Method and apparatus for determining machine health

Abstract

Systems and methods for determining machine health score are provided. They include a computer and a first sensor positioned to determine sensor data relating to a characteristic of a machine during a plurality of machine operating states. Logic and instructions executable by the computer are included for performing several steps including: determining the plurality of machine operating states; receiving an indication of an operating state of interest selected from the plurality of the operating states; receiving values of acceptable behavior of the machine during the machine operating state of interest; receiving second sensor data relating to a characteristic of the machine during the machine operating state of interest; and determining the machine health score by comparing second sensor data to values of acceptable behavior of the machine.

Description

[0001] Title:

[0002] Method and Apparatus for Determining Machine Health

[0003] Cross-Reference to Related Applications:

[0004] This application is related to and claims priority from US Provisional Application Serial Number 63 / 610,104 filed on December 14, 2023. This application is also related to US Applications entitled (1) “Method and Apparatus for Local Sensing” which received US Provisional Application Serial No. 62 / 739,419; (2) “Systems and methods to integrate environmental information into measurement metadata in an Electronic Laboratory Notebook Environment” which received US Provisional Application Serial No. 62 / 739,427 and US Application Serial No.16 / 589,347; (3) “Method and Apparatus for Process Optimization” which received US Provisional Application Serial No. 62 / 739,441 and US Application Serial No. 16 / 589,713; (4) “Method and apparatus for determining freezer status”, which received US patent no. 11,561,037 (P-012); (5) “Method and system for contextual notification”, which received US App. Ser. No. 18 / 026,990 (P-014); and (6) “Method and Apparatus for Noninvasive Determination of Utilization” which received US Provisional Application Serial No. 17 / 800,205 (P-018). These applications are incorporated in their entireties herein by reference for all purposes. All references mentioned herein, including for example websites, articles, reference books, textbooks, granted patents, and patent applications are incorporated in their entireties herein by reference for all purposes.

[0005] Background of the Invention:

[0006] As machines become more and more complex, there is an increasing need for a way to assess the health of the machine before it breaks down, goes out of calibration, or needs service. To stay competitive, companies need to have the machines they rely on running according to their performance specifications and to avoid unscheduled downtime if a machine were to break down (or operate outside of acceptable performance criteria). It is understood that machines will have downtime when they need to be taken out of operation for servicing, repairs, calibration, etc., but it is advantageous to plan the down time in advance to allow for workflows to be adjusted (for example, shifting the workload to other machines or ending a process at a good stopping point) instead of having to take a machine out of service unexpectedly when it may experience issues with little to no warning.

[0007] In a manufacturing context where production runs may run for hours or days continuously, a machine may be running outside of desired performance specifications unknown to the manufacturer. In this case, there is a risk that the output product may be manufactured out of tolerance and hence fail quality control testing, resulting in a financial loss as the product is scrapped and deemed not saleable.

[0008] There are currently common ways to reduce the risk of a machine operating outside of desired specifications, such as having a fixed schedule for servicing (e g. once a month, once a quarter, or once a year). This is not an ideal approach since machines may begin to operate outside of their performance specification in between the scheduled servicing times and because such an approach that is based on the time between servicing cycles assumes that the machine experiences relatively consistent usage and wear and tear between those times. This is not usually the case since the demands on a machine may vary over its use life.

[0009] An improved method of determining when to perform servicing is to estimate and / or measure how much a machine has been used (via usage or utilization metrics) and then schedule preventative servicing based on the machine exceeding a given amount of time or cycles. While this is an improvement since one can consider how much a machine has been used and attend to machines that are used for more time or cycles than others, it still lacks context around how heavily it may have been used (i.e. the amount of stress the machine may have experienced and not just the total time). For example, two machines may have been run for the same duration and / or number of cycles, but one of the machines may have been run at a high-intensity setting (resulting in more wear and tear on it) than the other machines which may have been run at a low-intensity setting (resulting in less wear and tear). This can be the case for machines that are designed to be operated in different states, such as a centrifuge, which can be run at different rotational speeds (RPMs), a robotic arm that can be run at different speeds or be used to carry different weights of materials. Furthermore, solely relying on measuring the utilization of a machine (total time or number of cycles) may not allow detection of early signs the machine is behaving in an anomalous manner, which can indicate that a larger issue may occur soon (especially before the next usage-based servicing was scheduled to occur). There is a need for improved systems and methods for detecting behaviors of a machine that can be a measure of the machine’s health and that can provide early indicators that the machine’s health is deteriorating (or has deteriorated).

[0010] Brief Summary of the Invention:

[0011] The present invention solves the problems in the art and provides systems and methods for detecting behaviors of a machine that can be a measure of the machine’s health and also that can provide early indicators that the machine’s health is deteriorating (or has deteriorated). Based upon these observations, alerts can be issued, and maintenance can be scheduled. In a first embodiment, the present invention provides a system for determining a machine’s health score. The system includes a computer and a first sensor. The first sensor is positioned to determine sensor data relating to a characteristic of a machine during a plurality of machine operating states. The computer comprises a processor, memory, and logic and instructions for performing the steps of: a. receiving, by the computer, first sensor data from the first sensor relating to a characteristic of the machine during each of the plurality of machine operating states; b. determining and storing in memory, by the computer, the plurality of machine operating states from the first sensor data received in step a.; c. receiving and storing in memory, by the computer, an indication of an operating state of interest selected from the plurality of the operating states determined and stored in memory in step b.; d. receiving and storing in memory, by the computer, values of acceptable behavior of the machine during the machine operating state of interest; e. receiving, by the computer, second sensor data relating to a characteristic of the machine during the machine operating state of interest; and f. determining and storing in memory, by the computer, the health score of the machine from the second sensor data by comparing the second sensor data to the values (or range of values) of acceptable behavior of the machine during the machine operating state of interest received and stored in step d., thereby determining the health score of a machine. In a preferred application of the first embodiment, the machine is a cold storage unit (CSU) having a compressor and / or pump (having on / cooling and off / warming operating states) and the first sensor is configured to determine a characteristic of power or energy consumption of the compressor and / or pump.

[0012] In a second embodiment, the present invention provides another system for determining a machine health score. The system includes a computer and a first sensor, where the first sensor is positioned to determine sensor data relating to a characteristic of a machine during a machine operating state of interest. The computer includes a processor, memory, and logic and instructions for performing the steps of: a. configuring the first sensor to determine and transmit the sensor data relating to a characteristic of a machine during the machine operating state of interest; b. receiving and storing in memory, by the computer, values (or range of values) of acceptable behavior of the machine during the machine operating state of interest; c. receiving, by the computer, the sensor data relating to a characteristic of the machine during the machine operating state of interest; d. determining and storing in memory, by the computer, the health score of the machine from the sensor data by comparing the sensor data to the values (or range of values) of acceptable behavior of the machine during the machine operating state of interest received and stored in step b., thereby determining the health score of a machine.

[0013] In a third embodiment, the present invention provides a system for determining a machine health score, the system comprising a computer and a first sensor, where the first sensor is positioned to determine sensor data relating to a characteristic of a machine during at least one machine operating state of interest. The computer comprises a processor, memory, and logic and instructions for performing the steps of: a. receiving, by the computer, first sensor data from the first sensor relating to a characteristic of the machine during the at least one machine operating state of interest; b. determining and storing in memory, by the computer, at least one machine operating state of interest from the first sensor data received in step a.; c. receiving and storing in memory, by the computer, an indication of the at least one operating state of interest; d. receiving and storing in memory, by the computer, values (or range of values) of acceptable behavior of the machine during the at least one machine operating state of interest; e. receiving, by the computer, second sensor data relating to a characteristic of the machine during the at least one machine operating state of interest; f. determining and storing in memory, by the computer, the health score of the machine from the second sensor data by comparing the second sensor data to the values (or range of values) of acceptable behavior of the machine during the at least one machine operating state of interest received and stored in step d., thereby determining the health score of a machine.

[0014] In a fourth embodiment, the present invention provides a cold storage unit (CSU) health score determination and modification system having: a computer; and a first sensor. The first sensor is positionable to determine sensor data relating to a characteristic of the CSU having a compressor and / or pump during on and off compressor / pump operating states. The computer includes a processor, memory, and logic and instructions for performing the steps of: a. receiving, by the computer, first sensor data from the first sensor relating to a characteristic of the CSU during the off and on compressor / pump operating states; b. receiving and storing in memory, by the computer, values (or range of values) of acceptable behavior of the CSU during the off and / or on compressor / pump operating states; c. determining and storing in memory, by the computer, the CSU health score by comparing the first sensor data to the values of acceptable behavior of the CSU during the off and on compressor / pump operating states received and stored in step b.; d. receiving, by the computer, second sensor data relating to events that affect or cause anomalies in the first sensor data; and e. adjusting or ignoring, by the computer, any or all of the first sensor data received by the computer in step a. and / or health score determined in step d. in response to second sensor data received in step d.

[0015] In yet further embodiments, the present invention provides methods for determining a machine health score, optionally using them in any of the systems herein described.

[0016] Brief Description of the Figures:

[0017] Figure 1 shows a temperature profile ( 1 A) and an energy consumption profile (IB) of an ultra-low temperature (ULT) freezer.

[0018] Figure 2 shows another temperature profile (2A) and another energy consumption profile (2B) of another ULT freezer.

[0019] Figure 3 shows an energy consumption profiling of another ULT.

[0020] Figures 4 to 8 show temperature profiles of other cold storage units (CSUs).

[0021] Figure 9 and 10 show block diagrams for preferred steps performable according to the preferred embodiments and systems of the present invention.

[0022] Figure 11 shows an example of a health score determined for an ULT freezer in accordance with the invention.

[0023] Figure 12 shows a block diagram of a preferred system according to the present invention.

[0024] Figures 13 to 15 show temperature profiles (top) and an energy consumption profiles (bottom) of ultra-low temperature (ULT) freezers. Detailed Description of the Invention:

[0025] Without being bound by particular modes of operation, the present invention will be described in the following detailed description which contains preferred embodiments of the invention. It is useful to understand whether a machine is operating as it should (e.g. healthy) or not (e g. not healthy and needs to be serviced and / or replaced). Prior art methods to determine if a machine is healthy, or not, include the measurement of some parameter from the machine and a simple cross-reference to determine if the parameters are within acceptable limits. Machines, however, operate in different states and can exhibit different behaviors in these different states. The present invention solves problems in the art and by providing methods and systems for assessing health of a machine by actively monitoring operating and usage patterns to identify early indicators of potential machine health deterioration (e.g. not just total time the machine was used, but also operating characteristics related to the stress and / or load on the machine) and how the machine is behaving (e.g. is it showing signs of potential damage before a significant deterioration of its performance or a breakdown).

[0026] Definitions:

[0027] A “cold storage unit” (CSU) includes any freezer, refrigerator, walk-in freezer, cold room, ULT freezer (ultra low temperature freezer including those commonly referred to as “-80 freezers” or “-80C freezers), cryogenic storage units, liquid nitrogen cooled storage units,.

