of 0.14, a 50th percentile of 0.19, and a 75th percentile of 0.275. In some embodiments, the aggregate analysis can also estimate the rate of change of the metric of interest and normalize the associated metric calculated in the local analysis. This can be used... Figure 7 Method 700 employs techniques to improve the confidence of diagnostic event detection and address errors that may not be directly observable in a single location. Instead, when formed over a set of locations, they are likely to be easily detected and statistically significant.

[0171] exist Figure 8D In the study, for a collection of 1872 sites in the UTC of the UK during two consecutive months of active heating in an independent circulation system, with each site having an observable zone, plot 860 uses plotted points 862 to show the relative change Δw of the active heat transfer coefficient at sample n. a,n probability density function P ΔWa,,n In this example, the quartile is defined by vertical line 863, which in turn defines the middle 50th percentile of the corresponding probability distribution within the deep shaded area 864. The relative change Δw in the active heat transfer coefficient. a,n It has probability distributions with the 25th percentile at -0.05, the 50th percentile at -0.02, and the 75th percentile at 0.005. Here, the relative change η of the active heat transfer coefficient among the site sets is estimated. i,c,n The expectation, and normalized to Figure 8D The results obtained from the local analysis are shown in the curve 860. The total analysis can improve the confidence of local diagnostic event detection.

[0172] In some embodiments, the results may be normalized to account for potential sources of error, which may include predicted weather estimates, thermal instrument observations, or potentially significant unobserved heat transfer due to the presence of additional controlled and unobserved zones. In this example, the relative change Δw of the active heat transfer coefficient for a typical or central location during a consecutive two-month period of interest. a,n It has an expected value of -0.02. Therefore, the normalized efficiency threshold ε can be adjusted to 0.025. a,c To account for the total expected efficiency loss (equal to 0.0253 in this example). Alternatively, the thermal coefficient... or relative change in thermal coefficient It can be normalized to account for any observed total change.

[0173] Controller device

[0174] Figure 9 This is a block diagram illustrating a machine 900 and a control device (hereinafter referred to as the "IVTMC controller") 901 in the form of internal volumetric thermal modeling. The control device 901 can learn from a set of weather estimates and thermal properties to predict and control energy consumption, power consumption, and / or temperature associated with one or more volumes or enclosed environments of the thermal system. In some embodiments, the machine 900 may represent a computer system within which a set of instructions can be executed to cause the machine 900 to perform any one or more methods discussed herein. In various illustrative examples, the IVTMC controller 901 may correspond to... Figure 1 The computing node 107 of system 100. For example, the IVTMC controller 901 may include instructions for implementing adaptive filter banks 145 for modeling thermal systems, such as Figure 1 Location 101.

[0175] The IVTMC controller 901 may be computer-based, which may include, but is not limited to, the following components: a computer system 902 connected to memory 929. In one embodiment, the IVTMC controller 901 may be connected to and / or communicate with entities such as, but not limited to: one or more users 933a from user input device 911; peripheral devices 912; optional encryption processor devices 926; and / or communication networks 913, such as local area networks (LANs), piconet networks, wide area networks (WSNs), wireless networks (WLANs), intranets, extranets, the Internet, etc.

[0176] The computer system 902 may include a clock 930, a central processing unit (“(multiple) CPUs” and / or “(multiple) processors”) (these terms are used interchangeably throughout the disclosure unless otherwise stated) 903, memory 929 (e.g., read-only memory (ROM) 906, random access memory (RAM) 905, etc.) and / or interface bus 907, and most commonly, though not strictly necessary, all interconnected and / or communicating via one or more motherboard system buses 904 having conductive and / or additional transport circuit paths through which instructions (e.g., binary encoded signals) can be transmitted to enable communication, operation, storage, etc. The computer system may be connected to a power supply 928; alternatively, the power supply may be internal. Optionally, an encryption processor 926 and / or a transceiver (e.g., an IC) 974 may be connected to the system bus. In another embodiment, the encryption processor and / or transceiver may be connected via interface bus I / O as internal and / or external peripheral devices 912. Conversely, the transceiver can be connected to (multiple) antennas 975 to enable wireless transmission and reception of various communication and / or sensor protocols.