[0028] The term “power” is used to denote quantities that are related to energy per time from electromagnetic sources including but not limited to current and voltage as well as quantities that are related to current, voltage, and power etc. Unless the context indicates otherwise and / or is expressly limiting the terms current, power, electricity, voltage, energy etc. are used interchangeably herein.

[0029] The terms “service / servicing” etc. as used herein means any action to prepare, maintain, set, clean etc. a machine into good or acceptable working order. These actions include and are not limited to maintenance, repair, calibration, cleaning, defrosting, removing build-up, replacing parts, replacing components, reprogramming, replenishing and / or refilling consumable items and / or materials, flushing out a system, and the like.

[0030] The term “machine” includes instruments, electronic, mechanical, electromechanical, manufacturing equipment, facilities equipment, HVAC systems, laboratory equipment, CSU, computers, equipment that comprises a computing device, software, cloud computing apparatus, vehicles including automobiles, busses, trains, planes, spacecraft, boats, ships, and the like. In certain embodiments the machines are powered by electricity, heat, solar, wind, wave, hydrogen, fossil fuels, and the like.

[0031] The term “health” refers to a set value or determination value which is indicative of the quality of operation of the studied machine, which can be represented by an operating quality value (OQV). As described herein, the health of a machine can be set at a known value or determined through observation of operating characteristics of the machines at one or more operating states.

[0032] The term “operating state” refers to known or determined values of machines abilities. A machine can have one or more different operating states which require the same or different power consumption rates. For example a fan can have two or more set operating states such as those selected from the group consisting of: on, off, low, medium, and / or high, each of which result in different power consumption. These states can be represented by an Operating State Value (OSV).

[0033] Reference throughout the specification to “one embodiment,” “another embodiment,” “an embodiment,” “some embodiments,” and so forth, means that a particular element (e.g., feature, structure, property, and / or characteristic) described in connection with the embodiment is included in at least one embodiment described herein, and may or may not be present in other embodiments. In addition, it is to be understood that the described element(s) may be combined in any suitable manner in the various embodiments.

[0034] Numerical values in the specification and claims of this application reflect average values for a composition. Furthermore, unless indicated to the contrary, the numerical values should be understood to include numerical values which are the same when reduced to the same number of significant figures and numerical values which differ from the stated value by less than the experimental error of conventional measurement technique of the type described in the present application to determine the value.

[0035] The present invention provides several examples and details regarding how these steps are performed across several different machine types and architecture. After understanding the present invention in its entirety, one skilled in the art would be able to apply the teachings herein made to other types of machines without undue experimentation. General Aspects of the Invention:

[0036] In a first aspect, the present invention provides systems and methods for determining health of a machine by determining operating states of the machine, receiving / determining an indication of an operating state of interest, receiving an indication and / or determining how a healthy machine should be operating during the operating state of interest, and comparing how the machine is operating during the operating state of interest to how a healthy machine should be operating during the same state.

[0037] In a second aspect, the present invention provides systems and methods for determining health of a machine by receiving sensor data regarding a characteristic of the machine when the machine is in the operating state of interest, determining how a healthy machine should be operating during the operating state of interest, and comparing how the machine is operating during the operating state of interest to how a healthy machine should be operating during the same state. In this second aspect, determination of the various operating states of interest can be avoided. Instead, the systems and methods can make use of a sensor configured to solely provide sensor data when the machine is operating in the state of interest.

[0038] The first and second aspects of the present invention are directed to machines in general. However, in preferred embodiments, the machines are cold-storage units (CSUs). In a specific third aspect of the invention, the present invention provides an improvement upon freezer health determination systems of the art which simply employ observations of freezer duty cycles in determination of freezer health. The third aspect of the present invention can used separately or combined with any of the first two aspects, or any other aspect of the present invention. In the third aspect, the present invention observes and makes use of external events (e.g. events cause of a user or otherwise, such as freezer door openings etc.) which may affect the quality of data received and which would otherwise have been used in raw form to make a determination of health. In this third aspect, if an event is detected and / or determined (e.g. door opening etc.), the raw data can be processed to account for the event and / or simply ignored / discarded in health determination. Alternatively, or coextensively, if an event is detected and / or determined health score determination metrics may be altered to account for the event. Data Sources:

[0039] The data for performing the machine health assessment can be determined and come from different sources. For example, the machine itself can output the data in a format that is processable by a computing apparatus (e.g. a data stream). Examples of such data streams include digital data transmitted by serial protocols including RS232, RS485, USB; other digital protocols including ethernet; wireless protocols including wifi, bluetooth, bluetooth low energy, ZigBee, LoRa, etc; and / or analog output formats such as a voltage or current output, 4-20 mA current loop output, etc. Alternatively, or coextensively, the machine can be associated with and monitored by one or more independent sensor devices (e.g. temperature sensors, power sensors, electromagnetic sensors, Hall effect sensors, vibration sensors, motion sensors, gyroscopic sensors, optical sensors, cameras, chemical sensors, biosensors, physical sensors, stress sensors, piezo sensors, etc). One ordinarily skilled in the art will recognize that a variety of sensor types and data streams can be used so long as the quantity being measured is relevant to assessing the health of the machine.

[0040] One example of how assessing the health of a machine can be accomplished is by: monitoring and recording the operational usage characteristics of the machine (e.g. heavy load or light load, and not just total duration or number of cycles it was used) by identifying different operating states in which the machine operates; identifying operating states that are relevant and / or are of interest in terms of machine health or may show early indications of health problems; determining acceptable and unacceptable behavior in pertinent operating states of interest by specifying criteria for identifying anomalous behavior within the described states. In the event unacceptable behavior is identified, an alert and / or notification can be triggered to prompt subsequent actions, including but not limited to relaying a message to an individual, another machine, or a computer for further intervention.

[0041] Certain preferred aspects of the present invention include the following conceptual framework: 1. Determine an operating state of a machine by using at least one sensor (or type of data) to classify and / or identify each relevant operating state that is of interest for estimating machine health; 2. Determine the characteristics of “acceptable behavior” in a given operating state; 3. Determine if the machine is operating in a healthy manner (ie. within acceptable criteria limits) or not. “Acceptable behavior” detection and / or determination can be done by the same one sensor (or data stream) or with two or more sensors or data streams. The second, or additional sensors (or data stream(s)) can be measuring the same (or related) quantity as the first sensor (or data stream) or it can measure a different quantity. In certain preferred embodiments, at least two different types of sensors (or data streams) can, in part or in whole, perform steps 1 and 2 above (i.e. sensors or data streams that measure different quantities, like current / power / energy / vibration for one measurement and temperature for the other). One can use sensors (or data streams) that measure the same type of quantity (e.g. two or more temperature sensors, two or more power sensors, two or more vibration sensors, etc) but measuring different operating parts of the machine.

[0042] In additional embodiments, machine operating characteristics can be measured over time by measuring and analyzing electrical input (via a voltage or current meter) to the machine over time; measuring and analyzing a portion of the interior temperature (via a thermocouple or temperature sensor) of the machine over time; measuring and analyzing the ambient temperature (via a thermocouple or temperature sensor) of the room surrounding the machine over time; measuring and analyzing the temperature (via a thermocouple or temperature sensor) of the machine over time; measuring and analyzing sound indicative of machine operating cycles over time; and / or measuring and analyzing vibration (via a microphone, waveguide, piezoelectric sensor, accelerometer, or other vibration, movement and / or sound sensor etc.) that may be indicative of machine operating states over time. Observing or measuring any one of these variables over time provides either direct or indirect information regarding the operating states and therefore provides either direct or indirect observation regarding operating states and / or cycling over time. For example, a temperature sensor placed in, on, or near the machine reveals an elevated temperature which is indicative of the machine being “on” and a cooler temperature when the machine is “off’. As another example a temperature sensor placed within a freezer space reveals a reduction in temperature which is indicative of the freezer / compressor being “on” and a rise in temperature when the freezer / compressor is “off’. As machine operating states and / or cycles lengthen, this can be equated with degraded health of the machine.

[0043] Determining Operating States of the Machine:

[0044] Employing the data from one or more different sources as described herein, the different operating states of a machine that are relevant to its health and performance can be defined. There are several scenarios of how a machine can be used and / or operated, resulting in different operating states. Tn a first example (e g. an exhaust fan running at a fixed speed or a vacuum cleaner that has one power setting) a machine has two states of operation: State 1 - Off with no energy consumption; and State 2 - On and consuming energy at a relatively constant level.

[0045] In another example a machine can have more than two states of operation, for example a machine with operating states including: State 1 - Off with no energy consumption; State 2 - On but consuming little energy such as being in a standby mode, sleep mode, low power mode, power-saving mode, hibernating mode, or similar; and State 3 - On and being actively used and consuming a higher level of energy (e.g. HVAC system, cold storage unit, fridge, freezer, oven, incubator, television, treadmill).

[0046] In this example it may not be necessary to differentiate between two or more operating states (i.e. it may be acceptable to the user to treat multiple operating states as equivalent to each other for the purposes of assessing the machine’s health). Here, a machine may consume different amounts of energy even when it is in one operating state, but that may not be relevant to the user who is seeking to determine the health of the machine. For example, a treadmill can be operated at different speeds, which would cause the motor to consume different levels of energy; however, a user may only be interested in the three designated states of 1) off, 2) standby mode and 3) running (at wherever speed). In this example, State 1 and State 2 may be considered to be equivalent states for the purpose of determining the health of a machine.

[0047] In additional examples a machine may have more than two states of operation AND each state needs to be identified and / or quantified separately. For example, a fan can have four operating states: off, high, medium, or low speeds, where each operating state is expected to cause a different level of stress and deterioration on the motor and associated components. In another example a centrifuge can be run at many different RPMs (revolutions per minute) and each rotation speed (or range of RPMs) can be designated as an operating state of interest. In yet a further example a temperature-controlled machine (such as a cold storage unit) can be set to different target temperatures at different times, where colder temperatures are expected to use more energy and / or cause more stress and / or deterioration to the compressor (or other electromechanical components) as compared with setting the unit to higher temperatures. In these examples it could be expected that higher RPMs, and / or energy consumption, could cause more stress and deterioration on the motor / compressor and associated mechanical components. The data received by one or more sources described above can be quantified into various operating states of the machine. An example of such quantification is by reference to a lookup / reference chart and / or reference to operating manuals available online or stored in memory. The quantified operating states can then be stored in memory and employed in further analysis of the machine.

[0048] Determining Machine Operating States:

[0049] Determining the operating states of a machine can be contingent upon its type and operational context. An approach to this determination varies based on the machine's characteristics and application. The following non-limiting examples show various parameters that can be measured to discern operating states of a machine and different sensor and / or sensor data streams that can be useful in the determination.