[0177] Power supply 928 can be any standard form of power supply for small electronic circuit board devices, such as alkaline batteries, lithium-ion batteries, lithium-polymer batteries, nickel-cadmium batteries, solar cells, etc. Other types of AC or DC power supplies can also be used. In the case of solar cells, in one embodiment, the housing provides a hole through which the solar cell can capture photon energy. Power supply 928 is connected to at least one of the subsequent components of the IVTMC interconnection, thereby providing current to all subsequent components. In an alternative embodiment, external power supply 928 is provided through a connection between I / O interfaces 908. For example, USB and / or IEEE 1394 connections carry data and power between these connections and are therefore suitable power supplies.

[0178] Multiple interface buses 907 can accept, connect to, and / or communicate with multiple interface adapters, typically but not necessarily in the form of adapter cards, such as, but not limited to, input / output (I / O) interfaces 908, storage interfaces 909, network interfaces 910, etc. Optionally, an encryption processor interface 927 can similarly be connected to the interface bus. The interface bus provides communication between interface adapters and with other components of the computer system.

[0179] Storage interface 909 can accept, communicate with, and / or connect to multiple storage devices, such as, but not limited to, storage device 914, removable disk devices, etc. The storage interface can employ connection protocols such as, but not limited to, Ultra Serial Advanced Technology Attachment (Packet Interface), Enhanced Integrated Drive Electronics (E)IDE, IEEE 1394, Fibre Channel, Small Computer Systems Interface (SCSI), Universal Serial Bus (USB), etc.

[0180] Network interface 910 can accept, communicate with, and / or connect to communication network 913. Through communication network 913, user 933a can access IVTMC controller 901 via remote client 933b (e.g., a computer with a web browser). The network interface can employ connection protocols such as, but not limited to: direct connection, Ethernet (thick cable, thin cable, twisted pair 10 / 100 / 1000Base T, etc.), token ring, wireless connection (such as IEEE 802.11ax), etc. If processing requirements specify greater speed and / or capacity, a distributed network controller (e.g., a distributed IVTMC) architecture can be similarly used for pooling, load balancing, and / or additionally increasing the communication bandwidth required by IVTMC controller 901. Furthermore, multiple network interfaces 910 can be used to interface with various types of communication networks 913. For example, multiple network interfaces can be employed to allow communication via broadcast, multicast, and / or unicast networks.

[0181] Input / output interface (I / O) 908 can accept, communicate with, and / or connect to user input device 911, peripheral device 912, encryption processor device 926, etc. The I / O can employ connection protocols such as, but not limited to: audio: analog, digital, mono, RCA, stereo, etc. User input device 911 is typically a peripheral device and may include: card reader, dongle, fingerprint reader, glove, graphics tablet, joystick, keyboard, microphone, (multiple) mice, remote control, retina reader, touchscreen (e.g., capacitive, resistive, etc.), trackball, tracking board, sensors (e.g., accelerometer, ambient light sensor, GPS, gyroscope, proximity sensor, etc.), probes, etc.

[0182] Typically, any mechanization and / or implementation that allows the processor to influence the storage and / or retrieval of information is considered memory 929. However, memory is a replaceable technology and resource, and therefore, any number of memory implementations can be employed in place of or coordinated with each other. It should be understood that the IVTMC controller 901 and / or computer systemization can employ various forms of memory 929. For example, a computer systemization can be configured where the operation of on-chip CPU memory (e.g., registers), RAM, ROM, and any other storage devices is provided by a punch tape or punch card mechanism; however, such an embodiment would result in an extremely slow operating rate. In a typical configuration, memory 929 will include ROM 906, RAM 905, and storage device 914. Storage device 914 can be any conventional computer system storage device. Storage devices may include computer-readable storage media 935; magnetic drums; (fixed and / or removable) disk drives; magneto-optical drives; optical drives (i.e., Blu-ray, CD-ROM / RAM / recordable (R) / rewritable (RW), DVD R / RW, HDDVD R / RW, etc.); device arrays (e.g., redundant arrays of independent disks (RAID)); solid-state storage devices (USB storage devices, solid-state drives (SSDs), etc.); other processor-readable storage media; and / or other similar devices. Therefore, computer systems typically require and utilize memory.