[0050] Camera Utilization: Leveraging a camera, whether for video or image capture, proves useful in identifying optical markers associated with the machine's operating state. Examples include, but are not limited to: Recognition of a switch position to infer the configured operating state; Detection of an indicator light status (on / off, color) to signify the current operating state; Analysis of machine motion to discern its operating state (for example, a robotic arm exhibiting horizontal or vertical movements), the direction of the robotic arm motion can be determined using video cameras and established image recognition methods;

[0051] Electromagnetic Properties: including as measuring electromagnetic parameters can indicate a machine's operating state, such as measuring power, current, voltage, energy, electric fields, and magnetic fields measurements, measuring visible light characteristics, including intensity, color, and flashing patterns, and measuring hyperspectral light (outside of the human visible spectrum); measuring electromagnetic radiation.

[0052] Physical, Mechanical, and / or Chemical Properties: Various chemical, mechanical and / or physical parameters can be measured to infer operating states. Examples include, but are not limited to determining / detecting: movement (vibration, rotation, any type of motion) using a motion sensor such as an accelerometer or gyroscope; sound level (including ultrasonic and those frequencies outside of human audible range); sound analysis (such as detecting and analyzing frequency spectra of sound signals); temperature monitoring; mass change measurement; air flow measurement; fluid flow measurement; detection of chemical changes; relative humidity and / or absolute humidity (e.g. for machines like humidifiers, AC units, or dehumidifiers); and pressure.

[0053] One ordinarily skilled in the art will recognize that there can be more other parameters that can be measured and quantified. Furthermore, one ordinarily skilled in the art will recognize that such data can be obtained from measurements made from sensors or devices that are independent of the machine under observation or that these measurements can be made by sensors or devices that have been built into the machine itself. In the former case, the independent sensors or devices can be placed in proximity to the machine, mounted on or within the machine, or placed at a location that is sufficient to measure the parameter of interest. In the latter case, the measurements can be transmitted from the machine via an output mechanism such as a digital or analog output port, or via a wireless method such as wifi, bluetooth, ZigBee, LoRa, or similar.

[0054] Determining the Machine Operating State of Interest:

[0055] As explained throughout the present specification, operation of a machine at various operating states has differing implications on the physical components of the machine. For example, a machine running in the highest power mode can be expected to result in the highest wear on physical components. Similarly, a machine run in a low power state may be equated with minimal wear on the physical components. In these situations, a user or programmed circuity may decide that the machine operation states of interest include when the machine is operating under medium and / or high power consumption conditions. This can be determined by and / or input into the computer and stored in memory for later use in the health determination protocol.

[0056] Determining Acceptable Behavior for a Given Operating State and / or Operating State of Interest:

[0057] In a further step, detection / quantification / determination of acceptable behavior (e.g. reference health value or range of values) of the machine in a given operating state can be determined and stored in memory. Different operating states may have different types of acceptable behavior that may need to be computed differently. Different operating states can have different acceptable limits for the same behavior (thus computed the same way in different states, but what is deemed acceptable may have different limits). The following examples are provided for determining acceptable behavior of the machine at a given operating state.

[0058] Example 1 : During State 1, the acceptable behavior may be that the fundamental frequency of vibration of a machine is between 30-50 Hz, but in State 2, the acceptable behavior may be that the fundamental frequency of vibration is between 20 - 70 Hz. In both states, the same mathematical calculation can be used as the score for determining machine health, but the acceptable range is different.

[0059] Example 2: During State 1, the acceptable behavior may be that the fundamental frequency of vibration of a machine is between 30-50 Hz, but in State 2, the acceptable behavior may be that the temperature of the machine is between 40C to 60C. Thus, each operating state has a different metric to be calculated using different sensor readings or data streams.

[0060] Example 3: During State 1, the acceptable behavior may be that the fundamental frequency of vibration of a machine is between 30-50 Hz but does not change by more than 5 Hz per minute, but in State 2, the acceptable behavior may be that frequency of vibration is between also 30-50 Hz but does not change by more than 2 Hz per minute. In both states, the same calculation can be used to determine the fundamental frequency of vibration, but how fast that fundamental frequency is changing is the indicator of acceptable behavior (which is different for each state). Thus, the rate of change (or variability) of the Health Score may also be used to identify when a machine’s behavior is of concern to a user.

[0061] Example 4: Acceptable machine operating characteristic levels are determined immediately after a scheduled service of the machine. The machine is assumed to be in good working order and the determined characteristic level is deemed, by a user or computer, to be representative of acceptable behavior of the machine during the respective operating state.

[0062] Example 5: A Reference Health Score can be used to establish a baseline of acceptable machine behavior (ie what Health Score range indicates healthy behavior and what Health Score range indicates unhealthy behavior?). One can determine this: (1) by theory (what’s theoretically expected according to theoretical models); (2) by running the machine in a known health state and determine it empirically and just having a fixed model of acceptable behavior; (3) by crowdsourcing acceptable behavior from many of different machines of the same / equivalent make / model / etc from which one is able to determine what a baseline healthy behavior should look like (4) by looking at historical data and crafting a “relative Health Score”. The look back period of option (4) can be a moving window (compare the machine’s behavior during the past one week to the behavior of the machine during the previous month). The look back period of option (4) can be a fixed time window relative to an event that is relevant to the machine’s health (compare the machine’s behavior during the past one week to the behavior of the machine during the first month after it was last serviced or calibrated). The look back period can be an absolute fixed time window (compare the machine’s behavior during the past one week to the behavior of the machine between Jan 25, 2023 and March 15, 2023) etc.

[0063] Determining a Health Score and / or overall health of a Machine:

[0064] A health score can be determined that provides an indication of the health of a machine. The health score can be determined by operating the machine at the operating state(s) of interest and receiving operational sensor data regarding an operating characteristic of the machine. This operational data can then be compared to data at the values (or range of values) of acceptable machine operation (e.g. reference health values or range of values) determined and stored in memory.

[0065] The determined health score can be compared to a reference health score, or range of scores, to determine whether the machine is currently in a healthy condition or not. The health score can be representative of the offset of the operational values to the reference values.

[0066] Instead of computing the Health Score only for states of interest, one can compute the Health Score for all operating states, but just ignore the irrelevant states or weight the health score from the irrelevant states less and weight the Health Score from the relevant state(s) more (for example do a weighted sum or weighted average).

[0067] The Health Score can be determined on a periodic basis (for example, every month, every week, every day, every hour, every minute, every second, or more or less frequently), or in response to an event that has occurred or will occur (for example, one can compute the Health Score prior to a scheduled servicing activity, or just after one) or on an ad hoc basis (that is, not on a periodic basis or not in response of anticipation of an event). Furthermore, the Health Score can be determined by and / or is selected from the group consisting of: an instantaneous calculation; a continuously computed value (e.g. using a moving window); performed at periodic intervals; computed by taking different windows of data; calculated when an operating condition is met (e.g. do it only when in a particular state, or when the machine changes operating state); calculated in response to an event that is internal to the machine’s operation or external to the machine’s operation (e.g. calculate the Health Score when the machine is powered on from an off state AND / OR calculate the Health Score of a machine when the ambient temperature of the room that the machine is located in exceeds a certain threshold).

[0068] In order to assess the health of a machine, healthy operation and unhealthy operation criteria / values can be identified to analyze regions of operation. It can also be useful to define boundary conditions between these two regions. Furthermore, it can be useful to not just define two regions of operation (health and unhealthy), but also the degree to which the machine is operating in the respective region. For example, if the health of a machine can be represented as a number from 0.000 to 100.000, then one may choose to define the following “Health Score” regions and corresponding actions to be taken: 0.000 - 25.000 = Extremely unhealthy: take machine offline and service immediately; 25.001 - 50.000 = Moderately unhealthy: can continue operating the machine, but schedule servicing within 2 weeks; 50.001 - 75.000 = Moderately healthy: can continue operating the machines, but schedule servicing within 6 weeks; 75.001 - 100.000 = Extremely healthy: machine can continue to operate, and no servicing is needed until further notice. Alternatively, one may choose Health Score regions and corresponding actions along the following: 0.000 - 30.000 = Extremely unhealthy: stop machine operation immediately and service it; 30.001 - 40.000 = Very unhealthy: stop the machine within 48 hours and service it; 40.001 - 60.000 = Moderately unhealthy: can continue running the machine for 1 week, but then stop machine operation and service it; 60.001 - 65.000 = Moderately healthy: no action needed, but inspect error logs for any indication of internal error codes; 65.001 - 100.000 = Healthy: no action needed. Myriad variation of health score regions and associated actions can be easily envisioned.

[0069] The health score can also be determined by a reference value compared against a measured value and could be some percentage of a known value indicative of machine health level. For example when it is determined that the measured operating value is greater than, equal to, or less than a reference value in a lookup table (or some function of a reference value in a lookup table such as 33%, 50%, 66%, 75%, 90%, 125%, 150%, 200%, 300%, 400%, 1000% etc. of some variable), it can be determined that the machine is in a particular health state.

[0070] It can also be useful to compute the Health Score for a machine over a period of time that is long enough to be representative of the machine’s performance.

[0071] In another example, one may choose to also consider the trend of the Health Score to provide additional information and potentially predict when a machine may be transitioning from a healthy operating region to an unhealthy one. For example, the Health Score for a machine may be in the highly healthy region but it is showing a downward trend towards the unhealthy region. In this scenario, even if the machine is currently exhibiting a Health Score in a healthy region, it is useful to have a notification sent to inform a user that the machine is expected to enter an unhealthy operating region in the near future.

[0072] Determining a Health Score and / or overall health of Specific Machines:

[0073] For machines that are used for maintaining temperatures below the ambient temperature of the environment they are placed in (e.g. cold storage, freezers, refrigerators, etc), the operating state of interest can be when the cold-inducing mechanism (e.g. compressor) is running or consuming energy (operating state) using a variety of measurement techniques (energy consumption, vibration of compressor, or the temperature itself going down). The value of acceptable behavior can be the behavior of internal temperature. Acceptable behavior can be equated with a determination that the internal temperature of the CSU is strictly decreasing temperature. For example, intermittent or continuous temperature sensor reading demonstrates that each successive reading is lower than the previous reading.

[0074] For machines that are used for maintaining temperatures above the ambient temperature of the environment they are placed in (e.g. incubators, ovens, etc.), the operating state of interest can be when the warm-inducing mechanism (e.g. heating coil) is running or consuming energy (operating state) using a variety of measurement techniques (energy consumption, the temperature itself going up, perhaps an indicator light turning on, etc.). The value of acceptable behavior can be the behavior of the internal temperature. Acceptable behavior can be equated with a determination that the internal temperature of the machine is strictly increasing temperature in warm storage. For example, intermittent or continuous temperature sensor reading demonstrates that each successive reading is higher than the previous reading.

[0075] For machines that have moving parts (e.g. centrifuges, die presses, conveyor belts, shakers / mixers, etc.), the operating state of interest is when the machine is running and / or running at a certain energy consumption (high / low etc.) using a variety of measurement techniques (energy consumption, temperature of the machine goes up, vibration of the machine, perhaps an indicator light turning on, etc.). The value of acceptable behavior can be a reference value (e.g. Vibration amplitude and / or frequency content) where higher or lower values (e.g higher harmonics) can indicate if the machine has impaired health or is behaving as expected.

[0076] Performing an action based on the Health Score:

[0077] The system and methods further comprise the steps of correlating the health score with a lookup chart and / or reference value to determine if a correlation exists (e.g. for example the machine is in an unhealthy state, healthy state, trending in a health or unhealthy direction etc.). If a correlation is determined to exist further action can be taken. Further actions can be selected from the group consisting of: triggering an alarm; turning the machine on / off; instituting a machine calibration routine; and issuing an alert. The alert can be sent to another computing device and / or a user (e.g audibly, visually, via voice, text, SMS, email, etc). The alert can also be an alert comprising a message (e.g. a contextual message) containing information selected from the group consisting of warning of an immediate or impending condition; the health score of the machine; the trend of health scores of the machine over time; the rate of change of health scores of the machine over time; instructions to service the machine; and an indication that the machine is operating in acceptable behavior conditions and no action is necessary.

[0078] Health Score determination systems and methods:

[0079] Figure 9 shows an example embodiment of monitoring a machine’s Health Score once. In a first step 900 the current operating state of the machine is determined. In a second step 901 it is determined whether or not the current operating state is an operating state of interest. If it is not the process end 907. If the machine is operating in an operating state of interest, a health score is determined 902, and then it is determined 903 whether the health score is within acceptable limits. If the machine is not operating within acceptable limits then an alert is issued 904 to perform a specific action A. If the machine is operating within acceptable limits, a further determination 905 can be made regarding a trend in behavior. If the machine is trending toward unacceptable behavior, an alert can be issued 906 to perform a different action B. If the machine is in and maintaining a trend within acceptable behavior values, then the process ends 908.

[0080] Figure 10 shows another example embodiment of monitoring a machine’s Health Score, however in this embodiment in a looping and / or continuous manner. One ordinarily skilled in the art will recognize that the process of Figure 10 can be repeated periodically at predefined intervals or aperiodically. Here, in a first step 1000 the current operating state of the machine is determined. In a second step 1001 it is determined whether or not the current operating state is an operating state of interest. If it is not the process loops back to step 1000. If the machine is operating in an operating state of interest, a health score metric is determined 1002, and then it is determined 1003 whether the health score is within acceptable limits. If the machine is not operating within acceptable limits then an alert is issued 1004 to perform a specific action A and the process reverts to step 1000. If the machine is operating within acceptable limits, a further determination 1005 can be made regarding a trend in behavior. If the machine is trending toward unacceptable behavior, an alert can be issued 1006 to perform a different action B. If the machine is in and maintaining a trend within acceptable behavior values, then the process reverts and loops back to step 1000.

[0081] The systems of the present invention comprise at least one sensor and a computer having a processor and memory. The sensor is positioned to determine a operating metric of the machine. The computer further comprises logic and instructions for performing any of the steps herein described, storing determined data / metrics / score etc., and storing sensor data in memory received from the machine and / or associated sensor.

[0082] Examples:

[0083] Without being bound by limitation, the present invention is further described in the following exemplary embodiments.

[0084] Example 1: Exemplary Process and system for determining a Health Score

[0085] In a particularly preferred embodiment, determining whether a machine is healthy or not depends on a number of factors. The process can be described by these high level steps: Identify what all the operational states of interest are; Identify what operational state the machine is in; Measure / detect / identify how the machine is behaving in that state; Compare its behavior to what has been defined to be healthy or unhealthy in that state; Do additional trend analysis if desired; Trigger an action if needed based on the Health Score.

[0086] Figure 12 illustrates a detailed example embodiment of the invention. In this example, sensor data are obtained related to a characteristic of a machine 1200 in at least one operating state, and more preferably at a plurality of operating states. This sensor data can come from a sensor that is part of the machine itself, such as Sensor 1 or from an independent sensor that is not part of the original construction of the machine, but that is positioned and configured to measure at least one characteristic that is indicative of at least one of the machine’s operating states and preferably a plurality of the machine’s operating state. For example, Sensor 2 1202 is placed in physical proximity to the machine 1200 and can be in contact with a part of the machine (such as an external case, an internal part, the power cord (or any extension power cords or multi-plug outlet devices that are in electronic communication with the machine), and the like. Also, the sensor can be placed at a location away from the machine but positioned in a way to measure at least one parameter that is indicative of the machine’s operating state. For example, Sensor 3 1203 can be a camera, other imaging device, microphone, and / or room temperature sensor that can be placed away from the machine but that is able to capture (e.g. optically, audibly, thermally) at least one parameter that is indicative of the machine’s operating state (such as capturing an indicator light, motion or vibration of a machine). Alternatively, data can be generated by the machine 1200 itself via its internal sensors and transmitted via an output port 1204.

[0087] Furthermore, the system of the invention may include an additional sensor (or multiple additional sensors) that is / are configured and located to measure at least one parameter that is indicative of an operating characteristic in at least one and preferably a plurality of operating states. For example, Sensor 4 1208 can be an additional sensor that is part of the machine construction itself; Sensor 5 1206 can be an additional independent sensor that is not part of the original construction of the machine and is placed in physical proximity to the machine 1200 and can be in contact with a part of the machine (such as an external case, an internal part, the power cord (or any extension power cords or multi-plug outlet devices that are in electronic communication with the machine), and the like. Sensor 6 1207 can be an additional sensor placed at a location away from the machine but positioned in a way to measure at least one parameter that is indicative of the machine’s operating characteristic.

[0088] Data from a first sensor can be configured to measure a parameter indicative of the machine’s operating state (e.g. Sensor 1 1201 or Sensor 2 1202) are transmitted to a Computing Apparatus 1205 (comprising a memory and processor and logic and instructions for performing any and / or all of the described steps) for analysis. Optionally, data from a second sensor (e.g. Sensor 3 1203, Sensor 4 1208, Sensor 5 1206, or Sensor 6 1207) are also transmitted to the computing apparatus 1205. The Computing Apparatus 1205 comprises a processor, memory, and logic and instructions to execute calculations to determine a value from the first sensor that is indicative of the machine’s 1200 operating state (the “Operating State Value, or OSV”), for example:

[0089] • The energy consumption of the machine is above a threshold for a minimum duration of time or for a predefined duration of time. This can be measured with a current sensor, a power sensor, a magnetic field sensor (such as a Hall Effect sensor), etc.

[0090] • The current flowing (or the associated electromagnetic fields) through the machine’s power cord is above a threshold for a minimum duration of time or for a predefined duration of time.

[0091] • The vibration level of the machine is above a threshold for a minimum duration of time or for a predefined duration of time. This can be measured with a motion sensor such as an accelerometer.

[0092] • The fundamental frequency of vibration of the machine is either higher than a first threshold value or lower than a second threshold value for a minimum duration of time or for a predefined duration of time. This can be measured with a motion sensor such as an accelerometer.

[0093] • The sound level of the machine is below a threshold for a minimum duration of time or for a predefined duration of time. This can be measured using a microphone.

[0094] • The temperature level of the machine is above a threshold value for a minimum duration of time or for a predefined duration of time. This can be measured with a temperature sensor such as a thermometer, a thermocouple, a thermistor, an RTD (Resistance Temperature Detector), etc. Then, the computing apparatus 1205 (via the processor, logic and instructions, and memory) compares the OSV to a predetermined value or range of values (optionally stored in memory) to determine the operating state of the machine, wherein each value or range of values is indicative of the operating state of the machine. Examples of this determination are given in the table below.

[0095] Once the computing apparatus 1205 determines the machine’s 1200 operating state, then the computing apparatus analyzes data from the second sensor obtained from a time window that at least partially overlaps (and preferably is within) the time window of the data that was used to calculate the OSV to determine a value that is indicative of the quality of performance of the machine (the Operating Quality Value, or OQV).

[0096] For example: • The energy consumption of the machine is oscillating between a maximum and minimum value, wherein the difference between these two values exceeds a threshold. This can be indicative of a faulty switch in the machine causing it to not draw consistent power.

[0097] • The current flowing (or the associated electromagnetic fields) through the machine’s power cord is oscillating between a maximum and minimum value, wherein the difference between these two values exceeds a threshold. This can also be indicative of a faulty switch in the machine causing it to not draw consistent power.

[0098] • The vibration level of the machine is not consistent and there are higher harmonics of the vibration frequencies that are being generated. This may be caused by a loosening of screws or other means of attachment in a machine. For example, in a centrifuge or a shaker incubator, the motion of the machine may become unbalanced due to loose screws or broken mechanical parts. When this occurs, the machine may vibrate at different frequencies in addition to (or instead of) the expected normal range of frequencies. One ordinarily skilled in the art will recognize there are several common mentors used for frequency analysis, such as Fourier Transforms and Fast Fourier Transforms.

[0099] • The temperature of a machine may be increasing or decreasing at a rate that exceeds a desired rate. Alternatively, the temperature of the machine may not be increasing or decreasing in a smooth manner. These behaviors may indicate an issue with a cooling mechanism of the machine.

[0100] Then, the computing apparatus 1205 compares the OQV to a predetermined value or range of values (stored in the memory of the computing apparatus 1205) to determine what level of quality the machine is operating in, and additionally if this level of quality is acceptable to a user or not. Examples of this determination are given in the table below.

[0101] One ordinarily skilled in the art will recognize that data from a second sensor may not be necessary depending on the nature of what needs to be measured to determine the operating state and the performance quality of the machine. In some cases, the same sensor data can be used for both determinations. For example:

[0102] • One single temperature sensor can be used to determine an operating state by using a threshold value: o If the temperature is above a predetermined threshold value for at least 10 mins, then determine that the machine is in State 1 (e.g. a “Heating State”). o Then, use the same temperature data during this same time interval of 10 mins and compute the rate of increase during this same time window. If the rate of change exceeds a second threshold value, then determine that the machine’s performance is acceptable. o Alternatively, if the rate of change does not exceed the second threshold value, then determine that the machine’s performance is not acceptable.

[0103] • One single sensor can be used to determine the energy consumption of a machine (such as via a current sensor, power sensor, or electromagnetic sensor). o If the energy consumption of the machine is above a certain threshold for at least

[0104] 13 mins, then determine that the sensor is in State 1 (e.g. a “Machine Running State”). o Then use the same sensor reading for the next 20 mins and perform a frequency analysis to quantify the rate of fluctuations (i.e. how quickly the energy consumption varies with time) in energy consumption during that next 20-minute window. If the rate of fluctuations in energy consumption exceeds a threshold frequency for more than 2 of those 20 mins, then determine that the machine is malfunctioning by not drawing a relatively consistent amount of energy. o Alternatively, if the rate of fluctuations of energy consumption is below a threshold frequency, then determine that the machine is operating as expected.

[0105] Example 2: Cold Storage Units (Freezers, ULT Freezers, Cryogenic freezers, Refrigerators) for example compressor or pump based ULT freezers.