[0183] Although the non-transitory computer-readable storage medium 935 is shown as a single medium in the illustrative example, the term "computer-readable storage medium" should include a single medium or multiple media (e.g., a centralized or distributed database, and / or associated caches and servers) storing one or more sets of executable instructions. The term "computer-readable storage medium" should also include any tangible medium capable of storing or encoding a set of instructions executable by a computer, which causes the computer to perform any one or more methods described herein. The term "computer-readable storage medium" should include, but is not limited to, solid-state memory, optical media, and magnetic media. A non-transitory computer-readable storage medium 935 that may contain instructions encoded thereon encoding any one or more methods or functions described herein includes implementations... Figure 1 The instructions for the adaptive filter bank 145 are for implementing the disclosed herein. Figure 5 Method 500 Figure 6 Method 600 and Figure 7Method 700. Any component can be stored and accessed from a storage device and / or from a storage device accessible via an interface bus. Although program components such as those in the component set are typically stored in local storage device 914, they can also be loaded and / or stored in other memories, such as: remote “cloud” storage facilities accessible via a communication network; integrated ROM memory 906; via an FPGA or ASIC implementing the component logic; and so on.

[0184] The memory 929 may contain a collection of program and / or database components and / or data, such as, but not limited to: operating system component 915; information server component 916; user interface component 917; database component 920; encryption server component 918; IVTMC component 919; etc. (i.e., collectively referred to as the component collection). The aforementioned components may be incorporated into (e.g., are sub-components of) IVTMC component 919, loaded from, or operably obtained from IVTMC component 919.

[0185] Operating system component 915 is an executable program component that facilitates the operation of the IVTMC controller 901. Typically, the operating system helps access I / O, network interfaces, peripheral devices, storage devices, etc. The operating system can be a highly fault-tolerant, scalable, and secure system, such as: Unix and Unix-like system distributions (such as AT&T's Unix; Berkeley Software Distribution (BSD) variants such as FreeBSD, NetBSD, OpenBSD, etc.; Linux distributions such as Red Hat, Debian, Ubuntu, etc.); and / or similar operating systems.

[0186] Information server component 916 is a stored program component executed by the CPU. The information server can be a traditional Internet information server, such as, but not limited to, Apache from the Apache Software Foundation, Microsoft's Internet Information Server, etc. User interface component 917 is a stored program component executed by the CPU. User interface component 917 can communicate with other components in the component set and / or with other components in the component set (including itself and / or similar facilities). Most commonly, the user interface communicates with operating system component 915, other program components, etc. The user interface can contain, convey, generate, obtain, and / or provide program components, systems, users, and / or data communications, requests, and / or responses. Encryption server component 918 is a stored program component executed by CPU 903, encryption processor 926, encryption processor interface 927, encryption processor device 926, etc. Encryption server component 918 facilitates secure access to resources on the IVTMC and facilitates access to secure resources on remote systems; that is, it can act as a client and / or server for secure resources.

[0187] Database component 920 may be embodied in a database and the data it stores. The database is a stored program component executed by the CPU; the stored program component partially configures the CPU to process the stored data. In one embodiment, database component 920 includes several tables 920a-f. The thermal coefficient vector table (TCV) 920a may include fields such as, but not limited to, tcv_id, tcv_type, tcv_value, and tcv_date. The thermal property (TP) table 920b may include fields such as, but not limited to, tp_id, tp_value, hvac_id, and zone_id. The weather estimation (WE) table 920c may include fields such as, but not limited to, we_id, we_time, we_value, and we_serviceProvider. The zone table 920d may include fields such as, but not limited to, zone_id, thermalDevice_id, hvac_id, and tvc_id. HVAC table 920e may include fields such as, but not limited to: hvac_id, hvac_model, hvac_zoneID, hvac_avgPowerConsumption, hvac_avgEnergyEfficiency, etc. Estimation (EST) table 920f may include fields such as, but not limited to: est_id, est_time, est_type, est_value, est_zoneID, etc. Any of the aforementioned tables can support and / or track multiple entities, accounts, users, etc.