[0106] ULT freezers are a well-known device and exceptional examples of machines suitable for the “machines” of the present invention. In the example below, the health determination systems of the present invention are employed to determine the health of a freezer / refrigerator before it fails (for example -80C freezers that have compressors that are known to fail). Information regarding these types of devices can be found at: https: / / labfreezers.net / blogs / blog / single- compressor-vs-dual-compressor-ultra-low-freezers. There are also freezers that have no compressor, but instead uses other means of cooling, for example Stirling engine. Information regarding these devices can be found at: https: / / newlifescientific.com / blogs / new-life-scientific- blog / dual-stage-cascade-vs-stirling-engine-ult-freezers. There are refrigerators / freezers that use a Peltier mechanism for cooling, but not common: https: / / labincubators.net / blogs / blog / peltier-vs- compressor-based-cooling. The present example, systems and methods can be employed with all of these devices.

[0107] In a first step the different operating states that are relevant to the compressor are determined. Note that some -80C freezers can have 2 compressors (one to bring it down to one temperature and a second one to bring it down further, called a dual-stage cascade or dualcompressor system).

[0108] In a common type of ULT Freezer, the mechanism to cool the interior of the freezer is comprised of one or two compressors (typically referred to as a single stage (or single compressor) or dual stage (or dual compressor) freezer. Examples of compressor-based ULT freezers are made by Thermo Scientific.

[0109] When the freezer is cooling, at least one compressor is running to bring down the temperature inside the freezer. Once the target temperature is reached, the compressor turns off. Thus, when the compressor is on, there is higher in energy consumption by the freezer, and when the compressor is not running, there is lower in energy consumption.

[0110] In this example, two distinct operating states can be observed that can be considered relevant: 1) The state in which the compressor is running; 2) The state in which the compressor is not running. Figure 1 illustrates an example of these two operating states: Figure 1A shows an example of a temperature profile 100 of a ULT freezer that has been configured to maintain a temperature between approximately -80C and -85C. Figure IB shows an example of the energy consumption profile 101 of the ULT freezer. Regions A, C, and E are when the compressor is not running and thus the energy consumption is low, and the temperature is seen to be rising. Regions B, D, and F are when the compressor is running and thus the energy consumption is higher, and the temperature is seen to be decreasing.

[0111] Figure 2 illustrates a more realistic energy consumption profile of a ULT freezer, which shows some variability in the energy consumption profile as well as a non-zero baseline when the compressor is not running. Figure 2A shows an example of a temperature profile 211 of a ULT freezer that has been configured to maintain a temperature between approximately -80C and -85C.

[0112] Figure 2B shows another example of the energy consumption profile 210 of the ULT freezer which illustrates other aspects of how the energy profile can vary: Energy profile 210 is at a non-zero level 203 when the compressor is off. This can be due to a low (but non-zero) level of energy consumption for other active systems in the freezer such as any microprocessors / microcontrollers, fans, another compressor that may be running (for example in dual-stage compressor units), any lights that may be on within the unit. The energy profde 210 in regions B, D, and F all show a different shape and / or level when the compressor is on: In Region B (between dotted lines 207 and 212), the compressor is running at level 202. In Region D (between dotted lines 208 and 213) the compressor energy consumption peaks higher to level 200 than in Region A but then decays down. In region F (after dotted line 209), the compressor runs at a steady level 201 that is higher than the steady level 202 in Region A. These examples are given to illustrate that freezer energy consumption may be variable over the course of typical operation. Furthermore, Figure 2 illustrates that when the compressor turns on (at times 207, 208, and 209), there may be a slight delay in the corresponding temperature drop (as shown by dotted lines 204, 205, and 206, respectively).

[0113] Figure 3 shows another example of energy consumption from an actual ULT. The energy consumption profile 301 in this example has been determined by measuring the strength of the magnetic field via a Hall Effect Sensor placed in proximity to the freezer’s power cord. Energy consumption profile 301 shows a relatively regular pattern. However, it is important to note that the energy consumption does not reach 0 when the compressor is off (as shown by the dotted circle 303). This is due to other components of the freezer continuing to consume energy even when the compressor is not running. Furthermore, the energy consumption profile 301 shows an initial peak and subsequent slow delay when it is in the running state (as shown in the dotted circle 302).

[0114] In a further step it is determined when a compressor is running (or not running). This can be determined based on electromagnetic measurements. Direct electromagnetics measurement can be done via an ammeter wherein the current measurement itself can be used as an indicator of whether the compressor is running or not. Additionally, since the voltage will be known for a given power source (i.e. wall plug) in a given country or region, the energy consumption can be calculated using well-known textbook methods (since the current value would then be proportional to the energy).

[0115] Electromagnetic measurement can be done by measuring power consumption using an “in-line” measurement device like the one from Eyedro: https: / / eyedro.com / product / ILM-W- 15A-NA / that measures the current that is flowing as well as the voltage and determines the power consumption. Energy consumption can then be calculated by standard textbook methods to compute the amount of energy that was used during a period of time if the power during that period of time is known.

[0116] Electromagnetic measurement can be done by measuring current using a split core sensor: https: / / www.magnelab.com / split-core-cts / , and / or by Measuring the magnetic fields via sensors that are responsive to magnetic fields such as Hall Effect sensors (such as is described in patent application [P-018]) or magnetometers placed in proximity (or preferably in contact with the power cord). The magnetic field strength at a particular location relative to the power cord will be proportional to the magnitude of the current for a given frequency, and for a given voltage, the magnetic field strength would thereby be proportional to the energy consumption. The magnetic field strength itself or the Hall effect sensor output can also be used to determine if the compressor is running or not.

[0117] The determination of whether / when a compressor is running (or not running) can also be based on physical and / or mechanical measurements such as vibration and / or sound. The compressor will vibrate when the compressor is running (e.g. when running it will vibrate more; when it is not running, it will vibrate less). The compressor will also make noise when running (e.g. the compressor will make more noise when running; when it is not running, it will make less noise).

[0118] The determination of whether / when a compressor is running (or not running) can also be based on temperature measurements. Such as based upon external temperature: when the compressor is running, the unit will generate more heat externally; when it is not running, it will generate less heat externally. In the alternative, based on internal temperature: when the compressor is running, the temperature inside of the freezer unit will decrease; when the compressor is not running, the temperature inside of the freezer will increase. In preferred embodiments where the machine is a CSU and as mentioned in this example, internal temperature is the preferred metric of observation and data collection.

[0119] The determination of whether / when a compressor is running (or not running) can also be based on measurement of other types of signals. For example, via a direct machine data port: Sometimes, the freezer may have a data output port that outputs operating codes that may tell what state the unit is in. In another example, measurement of light can be indicative of operating state and / or behavior (e.g. one may be able to see a light come on or radiation output when the compressor is running.

[0120] In this example of implementing the invention a Health Score is determined of a compressor-based cold-storage unit, such as freezers, refrigerators, ULT freezers, walk-in cold storage units, and similar. In this example, the goal is to identify when a compressor is running to cool the unit and whether: The compressor is working as intended (i.e. acceptable behavior and therefore healthy currently); The compressor is not operating as intended (i.e. not acceptable behavior and therefore unhealthy currently); The compressor’s behavior is trending towards becoming unacceptable (i.e. it’s health is deteriorating, but is still in the healthy region currently); and / or The compressor’s behavior is trending towards being acceptable while still currently unacceptable (i.e. its health is improving but is currently still unhealthy).

[0121] Again, a step in the process is to determine when the compressor is running. For example, over a period of time, temperature data is collected from inside the cold chamber and also data that is representative of the energy that the unit consumes is collected (e.g. from the current flowing through the power cord or measuring associated magnetic fields like we do with our Element-U, etc.).

[0122] The energy data is used to determine when a compressor is running by: making a measure of the energy and when it is above a threshold for a certain amount of time then conclude that a compressor is running; and designating those windows of time as a “Compressor Running” state. Some cold-storage units may have more than one compressor, and one ordinary skilled in the art will recognize that it is desirable to detect abnormal behavior in any of the compressors, so it may not be necessary to distinguish between the operating states of multiple compressors, but that one could choose to do so if desired.

[0123] In a next step, acceptable behavior criteria are determined during the Compressor Running state by: receiving temperature data during those same time windows as when the freezer is in the Compressor Running state; and defining acceptable behavior for the “Compressor Running” state being that the temperature follows an expected profile that decreases over the “Compressor Running” time window.

[0124] Expected / acceptable profiles can depend on many factors. In general, in CSU devices, it can be expected the temperature to decrease smoothly (and ideally in a strictly decreasing manner, information on this phenomena can be found at: https: / / mathworld.wolfram.com / StrictlyDecreasingFunction.html).

[0125] Figure 4 shows one idealized example of a temperature profile over time where the temperature profile 400 is strictly increasing in region A (when the compressor is not running), strictly decreasing 401 in region C (when the unit is in a “Compressor Running” state), and again strictly increasing in region E (when the compressor is not running). Region B shows the temperature 403 transitioning from increasing to decreasing when the unit starts up its compressor and begins to enter the “Compressor Running” state, and Region D shows the temperature 404 transitioning from decreasing to increasing (when the compressor turns off and stops running). Deviations from a strictly decreasing temperature profile indicates a decrease in compressor health.

[0126] Figure 5 shows an idealized example of what the temperature profile of Figure 4 can become if the compressor begins to behave in an unhealthy manner. In Figure 5, the temperature profile 505 is strictly increasing in regions A and E (when the compressor is not running), but in region C when the compressor is running (and hence in the “Compressor Running” state), the temperature is not strictly decreasing throughout the full time of region C. The temperature profile is decreasing in region 500, then increases in region 501, then decreases in region 502, and again increases in region 503, and finally decreases in region 504 before the compressor stops running at the end of region C. This behavior may be caused by any number of different factors, but the net result is that the temperature of the cold chamber in the unit does not decrease in a strictly decreasing manner. Thus, this type of increase and decrease fluctuation can be indicative of a compressor exhibiting unhealthy behavior. While this compressor does eventually bring the temperature down to the desired level, one can conclude that the compressor should be inspected and serviced to ensure better performance in the future. While it is visually evident what the temperature profiles of a healthy compressor and an unhealthy compressor would look like, it is advantageous and useful to develop an algorithmic method to quantify the extent of the temperature profile’s deviation from a desirable healthy behavior profile.

[0127] The profile of Figure 6 shows an actual temperature profile of a ULT Freezer with a setpoint of -80C that has a compressor showing signs of being unhealthy and at risk of failing. The compressor is not running in regions A, C, and E, and as such, the corresponding temperature profiles in those regions (615, 604, and 610, respectively) are seen to be increasing. There are minor time periods where the temperature profile in these regions are not strictly increasing, but given that this is actual temperature data, minor deviations from ideal behavior are to be expected. Furthermore, the time periods when the compressor is not running is not an operating state of interest and so for this example those regions are ignored.