[0188] IVTMC component 919 can be accessed via various components described herein (e.g., Figure 1 The comfort agent 120, comfort model 130, thermal model 140, thermal device 150, and weather model 160 convert user weather estimates and thermal attributes 920b into metrics such as estimated temperature, estimated energy consumption, and estimated power consumption. In one embodiment, the IVTMC component 919 acquires inputs (e.g., weather estimates 920c, thermal attributes 920b, etc.) and converts the inputs into outputs (e.g., estimated temperature, estimated energy consumption, estimated power consumption, etc.) via various components (e.g., the IVTMC component 919, etc.).

[0189] Various embodiments have been referenced in the foregoing. However, the scope of this disclosure is not limited to the specific embodiments described. Rather, any combination of the described features and elements, whether or not associated with different embodiments, is contemplated as an embodiment contemplated for implementation and practice. Furthermore, while embodiments may achieve advantages over other possible solutions or prior art, whether a given embodiment achieves a particular advantage does not limit the scope of this disclosure. Therefore, the foregoing aspects, features, embodiments, and advantages are merely illustrative and should not be considered as elements or limitations of the appended claims unless expressly stated in claim(s).

[0190] The various embodiments disclosed herein can be implemented as systems, methods, or computer program products. Therefore, aspects may take the form of entirely hardware embodiments, entirely software embodiments (including firmware, resident software, microcode, etc.), or embodiments combining software and hardware aspects, which may be collectively referred to herein as “circuit,” “module,” or “system.” Furthermore, aspects may take the form of computer program products embodied in one or more computer-readable media having computer-readable program code embodied thereon.

[0191] Any combination of one or more computer-readable media may be used. The computer-readable medium may be a non-transitory computer-readable medium. A non-transitory computer-readable medium may be, for example, but not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or any suitable combination thereof. More specific examples (not an exhaustive list) of non-transitory computer-readable media may include: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable optical disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. Program code embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wired, fiber optic cable, RF, etc., or any suitable combination thereof.

[0192] Computer program code used to perform the operations of various aspects of this disclosure can be written in any combination of one or more programming languages. Furthermore, such computer program code can be executed using a single computer system or through multiple computer systems communicating with each other (e.g., using a local area network (LAN), wide area network (WAN), the Internet, etc.). Although the various features described above are illustrated with reference to flowchart diagrams and / or block diagrams, those skilled in the art will understand that each block of the flowchart diagrams and / or block diagrams, and combinations of blocks in the flowchart diagrams and / or block diagrams, can be implemented by computer logic (e.g., computer program instructions, hardware logic, combinations of both, etc.). Typically, computer program instructions can be provided to processor(s) of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus. Furthermore, executing such computer program instructions using processor(s) produces a machine capable of performing the functions(s) or actions(s) specified in one or more blocks of the flowcharts and / or block diagrams.

[0193] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and / or operation of various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code, including one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative embodiments, the functions marked in the blocks may occur in a different order than those marked in the figures. For example, depending on the functions involved, two consecutively shown blocks may actually execute substantially simultaneously, or these blocks may sometimes execute in reverse order. It will also be noted that each block illustrated in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a system based on dedicated hardware, or a combination of dedicated hardware and computer instructions, that performs the specified function or action.

[0194] It should be understood that the above description is illustrative and not restrictive. Many other embodiments will become apparent from reading and understanding the above description. Although specific examples have been described in this disclosure, it should be recognized that the systems and methods of this disclosure are not limited to the examples described herein but can be implemented with modifications within the scope of the appended claims. Therefore, the specification and drawings are to be considered illustrative and not restrictive. Accordingly, the scope of this disclosure should be determined by reference to the appended claims and the full scope of their authorized equivalents.

Claims

1. A method for detecting diagnostic events in a thermal system, comprising: The controller device generates thermal coefficients at the adaptive filter bank to characterize the heat transfer of the volume associated with the thermal system; The controller device applies one or more filters to the thermal coefficient based on the sampling rate; In response to the application of the filter, the controller device generates one or more estimated thermal coefficient thresholds based on the sampling rate; The controller device determines whether at least one of the filtered thermal coefficients satisfies at least one of the estimated thermal coefficient thresholds; as well as Based on the determination, the controller device provides alarm information indicating diagnostic events associated with the thermal system. The process of generating the estimated thermal coefficient threshold also includes determining whether the sequence of thermal coefficient vectors supports the definition of a diagnostic event.