[0128] The compressor is in a running state in regions B, D, and F. As can be seen, the temperature profile in these regions is not strictly decreasing and in fact there are several time periods where there is a significant increase in temperature. For example, regions 601, 606, 609, 612 all show strictly decreasing temperature profiles, but regions 600, 602, 605, and 608 show significant increases in temperature. Regions 603, 607, 611, 613, and 614 show a number of temperature increases and decreases, causing the profile to appear “bumpy” or “noisy”, but nonetheless not strictly decreasing. The actual shape of the correlating profile can depend on many factors: Design of the unit itself such as the geometry of the cold chamber, the strength of the compressor, and whether there is a fan that is circulating inside the cold chamber; How the unit is used including what and how much is stored in the unit, placement of the temperature sensor, opening and closing of the door to the cold storage unit, the temperature set point; Maintenance of the unit including level of ice buildup and defrost activities (including automatic defrosting).

[0129] If the temperature profile can be altered due to events and / or activities in course of normal usage of the unit (such as opening and closing of the door, or defrost cycles), and if these activities can be independently detected (which they can) then corresponding data surrounding these events can either be disregarded or appropriately considered when assessing the health of the compressor.

[0130] In a further step, a health score is determined to quantify compressor behavior. Several methods to algorithmically quantify a Health Score of a compressor-based cold storage unit are available. The following subroutines illustrate several preferred methods of computing a health score.

[0131] Residuals method for determining a health score:

[0132] For each “Compressor Running State” time window, take the corresponding temperature readings and fit a line or curve using standard well-known textbook regression methods (e.g. least squares) and determine the scatter around the curve of best fit by determining the residuals. Figure 7 shows two examples of the temperature profile during one “Compressor Running State”. In Figure 7A, the points represent the sampled temperature data and line 700 represents a linear best fit line. In Figure 7b, the points represent sample temperature data and the curved line 701 represents a nonlinear curve of best fit. In both examples of Figures 7A and 7B, the sampled temperature data is not strictly decreasing as would be ideally expected with a healthy compressor.

[0133] Healthy compressors will lower the temperature when they are running, and thus, the fitted line or curve will be decreasing in value during the “Compressor Running State”. There can be many mathematical expressions that can be used to describe the shape of the temperature profile, and one ordinary skilled in the art will recognize that there are well-known and commonly used functions that can be used to fit the temperature data profile, including but not limited to: Polynomial functions of order 1 (a straight line) or greater order; An exponential or logarithmic function; An error function; Algebraic functions - https: / / en.wikipedia.org / wiki / Algebraic_function; Transcendental functions - https: / / en.wikipedia.org / wiki / Transcendental_function; Etc..

[0134] A health score can be determined by quantifying how scattered the data points are around the line or curve of best fit. One ordinarily skilled in the art will recognize that there are several well-known methods of quantifying the scatter including: A correlation coefficient, preferably the Pearson Correlation Coefficient, for linear regression; Sum of the residuals squared; and Root mean squared error method. The closer the Computed Health Score result is to the Reference Health Score value, the higher the health of the compressor.

[0135] Successive slopes method 1 :

[0136] Figure 8 shows the same temperature data points from Figure 7B. In this method, the slope between adjacent points is determined (can be adjacent points or skip a few points, or perform linear fits along a sliding window). For example in Figure 8, the slopes of the line segments connecting successive pairs of data points is computed: The slope of line segment 810 connecting points 800 and 801; The slope of line segment 811 connecting points 801 and 802; The slope of line segment 812 connecting points 802 and 803; The slope of line segment 813 connecting points 803 and 804; The slope of line segment 814 connecting points 804 and 805; The slope of line segment 815 connecting points 805 and 806; The slope of line segment 816 connecting points 806 and 807; The slope of line segment 817 connecting points 807 and 808; The slope of line segment 818 connecting points 808 and 809. The slope of each segment (810 - 818) will be negative, zero, or positive.

[0137] The sum or average of all the slopes of the segments together (or a weighted average or weighted sum if there is reason to do so) can be figured to determine the Computed Health Score result. It is expected that the slopes of all the segments will be negative for healthy cold storage units since it means the compressor continues to pull the temperature down during its entire running state. The closer the Computed Health Score result is to the Reference Health Score value, the higher the health of the compressor.

[0138] Successive slopes method 2:

[0139] This is a similar method as “successive slopes method 1” above but instead of adding or averaging the slope segments, a value +1 is assigned for each positive slope; 0 for each zero slope; -1 for each negative slope. Then sum of all the values or average of the values can determine the Computed Health Score. It is expected that the Computed Health Score will be more negative the healthier the compressor is. The closer the Computed Health Score result is to the Reference Health Score value, the higher the health of the compressor.

[0140] Successive differences method 1 :

[0141] Similar to the “Successive Slopes” method, but time can be disregarded - for example just determine if the temperature is going up, down, or is flat, regardless of the time it takes. This method could be considered computationally more efficient. In this method, the difference between adjacent, or somewhat adjacent, points is determined. Several ways to accomplish this are possible as it may be possible to skip a few points or average a few points together and then take the difference between successive averages.

[0142] For example as shown in Figure 8, the difference between adjacent data points is computed: Difference 1 = (Temperature of point 800) - (Temperature of point 801); Difference 2 = (Temperature of point 801) - (Temperature of point 802); Difference 3 = (Temperature of point 802) - (Temperature of point 803); Difference 4 = (Temperature of point 803) - (Temperature of point 804); Difference 5 = (Temperature of point 804) - (Temperature of point 805); Difference 6 = (Temperature of point 805) - (Temperature of point 806); Difference 7 = (Temperature of point 806) - (Temperature of point 807); Difference 8 = (Temperature of point 807) - (Temperature of point 808); Difference 9 = (Temperature of point 808) - (Temperature of point 809). Then the sum (or average) of the differences is calculated to determine the Computed Health Score.

[0143] It is expected that the difference calculations of all the segments will be positive for healthy cold storage units since it means the compressor continues to pull the temperature down during its entire running state. The closer the Computed Health Score result is to the Reference Health Score value, the higher the health of the compressor.

[0144] One example variation to this method is to perform the difference calculation by skipping points rather than performing the difference calculation on data points that are adjacent to each other in time. For example, one can take the difference between the first and third point, the second and fourth point, the third and fifth point and so on. In another example, one can take the difference between the first and fourth point, the fourth and seventh point, the seventh and tenth point and so on.

[0145] Successive differences method 2:

[0146] This approach is the same as the “successive differences method 1” above but instead of adding or averaging the computed differences, a value of -1 for each positive difference is assigned; 0 for each zero difference; and +1 for each negative difference can be assigned. Alternatively, +1 can be assigned for each negative difference and 0 can be assigned for any nonnegative difference (such as a positive difference or a zero difference). Then the sum or average of the values can be computed to determine the Health Score. Optionally, the Health Score can be normalized to a range that spans an expected maximum to minimum or expressed as a percentage of a normalized range. It is expected that the Health Score will be more positive the healthier the compressor is. The closer the Health Score result is to the Reference Health Score value, the higher the health of the compressor.

[0147] One example variation to this method is to perform the difference calculation by skipping points rather than performing the difference calculation on data points that are adjacent to each other in time. For example, one can take the difference between the first and third point, the second and fourth point, the third and fifth point and so on. In another example, one can take the difference between the first and fourth point, the fourth and seventh point, the seventh and tenth point and so on. Another example variation to this method is to assign -1 for every difference that is more positive than a first predetermined number; 0 for each difference which falls between the first predetermined number and a second predetermined number; and +1 for each negative difference that is more negative than the second predetermined number. If the first predetermined number was +0.2 C and the second predetermined number was -0.3 C, then the following would be the assigned numbers for the example differences below: Difference of -0.8 C = assign +1; Difference of +0.9 C = assign -1; Difference of -0.2 C = assign 0.

[0148] Frequency analysis method:

[0149] A frequency analysis such as a Fourier transform can be employed to determine a health score. The less smooth the data is (i.e. more scattered and not smoothly decreasing) the higher frequency components will result which equates to a less healthy compressor.

[0150] Example Based on Measured Data:

[0151] Figure 11 shows an example of a Health Score 1101 computed for an ULT freezer in accordance with the invention. In this example, the Health Score can range from 0 to 1, wherein a Health Score of 0.5 and below is deemed to be unhealthy and a Health Score of above 0.5 is deemed to be healthy. The temperature profile 1102 is relatively consistent in section A (where Health Score 1101 is approximately 0.8 and also relatively stable in the healthy region above 0.5. However, the Health Score 1101 begins to trend downward in section B (but is still in the healthy region above 0.5). Also in section B, the freezer temperature 1102 shows a dramatic shift up to a slightly higher temperature. The decrease in the Health Score 1101 in section B is an early indicator that the freezer’s health is starting to deteriorate. Finally in section C, the Health Score 1101 drastically and rapidly falls towards the boundary of 0.5. Also, in section C the freezer’s temperature 1102 shows an uncharacteristic upward trend indicative of a compressor struggling to maintain the desired temperature. Finally in section D the freezer was taken offline for servicing.

[0152] It is common for users of cold storage units to configure an alarm to be triggered if the temperature were to exceed a certain threshold. Most ULT freezers are set to maintain a temperature of -80C and below; as such, it is common to set an alarm threshold at -70C or -65C. In this example, the Health Score calculation was able to detect an unhealthy compressor which was showing signs of impending failure weeks before the freezer’s temperature alarm would have been triggered.

[0153] This example and the present invention apply to CSUs:

[0154] The present example and embodiments described herein can be employed with other CSUs including those that employ cooling via Peltier device and / or a Stirling Engine and / or cryogenics. Stirling engine cooling devices make use of cooling without compressors and are available at https: / / www.biolifesolutions.com / storage / stirling-ultra-low / . Cryogenic Freezers aka LN2 storage units, cooled by pumping in liquid nitrogen when the temperature gets above a certain threshold.

[0155] Operating characteristics of these devices can be determined, for example by measuring many of the same quantities mentioned above including: Current, magnetic fields, power, etc; Vibration / sound; Internal temp; External temp; and Data port etc. For example, in LN2 units, the relevant operating states include: State 1 : Active Pumping where liquid nitrogen is actively being pumped into the cryogenic freezer chamber where the pump mechanism is consuming energy to perform the pumping action; and State 2: No Pumping. Acceptable behavior in each state can be determined or coded as: State 1 : the temperature inside the cryogenic chamber decreases in a smooth and / or strictly decreasing manner until a lower temperature threshold limit is achieved; State 2: the temperature rises in a smooth and / or strictly increasing manner until an upper temperature threshold limit is achieved. A health Score for State 1 can be determined by ascertaining whether the temperature profile should be smooth and / or strictly decreasing. Detection of system failures can be made and could be indicative of: The liquid nitrogen pumping mechanism is does not pump enough liquid nitrogen; The liquid nitrogen pumping mechanism is does not pump the liquid nitrogen smoothly (i.e. the pump is starting to fail); The liquid nitrogen runs out but the pump is still trying to pump.

[0156] Factors Influencing a Health Score

[0157] As has been mentioned above, there can be factors and / or events which influence the health score calculation that are not related to the actual performance quality of a machine. For example: • In a temperature-controlled unit (e.g. Cold Storage Unit, freezer, refrigerator, incubator, oven, etc ), the opening and closing of the door (or not closing the door fully) can cause the temperature profde to change in response. Such a temperature change may be interpreted by the calculation as a true machine failure.

[0158] • In machines that have an element of motion (e.g. centrifuges, shakers, shaker incubators, washing machines, etc.), an imbalance of loads may cause the vibration profile to be different than expected.

[0159] Figure 13 shows both the temperature profile 1300 and the compressor energy consumption profile 1301 for a ULT freezer. In this example, the unit is kept closed and there are no door opening events. As such, the temperature profile and the compressor energy consumption profile both are relatively consistent in their shapes over time.

[0160] Figure 14 shows the same ULT freezer with user-generated activities. The temperature profile 1400 and compressor energy consumption profile 1401 both show relatively consistent shapes in region 1409 (for temperature) and region 1406 (for the compressor). However, the temperature rapidly increases when a first door opening event 1405 occurs and again when a second door opening event 1404 occurs. Immediately after door opening event 1045, the temperature is seen to fall rapidly for a short duration since the door is closed until the second door opening event 1404 pulls the temperature up again. After the second door opening event, the temperature falls rapidly for a short duration but then decreases at a slower rate in region 1402. This is due to the door not being fully closed (due to user error). A third spike in temperature 1403 indicates a third door opening and closing event.

[0161] Thus the normal use of the ULT freezer (doors being opened and closed, or not benign closed fully) can affect the shape of the temperature profile. Correspondingly, there can be an effect on the compressor energy consumption as well. Region 1406 shows a relatively consistent shape of the energy consumption, but regions 1407 and 1408 show elongated times of higher energy consumption. These are due to the compressor running for a longer duration to pull down the temperature after door opening events. In region 1407, the compressor is running for a longer period of time since the door was not fully closed and the internal temperature could not be stabilized as quickly as if the door were fully closed. In region 1408, the compressor runs for a longer period of time as compared to region 1406 but not as long as region 1407 because the door was closed fully and the temperature spike 1403 could be returned to the target range relatively quickly.

[0162] Thus, user behavior can affect the shape of the temperature profile and could lead to an erroneous Health Score calculation since the shape of the temperature profile would not be indicative of the quality of the compressor’ s performance.

[0163] Also, some freezers employ an automatic defrost cycle where the temperature rises to a predetermined level for a period of time. This also can affect the shape of the temperature profile and can lead to an erroneous Health Score calculation. Figure 15 shows temperature and compressor profiles before (region A), during (region B), and after (region C) of a defrost cycle:

[0164] • Region A: Before a defrost cycle. Temperature profile 1502 is relatively “saw tooth” in shape and compressor profile 1503 shows relatively consistent high and low energy consumption regions.

[0165] • Region B: During a defrost period. Temperature profile 1500 rises to a higher temperature and remains relatively constant. Compressor profile 1501 does not show the characteristic high and low switching behavior.

[0166] • Region C: After a defrost cycle. Temperature profile 1504 returns back to a characteristic sawtooth shape. Compressor profile 1505 also returns to its characteristic high and low switching profile.

[0167] As can be seen in Figure 14, the shape of door opening / closing events can appear to be unique shapes that are different from the normal temperature profile shape when the door is kept closed and as can be seen in Figure 15, a defrost cycle can also show markedly different profile shapes. Thus, in many cases, these door opening / closing events and defrost events can be identified independently.

[0168] One ordinarily skilled in the art will recognize that there are many well understood methods in data science that can be used to identify unique profile shapes in time-series data, including but not limited to:

[0169] • Frequency analysis (e g. Fourier transform, Fast Fourier Transform) to identify spikes in temperature (since these spikes will have different frequency content as compared to normal door-closed operation)

[0170] • Machine learning methods trained to identify the shapes of the temperature spikes associated with normal door opening / closing events, defrost events, doors opened but not closed fully, etc. Suitable example of how to identify different events in the temperature data can be found in patent applications and granted patents entitled “Method and apparatus for determining freezer status”, which received US patent no. 11,561,037 (P- 012) and “Method and system for contextual notification”, which received US App. Ser. No. 18 / 026,990 (P-014).

[0171] Thus, in many cases, it would be possible to identify user-generated events (or machinegenerated events that are not indicative of the quality of performance, such as defrost cycles) and address them in calculating a Health Score:

[0172] • In some cases, calculating a Health Score including these events would be acceptable if the impact of these events does not materially affect the Health Score.

[0173] • If the events can be independently identified, then exclude those time regions from the analysis of the Health Score.

[0174] • If the events can be independently identified, then calculate a Health Score separately for just those events. This allows for more than one Health Score and provides more insight into the machine’s performance quality. For example, how quickly the freezer’s temperature recovers after a door opening / closing event can be a separate indicator of the compressor’s performance quality.

[0175] One interesting observation in the example of Figure 14 is that the compressor’s duty cycle also changes in response to user-generated behavior. There are several known methods of assessing the performance quality of a compressor-based cold storage units as described in US Patent Nos. 10,337,964 and 10,837,873. One ordinarily skilled in the art will recognize that determining the performance quality of a compressor solely on a measure of its duty cycle can have severe limitation since as shown in Figure 14, changes to the compressor duty cycle can be caused by user behavior. Thus, being able to detect user behavior (such as door opening / closing events) or other events not indicative of compressor performance quality (e.g. defrost cycles) and incorporating an appropriate consideration of such events (as described above) when a measure of performance quality is calculated is an important improvement to duty -cycle based calculations.

[0176] Example 2: Incubator and / or oven health Environmentally controlled chambers are useful in many applications, such as cell culture incubators and ovens. The present system and methods are also useful in determining health of the types of machines.

[0177] In a first step different operating states of these environmentally controlled machines are determined. Operating states of these machines vary and depend on manufacturer / model specification. However, a few commonalities exist. For example, with incubators: CO2 is pumped in when the CO2 drops below a certain level; Temperature is regulated; humidity can be actively regulated; a shaker may be presented and intermittently / constantly activated. See for example incubators provided by Infers Multitron found at https: / / www.infors- ht.com / en / shakers / incubator-shakers / multitron / . In another example, ovens have many common aspects which are measurable including: Heating mechanism is on or off or in some other intermediate state; and Circulating fan can be on, off, or set to a variable speed. There can be more than one heating state. Some ovens can use multiple heating elements and can turn them on / off independently. For example, one heating element can be used for lower temperature settings, but two or three (or more) heating elements can be used to increase the temperature to higher levels.

[0178] Accordingly, a few operating variables and parameters can be monitored by a sensor, sensor data provided to received and received by the computer and used in the determination of machine operating states. These are included in the table below.

[0179] Once the operating states of the machine are determined, operating states of interest can be identified, and values of acceptable behavior in the operating states of interest can be determined and / or received and stored in computer memory. The machine is then run in normal operation / setting in its various operating states while sensor data is determined by the respective sensor and received by the computer. A health score, or several health scores, of the machine can then be calculated from the sensor data received during normal operation by comparing the sensor data to the values of acceptable behavior stored in memory in the various operating states of interest. A further determination can be made with respect to the determined health score, or the several health scores, and further action can be taken. For example, if the individual health score is outside of a range of acceptable values further action can be taken. Alternatively, if the trend of health scores is trending in a specific direction toward healthier or unhealthier behavior, further action can be taken.

[0180] Example 3: Manufacturing and Facility Machines Health

[0181] In general, machines used in manufacturing are expected to be running without unexpected failures nearly all the time. As such, being able to estimate the health of such machines is important and useful. Furthermore, manufacturing facilities have specialized environmental control and conditioning machines (such as HVAC systems, air purifiers, humidifiers, dehumidifiers, etc.). Even though these machines may not be on the production floor itself, they are nonetheless critical in ensuring manufacturing quality. As such, the health and general performance quality of these facility machines is also important. The systems and methods described herein can be employed to determine the health of these machines and provide further action depending on the outcome of the determined health.

[0182] Examples of variables, operating states, and sensor measurable values are provided below that can be employed in the systems and methods described herein in the determination of health of manufacturing and facility machines.

[0183] Example 4: Centrifuge Health As with other embodiments described herein, the systems and methods of the present invention can be employed to determine health of other laboratory devices such as a centrifuge. A variety of sensors, and corresponding data, can be positioned and determined regarding determination of operating states, acceptable behavior, actual behavior, and the health and / or health score of the machine.

[0184] A power sensor can be employed to determine the operating state and another sensor can be employed to determine acceptable and / or actual behavior of a machine such as a centrifuge. Or vice versa where a different sensor is employed for determining state while a power sensor is employed to determine behavior. For example: a motion sensor (e.g. vibration sensor or accelerometer) or an acoustic sensor (e.g. microphone) can be employed to determine when a centrifuge is running, and a sensor to measure power draw can be employed to determine if more power than expected is being used. If more power is being used, this may indicate that the motor has to work harder to achieve the same RPM which may be indicative that the health of the centrifuge is falling.

[0185] Example 5: Additional Embodiments

[0186] Employ single sensor measurement in the systems and methods of the present invention:

[0187] In some cases, it may be possible to measure only one parameter and infer both the operating state and the machine’s health from a single sensor. In one example, for a compressorbased CSU, if the temp is decreasing then it can be assigned that during that time period the operating state is “Compressor Running”. The duration of the temperature decrease (from its peak temperature to its trough) can be analyzed to determine when and for how long the compressor has been in the “Compressor Running” state. During the “Compressor Running” operating state, one can employ any of the methods described herein to determine if the temperature is strictly decreasing, and if not, by how much the temperature profile deviates from strictly decreasing behavior in determination of health and / or health score of the machine. Further action can then be taken depending on a further comparison of the health and / or health score of the machine.

[0188] Employ multiple sensor measurements in the systems and method of the present invention: One ordinarily skilled in the art will recognize that more than one sensor or data stream (or combinations thereof) can be used to determine an operating state and / or determining the behavior of a machine in an operating state.

[0189] For example, both vibration and energy usage together can be used to identify if a machine is in a running state (by using an accelerometer and a current sensor, respectively). If both sensor data streams record data above a certain threshold, then one can determine that the machine is in a running state.

[0190] Any combination or all of the sensor data, machine operating states, the machine health score, and / or actions can be stored in computer memory in a data file for subsequent use in any method or system herein described.

Claims

Claims:

1. A system for determining a machine health score, the system comprising a computer and a first sensor, wherein: the first sensor is positioned to determine sensor data relating to a characteristic of a machine during a plurality of machine operating states; the computer comprises a processor, memory, and logic and instructions executable by the processor for performing the steps of a. receiving, by the computer, first sensor data from the first sensor relating to a characteristic of the machine during each of the plurality of machine operating states; b. determining and storing in memory, by the computer, the plurality of machine operating states from the first sensor data received in step a.; c. receiving and storing in memory, by the computer, an indication of an operating state of interest selected from the plurality of the operating states determined and stored in memory in step b.; d. receiving and storing in memory, by the computer, values of acceptable behavior of the machine during the machine operating state of interest; e. receiving, by the computer, second sensor data relating to a characteristic of the machine during the machine operating state of interest; f. determining and storing in memory, by the computer, the health score of the machine from the second sensor data by comparing the second sensor data to the values of acceptable behavior of the machine during the machine operating state of interest received and stored in step d., thereby determining the health score of a machine.