2. The method according to claim 1, wherein, Generating the thermal coefficient also includes: Identify thermal models for applications using thermal systems; Based on the thermal model, the estimation error of the reference signal relative to the main signal associated with the volume is determined; as well as Based on an adaptive filter, the thermal coefficient is adapted to satisfy a solution associated with the volume, taking into account the estimation error.

3. The method according to claim 1, further comprising: The sampling rate is adapted based on the filter operation associated with at least one of the filters.

4. The method according to claim 3, wherein, The filter operation includes at least one infinite impulse response filter.

5. The method according to claim 1, further comprising: Determine whether at least one thermal coefficient exceeds an upper or lower limit window associated with at least one estimated thermal coefficient threshold.

6. The method according to claim 1, wherein, The alarm message indicates abnormal operation of the device associated with the volume.

7. A system for detecting diagnostic events in a thermal system, comprising: A memory that stores multiple thermal coefficient data; and A controller device, operatively coupled to the memory, for: Thermal coefficients are generated at the adaptive filter bank to characterize the heat transfer of the volume associated with the thermal system; Apply one or more filters to the thermal coefficients based on the sampling rate; In response to the applied filter, one or more estimated thermal coefficient thresholds are generated based on the sampling rate; Determine whether at least one of the filtered thermal coefficients satisfies at least one of the estimated thermal coefficient thresholds; as well as Based on the determination, alarm information indicating diagnostic events associated with the thermal system is provided. In order to generate the estimated thermal coefficient threshold, the controller device is further configured to determine whether the sequence of thermal coefficient vectors supports the definition of a diagnostic event.

8. The system according to claim 7, wherein, In order to generate the thermal coefficient, the controller device is further configured to: Identify thermal models for applications using thermal systems; Based on the thermal model, the estimation error of the reference signal relative to the main signal associated with the volume is determined; as well as Based on an adaptive filter, the thermal coefficients are adapted to satisfy a solution associated with the volume, taking into account the estimation error.

9. The system according to claim 7, wherein, The controller device is also configured to: The sampling rate is adapted based on the filter operation associated with at least one of the filters.

10. The system according to claim 9, wherein, The filter operation includes at least one infinite impulse response filter.

11. The system according to claim 7, wherein, The controller device is also configured to: Determine whether at least one thermal coefficient exceeds an upper or lower limit window associated with at least one estimated thermal coefficient threshold.

12. The system according to claim 7, wherein, The alarm message indicates abnormal operation of the device associated with the volume.

13. A non-transitory computer-readable storage medium comprising executable instructions, which, when executed by a controller device, cause the controller device to: Apply one or more filters to the thermal coefficients based on the sampling rate; In response to the applied filter, one or more estimated thermal coefficient thresholds are generated based on the sampling rate; Determine whether at least one of the filtered thermal coefficients satisfies at least one of the estimated thermal coefficient thresholds; as well as Based on the determination, alarm information indicating diagnostic events associated with the thermal system is provided. In order to generate the estimated thermal coefficient threshold, the controller device is further configured to determine whether the sequence of thermal coefficient vectors supports the definition of a diagnostic event.

14. The non-transitory computer-readable storage medium according to claim 13, wherein, To generate the thermal coefficient, the controller device is further configured to: Identify thermal models for applications using thermal systems; Based on the thermal model, the estimation error of the reference signal relative to the main signal associated with the volume of the thermal system is determined; as well as Based on an adaptive filter, the thermal coefficients are adapted to satisfy a solution associated with the volume, taking into account the estimation error.

15. The non-transitory computer-readable storage medium according to claim 13, wherein, The controller device is also used for: The sampling rate is adapted based on the filter operation associated with at least one of the filters.

16. The non-transitory computer-readable storage medium according to claim 15, wherein, The filter operation includes at least one infinite impulse response filter.

17. The non-transitory computer-readable storage medium according to claim 13, wherein, The controller device is also used for: Determine whether at least one thermal coefficient exceeds an upper or lower limit window associated with at least one estimated thermal coefficient threshold.

Images