2. The system of claim 1, wherein step d. is performed by the step of: receiving and storing in memory, by the computer, values of acceptable behavior of the machine during the machine operating state of interest, wherein the values are received from a user and / or are determined by the computer by parsing a lookup chart.

3. The system of claim 1, wherein step d. is performed by: receiving and storing in memory, by the computer, third sensor data relating to a characteristic of the machine during the machine operating state of interest; and determining and storing in memory, by the computer, values of acceptable behavior of the machine during the machine operating state of interest from the third sensor data received in step d, wherein: the third sensor data is obtained when the machine is in a known acceptable working condition; and / or the third sensor data is an average or sampling of sensor data received prior to performing step (d).

4. The system of claim 1, wherein step f. is performed by the computer by: determining the magnitude of offset of the second sensor data to the values of acceptable behavior of the machine; and correlating the magnitude of offset of the second sensor data to a health score of the machine.

5. The system of claim 1, wherein the computer comprises logic and instructions executable by the processor for performing the additional steps of: g. repeating steps e. and f., by the computer, to determine a plurality of health scores of the machine over time from the second sensor data; h. a further step selected from the group consisting of: i. determining, by the computer, from the plurality of health scores a trend in health scores of the machine over time; and ii. determining, by the computer, from the plurality of health scores a rate of change in health scores over time; i. determining, by the computer, from the plurality of health scores determined over time from step k., the trend of health scores of the machine overtime and / or the rate of change in health scores over time, if a correlation exists; and j . if a correlation is determined to exist in step m., taking a further action selected from the group consisting of: triggering an alarm; turning the machine on / off; instituting a machine service routine; and issuing an alert, optionally wherein a correlation is determined to exist in step i. and an alert is issued, wherein:the alert is sent to another computing device and / or is sent to a user (e.g audibly, visually, via voice, text, SMS, email, etc); and / or the alert is a message containing information selected from the group consisting of warning of an immediate or impending condition; the health score of the machine; the trend of health scores of the machine over time; the rate of change of health scores of the machine over time; instructions to service the machine; and an indication that the machine is operating in acceptable behavior conditions and no action is necessary, and optionally wherein the computer comprises a processor, memory, and logic and instructions executable by the processor for performing the steps of: predicting a future machine health score, by the computer, from the plurality of health scores determined over time from step g., the trend of health scores of the machine overtime, and / or the rate of change in health scores over time; and performing an additional action depending on the predicted future health score, optionally wherein the additional action is selected from the group consisting of: triggering an alarm; turning the machine on / off; instituting a machine service routine; and issuing an alert, and optionally wherein: the alert is sent to another computing device and / or is sent to a user (e g audibly, visually, via voice, text, SMS, email, etc); and / or the alert is a message containing information selected from the group consisting of: warning of an immediate or impending condition; the health score of the machine; the trend of health scores of the machine over time; the rate of change of health scores of the machine over time; instructions to service the machine; and an indication that the machine is operating in acceptable behavior conditions and no action is necessary.

6. The system of claim 1, further comprising fourth sensor data relating to events that affect or cause anomalies in the first sensor data, second sensor data, and / or third sensor data, if present, wherein the first sensor data, the second sensor data, and / or the third sensor data, if present, AND / OR the health score determined in step f. are adjusted or ignored by the computer in response to fourth sensor data.

7. The system of claim 1, wherein: the second sensor data, third sensor data, if present, and / or fourth sensor data, if present, are received by the computer from the first sensor, and / or the system further comprises a second sensor positioned to obtain sensor data relating to a characteristic of the machine during the plurality of machine operating states, and wherein the second sensor data, third sensor data, if present, and / or fourth sensor data, if present, are received by the computer from the second sensor.

8. The system of claim 1, wherein the first sensor data, the second sensor data, third sensor data, if present, and / or fourth sensor data, if present, are sensor measurements selected from the group consisting of electrical input to the machine; interior temperature of the machine; ambient temperature surrounding the machine; temperature of the working components of the machine; and sound and / or vibration.

9. The system of claim 1, wherein the computer comprises logic and instructions for performing the additional steps of: k. correlating the health score determined in step f. with a lookup chart to determine if a correlation exists; and l. if a correlation is determined to exist in step o., taking a further action, by the computer, selected from the group consisting of: triggering an alarm; turning the machine on / off; instituting a machine calibration routine; and issuing an alert, optionally wherein the correlation is determined to exist in step k. and an alert is issued, wherein: the alert is sent to another computing device and / or is sent to a user; and / or the alert is a message containing information selected from the group consisting of: warning of an immediate or impending condition; the health score of the machine; the trend of health scores of the machine over time; the rate of change of health scores of the machine over time; instructions to service the machine; and an indication that the machine is operating in acceptable behavior conditions and no action is necessary.

10. A system for determining a health score of a cold storage unit (CSU) comprising the machine health score determination system of claim 1, wherein: the machine comprises a cold storage unit (CSU) comprising a compressor and / or pump (compressor / pump), the first sensor comprises a sensor selected from the group consisting of: a temperature sensor positioned to determine interior temperature of the CSU; a voltage and / or current sensor and / or electromagnetic sensor positioned to determine electrical input to the compressor / pump; a temperature sensor positioned to determine ambient temperature of room surrounding the CSU; a temperature sensor positioned to determine temperature of the compressor / pump; vibration and / or sound sensor positioned to determine sound and / or vibration of the CSU compressor / pump; the first sensor data comprises data from the selected first sensor; the operating state of interest is on / cooling; and the plurality of machine operating states of the freezer comprise: on / cooling; and off / warming.

11. The system of claim 10, further comprising fourth sensor data relating to user events that affect or cause anomalies in the first sensor data, second sensor data, and / or third sensor data, if present, wherein all or a portion of the first sensor data, the second sensor data, and / or the third sensor data, if present, AND / OR the health score determined in step f. are adjusted or ignored by the computer in response to fourth sensor data, optionally wherein the events that affect or cause anomalies are selected from the group consisting of: opening the door of the CSU; placing an object in the CSU that is a different temperature than the interior chamber temperature; opening and closing of the door; opening the door and leaving the door open; opening the door and partially closing the door; performing a calibration protocol; and performing a defrost protocol.

12. The system of claim 10, wherein the plurality of machine operating states further comprise an automatic defrost process operating state.

13. The system of claim 13, wherein the acceptable behavior during the operating state of interest is that the temperature of the CSU enclosed space is strictly cooling.

14. A system for determining a machine health score, the system comprising a computer and a first sensor, wherein: the first sensor is positioned to determine sensor data relating to a characteristic of a machine during a machine operating state of interest; the computer comprises a processor, memory, and logic and instructions for performing the steps of: m. configuring the first sensor to determine and transmit the sensor data relating to a characteristic of a machine during the machine operating state of interest; n. receiving and storing in memory, by the computer, values of acceptable behavior of the machine during the machine operating state of interest; o. receiving, by the computer, the sensor data relating to a characteristic of the machine during the machine operating state of interest; p. determining and storing in memory, by the computer, the health score of the machine from the sensor data by comparing the sensor data to the values of acceptable behavior of the machine during the machine operating state of interest received and stored in step h., thereby determining the health score of a machine.

15. A system for determining a machine health score, the system comprising a computer and a first sensor, wherein: the first sensor is positioned to determine sensor data relating to a characteristic of a machine during at least one machine operating state of interest; the computer comprises a processor, memory, and logic and instructions executable by the processor for performing the steps of:q. receiving, by the computer, first sensor data from the first sensor relating to a characteristic of the machine during the at least one machine operating state of interest; r. determining and storing in memory, by the computer, at least one machine operating state of interest from the first sensor data received in step q.; s. receiving and storing in memory, by the computer, an indication of the at least one operating state of interest; t. receiving and storing in memory, by the computer, values of acceptable behavior of the machine during the at least one machine operating state of interest; u. receiving, by the computer, second sensor data relating to a characteristic of the machine during the at least one machine operating state of interest; v. determining and storing in memory, by the computer, the health score of the machine from the second sensor data by comparing the second sensor data to the values of acceptable behavior of the machine during the at least one machine operating state of interest received and stored in step t., thereby determining the health score of a machine.

16. A cold storage unit (CSU) health score determination and modification system comprising: a computer; and a first sensor, wherein: the first sensor is positioned to determine sensor data relating to a characteristic of the CSU during on and off compressor / pump operating states; the computer comprises a processor, memory, and logic and instructions executable by the processor for performing the steps of: a. receiving, by the computer, first sensor data from the first sensor relating to a characteristic of the CSU during the off and on compressor / pump operating states; b. receiving and storing in memory, by the computer, values of acceptable behavior of the CSU during the off and / or on compressor / pump operating states; c. determining and storing in memory, by the computer, the CSU health score by comparing the first sensor data to the values of acceptable behavior of the CSU during the off and on compressor / pump operating states received and stored in step b.;d. receiving, by the computer, second sensor data relating to events that affect or cause anomalies in the first sensor data; and e. adjusting or ignoring, by the computer, any or all of the first sensor data received by the computer in step a. and / or health score determined in step d. in response to second sensor data received in step d.

17. The system of claim 16, wherein values of acceptable behavior of the off and / or on freezer states include: compressor / pump duty cycle and internal temperature of the CSU, wherein the values of acceptable behavior of the off and / or on freezer states include compressor / pump duty cycle, and wherein the duty cycle is determined by the computer by comparing the length of time the compressor / pump is on to the length of time the compressor / pump is off.

18. In CSU monitoring applications employing a computer to receive and analyze sensor data relating to compressor cycling over time to determine CSU health, the improvement comprising: the computer comprises a processor, memory, and logic and instructions executable by the processor for performing the steps of: a. receiving, by the computer, sensor data relating to events that affect or cause anomalies in the sensor data relating to compressor cycling over time; and b. adjusting or ignoring, by the computer, any or all of the sensor data relating to compressor cycling over time in response to sensor data received in step a. AND / OR adjusting the determined CSU health in response to sensor data received in step a.

19. A method for determining a machine health score, the method comprising: a. associating the system of claim 1 with a machine;b. receiving first sensor data stream, second data stream, third data stream, if present, and fourth data stream, if present, by the computer; and c. determining the machine health score.

20. A method for determining a machine health score, the method the steps of: a. positioning a first sensor first sensor to determine sensor data relating to a characteristic of the machine during a plurality of machine operating states; b. receiving first sensor data from the first sensor relating to a characteristic of the machine during each of the plurality of machine operating states; c. determining the plurality of machine operating states from the first sensor data received in step b.; d. selecting an operating state of interest selected from the plurality of the operating states determined and stored in memory in step c.; e. receiving values of acceptable behavior of the machine during the machine operating state of interest; f. receiving second sensor data relating to a characteristic of the machine during the machine operating state of interest; and g. determining the health score of the machine by comparing the second sensor data received in step f. to the values of acceptable behavior of the machine during the machine operating state of interest received and stored in step e., thereby determining the health score of a machine.

Images